# GLOSSARY OF TERMS AND TECHNIQUES FOR POLITICAL ANALYSIS

Document Type:

Collection:

Document Number (FOIA) /ESDN (CREST):

CIA-RDP85T00875R001100020001-3

Release Decision:

RIPPUB

Original Classification:

K

Document Page Count:

97

Document Creation Date:

November 9, 2016

Document Release Date:

February 18, 1999

Sequence Number:

1

Case Number:

Publication Date:

January 1, 1974

Content Type:

STUDY

File:

Attachment | Size |
---|---|

CIA-RDP85T00875R001100020001-3.pdf | 4.43 MB |

Body:

Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Glossary of
Terms and Techniques
for Political Analysis
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001.-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Preface
25X1A
For some years a move has been underfoot to turn POLITICAL
science into political SCIENCE. The traditional historical-descriptive
approach has been labeled inadequate and sophisticated behavioral and
management science tools have been brought to bear on political problems.
In the rush toward science, jargon has proliferated. Mathematical,
statistical, and computer terms, the holy water of the new era, are sprinkled
generously throughout articles in the professional journals. At best the new
political science vocabulary increases precision, at worst it hinders
communication. Even when it is to be deplored, however, it cannot be
dismissed because many of the leading figures in the discipline use this
jargon and many of the important new works are couched in such language.
Thus this glossary.
The glossary is divided into two alphabetic sections. The first section
defines approximately 500 terms-a vocabulary list of words and phrases
which are used by political scientists, particularly when describing many of
the new techniques of political analysis. The second section explains
approximately 100 techniques-rncthods or approaches for analyzing
political data. The terms appear in lower case letters, the techniques are
capitalized.
An effort has been made to be comprehensive and it is hoped that this
glossary will ease the way through most political science books or journal
articles. Comments and additions, however, should be sent to or 25X1 A
of the Office of Current Intelligence, room 7,
143-5204.
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
TERMS
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
a posteriori--determined after the fact. An a posteriori probability is
one determined by empirical observation. For example, experience tells us
that a flipped coin lands heads one half of the time.
a priori-determined in advance. An a priori probability is set strictly
by definition of the mathematical system. For example, a football coach
could specify before the game that each of his two quarterbacks will play 50
percent of the game.
abscissa-the horizontal axis on a graph, designated X. (see: ordinate)
ORDINATE
/ /ABSCISSA
absolute value--the value of a number with the positive or negative
sign omitted. Absolute value is ideatif%ed by two parallel lines around the
number. Thus /8/ (read "the absolute value of eight") and / -8/ (read, "the
absolute value of negative eight") are both equal to 8. Absolute value
measures the size of a number or movement, without measuring its
direction. For example, 161-3+ 1-41-1-51=2
acceptance region-one of two mutually exclusive regions in a sample
space used in the testing of statistical hypotheses. If the sample point falls
into one (the region of acceptance.) the hypothesis is accepted as true; if it
falls into the other, the hypothesis is rejected. (see: sample space, hypothesis
testing)
accuracy-in statistics tP e closeness of computations or estimates to
the exact or true values.
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
actor-any individual or social group that affects the decision process
within a political system. As in the theater, the political actor "plays a
role" and this role and attendant behavior can be described, analyzed,
explained, and predicted for political systems at all levels--local, state,
national, and international. The rational actor model focuses on the whole
state as a single actor, interacting with other states in the international
system. (see: rational actor model)
admissible strategy-a strategy with at least as good an overall out-
come as any other possible strategy but which may be surpassed by a better
strategy for some particular situation. (see: GAME THEORY)
ADP (automatic data processing)-the name used to denote the type of
computer equipment which processes data automatically as opposed to
manually. (see: computer)
aggregate data-data which summarizes the characteristics of a
number of cases, such as trade or census figures, but provides no informa-
tion on individual cases, such as separate trade transaction figures or
specific counts of people per household. (see: SURVEY RESEARCH)
aggregation of interests-the process by which two or more political
parties, interest groups, governmental bodies, or individuals combine their
demands to build a consensus in support of a particular policy or political
objective.
aggregative index-an index number which is constructed by
combining a number of items as distinct from picking out a representative
item. An aggregative price index, for example, gives the total of prices on
all commodities as opposed to picking out certain representative
commodities.
aggregative model-a model of a political or economic system in which
the variables are constructed from groups of individual variables, as when
an average value of military weapons is substituted for the actual prices of
the weapons. (see: model)
algorithm-a set of mathematical rules for a given procedure; a
sequence of statements defining a computer program.
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
alpha error-in hypothesis testing, the probability of rejecting a
hypothesis when it is actually true. (see: beta error)
alphanumeric-a name or label which contains both letters and
numerals used in computer programs.
alternative hypothesis-in the testing of hypothesis, any admissible
hypo+..i,c is alternative to the one under test.
amplification-(see: distortion)
analog computer-a device which simulates some mathematical
process or relationship so that the results of the process can be observed as
physical quantities, such as voltage or current. (see: digital computer)
analysis-the methodical investigation of a problem, and the
separation of the problem into smaller related units for more detailed study.
analytical model-a model in which interactions are expressed as
mathematical equations. (see: model)
anomic group-largely spontaneous, temporary groups formed to
express discontent through riots or demonstrations. (see: GROUP
THEORY)
antilogarithm or antilog-the number from which a logarithm (or log)
is derived. For example 100 is the antilog of log 2 since 102=100. (10 is
called the base.) (see: logarithm)
approach-a strategy of analysis which provides a unique viewpoint
and a set of tools or methods for the examination of political phenomena.
The psychological approach, for example, focuses on the influence of
personality traits on political behavior and uses psychoanalytic and other
techniques to study personality. (see: method)
argument-an independent variable (see: independent variable)
array-a table for the presentation of data in a row and column format.
A correlation matrix, for example, is a two dimensional description of the
association between variables. The intersection of row B and column 4
shows a correlation of .6 between these two variables. (see: contingency
table, matrix, variable)
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
1,00
32
-,00
.22
.32
1,00
-.08
.00
-.00
-08
100
-.10
22
00
-.10
1,00
articulation of interests-the making of a demand on a political system;
the means by which attitudes, opinions, and beliefs are transformed into
expressed demands for governmental action. In structural-functional
analysis, the articulation of interests i3 recognized as a basic input activity in
a typical political system. (see: input, SYSTEMS ANALYSIS,
STRUCTURAL FUNC TIONAL ANALYSIS)
artificial intelligence-the capability of a machine to perform functions
that are normally associated with human intelligence, such as reasoning,
learning, and self-improvement.
association-the degree of dependence or independence which exists
between two or more variables whether they are measured quantitatively or
qualitatively. If two variables are dependent, one upon the other, a change
in the value of one will effect a change in the value of the other. (see:
CORRELA TION)
asymptote-a straight line always approaching but never meeting a
curve; tangent at infinity.
CURVE
ASYMPTOTE \
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
attitude scaling-measures and compares political and social attitudes
by arranging the answers to written or oral questionnaires according to the
intensity of feeling revealed on a scale of "strongly agree" to "strongly
disagree." When attitude scales are applied to legislators and judges, for
example, the results are used to identify voting blocs, the impact of party on
voting decisions, and the reasons for inconsistent voting patterns. (see:
GUTTMA N SCALING)
attribute-in political science, a particular kind of qualitative political
variable that refers to the structural characteristics of nations, for example,
the power of the executive. In statistics, a qualitative characteristic of an
individual person or case that doesn't change-such as sex-as
distinguished from a variable which can take different values at different
times-such as age. (see: variable)
Aumann-Maschler theory-a type of game theory which does not
attempt to predict which coalition will form but only to determine what the
payoffs will be once a coalition is formed. The theory takes only the
strengths of the players into account; all considerations of fair play an J
equity are put aside. (se- T. GAME THEORY)
authority-political influence derived from a willing acceptance by
others of one's legitimate right to make rules or issue commands and to
expect compliance. The authority relationship is voluntary, as contrasted
with power which is based on the use of physical coercion or material
resources. (see: power)
autocorrelation-a measure of the degree to which a series of numbers
correlates with itself. It shows, for example, the amount of change over
time measured by the consistency in a series of numbers, and indicates,
therefore, whether the present values are significantly different from p :3t
ones. If a variable changes little over time, autocorrel.ation will be high.
For example, the number of people who vote in a certain precinct will vary
according to the weather, the issues, etc., but it may remain basically the
same from year to year. In this case, autocorrelation will be high. (see:
CORRELA TION)
autonomous equations-in econometrics, an equation which describes
the behavior of one particular group or sector of the economy. (see:
ECONOMETRICS)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
autoregression-the generation of a series of observations whereby the
value of each observation is partly dependent upon the values of those which
have immediately prececded it.
average-(see: mean)
Bales' typology-a system designed for the general study of smali
groups which classifies observed behavior into positive reactions, attempted
answers, questions, and negative reactions. (see: GROUP THEORY)
band-a group of circular recording tracks on a computer's data
storage device, such as a drum or disc. (see: computer)
bar chart-a graphical representation of frequencies or magnitudes by
rectangles drawn with lengths proportional to the frequencies or magnitudes
concerned.
APRIL
MARCH
FEBRUARY
JANUARY
DECEMBER
0 20 40 60 80
100
LEVEL OF ISRAELI/SYRIAN HOSTILITY
base-a number or magnitude used as a standard of reference.
base line-(see: abcissa)
base period-the period of time for which data used as the base of an
index number, or other ratio, have been collected. It can be as short as one
day, or as long as the average of a group of years.
Bayes Theorem-(sec: BA YES IA N A NA L YSIS)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
behavior pattern.-v,,ty regular or recurring form of human activity.
Political behavior patterns may range from those based on largely internal,
psychological responses (thought, perception, judgment, attitude) to overt,
observable physical responses (voting, protesting, lobbying, campaigning).
Human behavior patterns provide the basis for the scientific study of
politics.
behavioral s6~,?:nce-(see: social science, BEHA VIORALISM)
behaviorism-which should be distinguished from "behavioralism," is a
formal school of psychology which holds that only overt, observable data in
the form of stimuli and responses are meaningful for analysis, whereas
mental factors such as attitudes, opinions, and beliefs are meaningless. (see:
BEHA VIORALISM)
benchmark problem-a problem used to evaluate the performance of
computers relative to each other. (see: computer)
Bernouli Theorem-sometimes called the law of avenges or law of large
numbers, it states that as the number of trials of an event rises indefinitely,
the probability of each observation will come to approach the established
probability. For example, if you flip a coin long enough, heads will
eventually occur 50% of the time, even if you begin with long irregular runs
of tails.
beta coefficient-in statistics, the coefficient in regression equations,
often denoted by the letter "B." The value of the beta coefficient is a
measure of the influence that the independent variable has upon the
dependent variable. (see: coefficient, REGRESSION)
beta error-when testing a hypotheses, an error incurred by accepting a
hypothesis when it is actually incorrect. (see: alpha error)
bias--in statistics, any effect which distorts a statistical result totally as
distinct from a random error which may distort on any one occasion but
balances out on the average.
bimodal distribution-a frequency distribution in which two intervals
have more observations than any other interval. For example, a distribution
of people's heights may have two modes, one primarily of men and the other
primarily of women. (see: distribution, J-curve. ;:ode)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
NUM[3GR cc:
OBSERVATIONS
C
MODES
\
IL
binary notation-a system of writing numbers using only 0 and 1. The
numbering system we are most familiar with uses base 10. Thus the number
I I I is read: one one, one ten, and one hundred. The number 1000 is read: no
ones, no tens, no hundreds, and one thousand. Binary notation uses the base
2 and each column to the left of the decimal point is 2 raised to a
successively higher power. The one's column in the normal number system
is equivalent to 2?, the ten's column is 21, the hundreds column is 22, the
thousands column 23, and so on. Thus in binary notation, 111 is read: one
211, one 21, and one 22. Adding these, 11: is equal to the normal number 7
(2?=1+21=2+2Z=4). Similarly the number 8 is written 1000 in binary
notation: no 2?, no 2', no 22, and one 23 (2X 2X 2 =8).
binary sequence-any sequence, each number of which can take one
of two possible values.
binomial distribution-the probabilities of getting a certain number of
successes from a given number of trials. For example, the probabilities of
getting 0, 1, 2, or 3 heads when a balanced coin is flipped 3 times are MI, %,
%, and '/e respectively. 1 his is found by listing all the possible combinations
(eight) of the three flips and then counting how often 0, 1, 2, or 3 heads
occur:
H
H H
T
H H
H
H T
T
T H
H
T H
T
H T
H
T T
T
T T
bipolar factor-in factor analysis, a factor which is positively
correlated with some variables but negatively sorrel-ited with others. When
such a factor is identified, it is regarded as expressing a property which may
have a negative as well as positive intensity. For example, cowardice as
opposed to bravery, cowardice being regarded as a quality in itself and not
as the mere absence of bravery. (see: FACTOR ANALYSIS)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
biserial correlation-a coefficient designed to measure the correlation
between two qualities, one of which is measurable, the other a simple
dichotomy according to the presence or absence of an attribute. A
correlation measuring the strength of a government, for example, according
to whether it is a military or non-military one, is biserial. (see: attribute,
CORRELA TION)
bit-an abbreviation for binary digit, which is a digit of a number
written in the scale of two, usually using only 0 and 1. In computer
language, the term also refers to a single piece of information stored and
conveyed by an electrical impulse. (see: binary notation)
bivariate analysis-a generic term for any analysis which takes into
account two variables. (see: variable)
block-any group of items under treatment or observation. A block
may comprise a group of contiguous plots of land, all the votes of a given
area, all the psychological characteristics of a given family. The general
purpose of dividing all the material under study into blocks is to isolate the
sources of differences.
block diagram-a diagram of a system, instrument, computer, or
program in which selected portions are represented by annotated boxes and
interconnecting lines. (see: computer)
Boolean algebra, Boolean logic-a branch of algebra in which the
elements are sets instead of numbers, and a branch of logic in which the
elements are sets instead of sentences. Boolean operations include union
and intersection instead of addition or subtraction. (see: Venn diagram)
INTERSECTION (Af1B)
bootstrap-a technique or device designed to bring itself into a desired
state by means of its own action. A machine routine, for example, whose
first few instructions are sufficient to bring the remainder into the com-
puter from an input device. (see: computer)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
capability analy:4is-an assessment of a state's ability to achi!:ve an
objective vis-a-vis other states through the application of military, political,
economic, psycho!ogical, and other forms of power and influence.
Capability analysis provides decision-makers with the comparative data
that will enable them to make a policy choice based on the, relative
feasibility of alternative courses of action.
card image-the equivalent amount of information on a magnetic
computer tape that can be stored on one punched card. (see: computer)
categorical distribution-a distribution in which the data are sorted
into categories according to some qualitative description rather than by
wamerical value.
causal chain model-a large-scale economic model which looks at the
changing pattern of the model's variables over time. (see- model, variable)
causal model-a simplified description of reality which isolates the
causal relationships between a few variables-X, Y, and Z, for example. A
causal model does not seek to prove that X necessarily causes Y, but merely
that the variations in X can be used to explain the variation in Y. (see:
model, causality)
causality-the relationship which links two or more political variables
together in a cause-effect sequence to generate a particular event. Analysis
of causality seeks to identify the antecedent action or change in one vari-
able, called the independent variable, that produces or helps to produce
a change in a second variable, the dependent variable. Causality can seldom
be "proved" in the strictest sense, but it usually inferred from observed
sequential relationships. (see: independent and dependent variable)
cell frequency-when a frequency distribution is classified into
categories, the subcat~ ' 1?ories are called cells; the frequency with which
observations fall into a particular cell is the cell frequency. (see: distribution)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
MILIIANY NOW
MILIfANY
C!;LL
I RL?OUI NCY
POLITICAL LEANINGS
or MAJOR (JOVERNMENT Ll:, DER5
central limit theorem-the theorem which gives the normal distribution
its central place in the theory of probability and the theory of sampling. In
its simplest form th. thcol'em states that the variances of a given group of
data will tend to be normally distributed-that is most of the cases will tend
to cluster at a mid-point with fewer cases at either extreme-as the amount
of data tends to infinity. (see: distribution, normal distribution, variance)
central processing unit-the unit of a computing system that includes
the circuits controlling the inte=rpretation and execution of instructions to
the computer. (sec: computer, program)
central tendency-a characterization of the "middle" value of a variable,
or averages of various types which summarize the information contained in
a set of data. (see: measures of central tendency, variable)
chain-a sequence of terms, values, or items, such that each item
depends in some defined way upon the previous items in the series. The
most common of these is the Markov chain. (see: MA RKO VA NA L YSIS)
chain index-an index number in which the value at a given period is
related to a base in the previous p.;riod, i.e., Argentine political stability in
1972 compared with 1971 and 1973 compared with 1972, as distinct from
an index number which is related to a fixed base, i.e., Argentine political
instability in 1971, 1972, and 1973 related to 1968. The comparison of non-
adjacent periods is usually made by multiplying consecutive values of the
index numbers. For example, if' the value of Argentine political instability,
in 1973 based on 1971 is 500, and the value for 1971 based on 1968 is 400,
the chain index for 1973 based on 1968 is 400X500= 2,000 (divided by 100,
since the index numbers were based on 100 as a standard). (see: index
number)
1 13
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
chance--the quality of occurring randomly, or through unknown
causes that makes a particular political event unpredictable. (see: rundo,n)
chauice variation-variation in statistical obser'iations due to the action
of'random, as distinct from systematic or predictable, factors. (see: randonr)
channel drift-(see: distortion)
Chapman-Kolmogorov equations--a set of equations used in the theory
of stochastic (random) processes, giving the state of a system at a certain
time in terms of the known states at previous times. (see: stochastic,
system)
chi square-(pronounced "kyc square") a test determining whether
some observed distribution differs significantly from a purely random
distribution, A chi square test involves algebraic computation followed by
interpretation from a special chi square table found in the appendix to many
statistics texts. Scores tell what percentage of the time the observed
distribution could be reached purely by chance. If the data reaches a
confidence level of .95, for example, we may be 95 percent certain that the
observed distribution was caused by factors other than chance, bra^ause
random events would have produced this type of distribution only 5 percent
of the time. (see: confidence level, significance level)
circular formula, circular series-an operation which gives 'every item
in a series an upper and lower boundary by making the first itei also the
last. For example, in a series of six terms there are only five differences
between them. If, for analytical purposes it is desirable to have six
differences, the first term is made a "pseudo" seventh term. This is
equivalent to regarding the series as circular. Any resulting formula in the
analysis is also said to be circular. The device is used in serial correlation
analysis. (see: CORRELATION)
class, class boundaries, class interval, class frequency, class marks-the
total number of items in any set of data may be grouped according to
convenient divisions of the data to make subsequent analysis less tedious.
Such a group is called a class. The highest and iowest values in each class
are called the class boundaries, the difference between the highest and
lowest values is the class interval, and the mid-point of the class intervals is
the class mark. The number of items falling into each class is called the
class frequency.
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
cluster sainlpling-a sampling technique which selects a specified
number of random clusters, or groups of observations, for study. This
sampling procedure is esp',cially useful in cases where the cost of searching
for samples is relatively high, but the cost of analyzing selected samples is
relatively low. For example, in checking the condition of railroad ties,
inst,~,ad of sampling one tic in each of 1000 random locations, it may be
easier and cheaper to select only 100 random locations and then inspect 10
ties in each. (see: CLUSTER ANALYSIS, SURVEY RESEARCH)
coding-the classification and sorting of political or social data into
categories or ranges in order to simplify and improve analysis of the data.
In content analysis, for example, the units of analysis (words, themes, whole
message items) must be examined and assigned to categories such as
friendly, moderate, or hostile. (see: CONTENTANALYSIS)
coefficient-a multiplier. In 5Y (5 x Y), 5 is the coefficient and Y is the
variable. (see: continuous variable, discrete variable, variable)
coefficient of determination-(sec: R2, R-squared)
coefficient of reproducibility-in the GUTTMAN SCALING
technique, a w,'y of measuring whether the scale will be analytically useful.
A coefficient (correct responses divided by total responses) of .90 or better
indicates that the scale has indicated a common attribute of all the items on
the scale. This is regarded as an effective application of the Guttman
technique. (see: GUTTMAN SCALE)
cognitive dissonance-a psychological term for the normal human
tendency to reduce inconsistencies which may arise when new knowledge
conflicts with values, environment, or behavior, by either changing these
factors or by selectively ignoring the new facts. Cognitive dissonance theory
is particularly useful in leadership and decision-making studies. Analysis of
a leader's cognitive dissonance behavior patterns, for example, should
reveal whether that leader would opt for negotiations or hostilities when
confronted by a reviled traditional enemy with superior military
capabilities.
cohort-an aggregate of individuals who have had some common
experience during, the same time period. (see: COHORT ANALYSIS)
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
common factor-in factor analysis, t.ny factor which appears in two or
more variables. II' the factor appears in all variables, it is called it general
factor. II' it is common to it group of variable:; it is called it group factor.
(see: factor, PA CTOR A NA 1, Y,S'IS, variable)
complete system of equations-in econometrics, it term used to denote
ail the equations determining the behavior of an economic system or part of
such it system. (see: ECONOM1sTRIC'S)
computer---an electronic/mechanical mechanism which can perform
arithmetic operations, store information of various kinds, and compare
information with other information, It performs exactly the instructions it
is given, usually in the form of' a program which is simply a list of those
instructions. (see: program)
concept--a mental image or construct formed by generalizing from the
characteristics or a class or category of things. A concept is an abstraction
to which it descriptive label is attached, and the label may then he applied to
individual members of the class to which the concept refers.
conceptual framework-a set of related concepts that provides an
analytic scheme for political research and gives the analyst an overview of
his research material, so that data can be systematically sorted, arranged,
and examined. (see: concept, methodology, approach)
conditional probability-the probability of occurrence of one event,
given that some other event has occurred or will occur. P(A/B) is read "the
probability of A, given that B has occurred or will occur." For example, w;,
may assess the probability that the Redskins will win this weekend (event A)
as 80 percent, or P(A) = .8. But we may think that the probability of them
winning if Larry Brown is unable to play (Brown's injury being event B) is
lower, say, 60 percent or P(A/B) = .6. If, on the other hand, the events are
independent, then they do not influence the probability of each other. In
that case, the conditional probability of one, given the other, is still equal to
the original probability of the first. For example, the probability that the
Redskins will win (80%) is independent of whether or not " go to the game
(my presence being event C.) So, P(A/C) = P(A, = 80% or .8. (see: joint
probability)
Approved For M ease 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
c'mfldence level-the statistical probability derived from a significance
test :ouch is chi square. It tells how often the observed behavior would occur
sorely due to chance, and how certain we may be that non-random causes
,are operating. (see: significance test)
confluence analysis-a method of overcoming certain difficulties in
REGRESSION analysis when er"ors of observation produce spurious
linear relations - n the observed independent variables. (see: REGRESSION,
independent variable)
confounding-limiting the size of samples in large FACTOR
ANALYSIS experiments by omitting some of the cases which relate to
particular interactions deemed unimportant or of little practical significance
from a policy point of view. (see: FACTOR ANALYSIS, block)
constraining equation--a mathematical expression of a resource
limitation beyond which no conceivable solution is feasible. For example, in
computing possible US GNP levels, one constraining equation might
express the fact that the number of possible man-hours of labor input is
limited by the population size. (see: of jective function)
contingency table-an array of data itt rows and columns. The simplest
contingency table is a 2x2 matrix, illustrating two dichotomous variables
and the frequencies of observation of each. For example, a 2x2 contingency
table may show 20 observations of national unemployment and inflation
rates. It shows, for example, that there were 6 cases when inflation of
greater than 6 per cent was accompanied by unemployment of under 4 per
cent. (see: array)
Annual Inflation Rate
More than Loss than
6% 6%
More than
Notional 496
Unemployment
Rate
Lass than
4%
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
continuous variable-a variable which can assume fractional values
as well as whole numbers. It describes quantities; which may be infinitely
divided, like distances or time, as opposed to discrete variables, such as
men or machinery, which c, -,,n not be fractions. (see: discrete variable,
integer, real number)
continuum-a continuous range of related phenomena located along a
line between two extreme values of the characteristic to which they are
related. Events or behavior ranked on a continuum can be compared or
quantified by assigning numerical values to each position along the
continuum.
control group-a group representing the standar or normal level of
behavior or activity wP:ich is not subjected to experimentation but is
compared to the experimental group in order to assess the results of change.
covariance-the association between two variable s, as shown on a
scatterplot. CORRELATION is a measure of covariance, REGRESSION
is a description of it. (see: CORRELATION, REGRESSi ON)
coverage-used in sampling in two senses: (1) to denote the scope of the
material collected from the sample members; (2) to mean the extent or area
of the population examined. (see: sample, population, SUR VEY
RESEARCH)
Cramer's V-a chi square based measure of association used to
determine the strength of relationship between two tables of different sizes.
Cramer's V varies between 0, meaning no relationship, and 1, indicating
maximum relationship. (see.: measure of association, chi square)
critical region-one of two mutually exclusive regions in a sample
space used in the testing of statistical hypotheses. If the sample point falls
into one (the region of acceptance) the hypothesis is accepted; if it falls into
the other, the hypothesis is rejected. The second region is customarily
denoted as the "critical region." (see: sample space)
cross-correlations-correlations between series ordered in time, or
space. Thus, if v,, and v2, and v3 are variables in one series and w,, and w2,
and w3 are in another, correlations between v1, and w, and between v2 and
w2, etc., are cross-correlations.
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
culture--the aggregate of learned, socially transmitted behavior
patterns characteristic of a group, which is developed and maintained
through formal and informal learning, language, knowledge, folkways,
beliefs, customs, traditions, and institutions. A political system is shaped by
related cultural factors, and may it turn, promote cultural change by
influencing the behavior patterns ol'the society.
curvilhecar-a relationship between two variables displayed on a graph
by curved lines. (see: linear)
cybernetics-originally a science dealing with complex electronic
calculating machines, cybernetics has evolved into the study of information
channels, decision-making processes, feedback mechanisms, and other
methods by which a government or organization perceives, responds, and
en forces. (see: COMM UNICA TIONS THEORY
J) value-the largest absolute difference between the cumulative
observed frequency of a variable and its cumulative expected frequency (or
frequency if the variable were governed solely by chance), which is used in
determining the significance of ordinal data. (see: Kohnogorov-Smirnov
D, data-ordinal, significance level, variable)
data-the facts, statistics, and other forms of information that provide
the raw material for analysis. There are four levels of data. Ratio scale data
have a true zero point and equal intervals between data- for example, data
on income levels from $0 to $100,000 with an equal interval between each
dollar. Interval data lack a true zero point but have equal intervals.
Temperatures are an example of interval data because the interval between
each degree is the same but there is no true zero point. Ordinal data lack
both a true zero point and equal intervals. The class rank of students is of
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
this typo---in a class of 100, for example, the student ranked 100 is not
necessarily twice the achiever of the student ranked 50. Nominal data do
not even meet the requirement of order met by the three previous forms of
data. The data may be "yes-no" answers to questions, or data arbitrarily
assigned numbers for coding purposes (as localities in the US are assigned
zip codes).
decision function-a rule of conduct which, at any stage of a sampling
investigation, tells the statistician whether to take further observations or
whether enough information has been collected to perform the statistical
tests he wishes to make. (see: sample, SUR VEY RESEARCH)
decomposition-the breaking down of a problem into its constituent
parts. Used in time series analysis to mean the act of splitting a time series
into four-a long-term movement or trend; oscillations of more or less
regular period and amplitude about this trend; a seasonal component; and a
random or irregular component-by the use of statistical methods. (see:
TIME SERIES' A NA L YSIS)
deduction-the process of reasoning from the general to the specific
and drawing conclusions from applicable premises. Deduction is used in
political analysis when the primary objective is to explore and understand
the implications of premises or to formulate hypotheses based en what is
already known; the inductive method is used to test hypotheses once formed.
(see: induction, hypothesis testing)
degrees of freedom-the number of elements in a set of data that can
vary and still permit certain conditions to be met. Generally, the number of
degrees of freedom is one less than the number of observations in the set. If
four elements in a given set lave a mean of 16, for example, the first three
may take any values-e.g., 4, - 21, and 68-totalling 51-but then all three
available degrees of freedom have been used. The fourth value must now be
13 in order for the mean of 16 to be met (51+ 13=64/4= 16). The number of
degrees of freedom is used in chi square and other statistical operations.
(see: chi square)
delta (A)-an increment of change. If variable x goes from a value of 8
to a value of 14, the change in x (read, "delta x") is 6 (x, -x2 =Ax). At
often symbolizes a charge in time. Delta is also shown as d, especially in
the formula for the slope of a line; dy/dx, where the slope equais the
change in y divided by the change in x. (see: slope)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
demand-a measure of an individual's or a society's willingness and
ability to purchase and consume a good at a given price. (see: supply)
dependent variable-the event or condition the analyst wishes to explain
or predict in any research problem, as opposed to the independent variable,
which is the predictor or explicator. The relationship between dependent
and independent variables is not necessarily causal, but causality is usually
inferred. (see: variable, independent variable, intervening variable)
derived statistics-obtained by an arithmetical operation from the
primary observations. For .-xaniple, population figures are primary and so
are geographical areas, but population-per-squat.; ile is a derived statistic.
descriptive statistics-a branch of mathematics designed to summarize
or describe the important features of a large amount oi' data without
inferring anything that pertains to more than the data, as opposed to
inferential statistics which involves generalizations, predictions, and
subjective evaluation of the data. (see: inprential statistics)
descriptive survey-a sample survey where the principal objective is to
estimate the basic statistical parameters (means, totals, ratios) of the
population or its sub-divisions. (see: SUR VEY RESEARCH, parameter)
deterministic model-an unchanging model which contains no random
elements and for which the future course of the system is determined by its
structure at some fixed point in time. (see: model)
development theory-the study of growth and change within political,
social or economic systems, or change from one system to another,
generally toward greater governmental capacity to cope with the demands
made upon it. The terms traditional, transitional, and modern are
commonly used to designate societies in different stages of development.
deviate-the value of a variable measured from some standard point Ur-
location, usually the mean. (see: variable, mean)
deviation-(see: standard deviation)
diachronic analysis-(see: TIMESERIES ANALYSIS)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
dichotomous clas,4f1cation-the ordering of' phenomena into two
mutually exclusive groups (military/ non-military, for exampla). The
purpose of' such classification is to gain an understanding of political
phenomena by relating them to overall descriptive categories for purposes
of contrast and comparison.
digital computer-a computer which handles data in numeric form and
processes data by operations based on counting. (see: analog computer)
direct sampling-a term used when the sample units are people and not
some kind of record relating to them, such as census forms. (see: SUR VEY
RE, SEARCH, sample)
discrete variable-a variable which can take only integer or whole
number values-the number of weapons in a cache, for example-as
opposed to a continuous variable which can be either a whole number or a
fraction such as the average age of enemy soldiers fighting each month. (see:
variable, continuous variable, integer, reel number)
discrimination-in multivariate analysis, a method for allocating items
known to come from two or more populations to the correct population with
a minimum amount of misclassification. (see: multivariate analysis, sample,
population)
dispersion-the degree of heterogeneity of a set of observations. T!1ete
are various numerical alternatives for measuring dispersion, inc;uding
range, mean deviation, and standard deviation.
dispositions-personality traits that shape the individual's values,
attitudes, and opinions toward political phenomena and his actions in the
political arena. (see: personality trait)
distortion-occurs when information received by a system is changed
or misinterpreted before it can be acted upon. Amplification of positive
feedback is an example of distortion which involves an upward or dow:iward
spiral of self-reinforcing actions which lead to further actions of t.';z same
sort. Negative-only feedback, the opposite of amplification involves
self-corrective actions which keep the system in equilibrium. Channel drift is
a form of distortion where a political system attuned to one channel of
information input may gradually shift toward another channel with
unintended system consequences. Such consequences include, for example,
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
the societal and political changes which occur when newspapers begin to
replace word-of-mouth communication in lesser developed countries, or
when electronic media replace m.wspapcrs in developed countries,
Short-circuiting is a distortion where data is not subject to all phases of the
input-transformation-output system. (see: SYSTEMS ANALYSIS,
COMM UNICA TIONS THEOR Y)
distribution-the grouping of large sets of data into a number of classes
by size, frequency, or some other variable. The information can be
presented in a chart, table, or graph. When presented in graph form, it is
known as a distribution curve (see: frequency, frequency polygon, ogive,
normal distribution)
FREQUENCY
DISTRIBUTION
I.Q.
SCORES FREQUENCY
0 - 40 5
41 80 20
81 120 50
121 - 160 20
161 - 200 5
Dollard-Doob hypothesis-(see: frustration-aggression hypothesis,
CONFLICT BEHA VIOR ANAL YSIS)
domain-the values that an independent variable may take.
dominating strategy-a strategy in game theory which is in no respect
worse than all other strategies and in at least one respect better. (see:
GAME THEORY)
down cross-the point where a time series, measured about its mean,
changes in sign from positive to negative. Correspondingly, a point where it
changes from negativc to positive is called an up cross. (see: VME SERIES
ANALYSIS, mean)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
dyad-a linked pair of entities-two states, leaders, political syst';ms,
interest groups, departments within the government, etc.-that exhibit in
common some variable which is under analysis.
dynamic model-a mathematical computer model which is not derived
statistically from time-series but which starts with a statement of system
structure and by looking at multiple feedback relationships determines the
dynamic consequences when thr, assumptions within the model interact with
one another. A dynamic model ,foes not primarily seek to optimize decision
making or maximize system performance as in linear programming models
but rather to identify significant causal relationships to help the analyst gain
additional insight into system behavior. (see: feedback loop, system,
LINEAR PROGRAMMING)
efficiency-the capacity to produce desired results with a minimum
expenditure of resources; or the ability to combine given resources to
maximum advantage.
eigen value-in FACTOR ANALYSIS, the sum of the factor loadings
squared which measures the amount of vaJation in the variables ; !counted
for by the factor. (see: factor, loading, FACTOR ANALYSIS)
elasticity-a measure of the sensitivity of one variable to change in
another-commonly used in economics to demonstrate the change in
demand or supply in response to price variation. The formula for computing
elasticity of demand, for example, is the percentage change in quantity
demanded divided by the percentage change in price. Thus, if a 20 percent
price drop causes a 20 percent boost in demand, elasticity is equal to one,
and the total amount spent on the good (equal to quantity times price) stays
constant. If a 20 percent price cut causes a greater than 20 percent rise in
demand (so that th,; total amount spent on the good rises) elasticity is
greater than one, and demand is said to be "price elastic." Demand is
inelastic when a 20 percent price cut causes a less than 20 percent rise (and
total revenue falls).
empiricism-pursuit of knowledge by observation and
experimentation. In political science, empirical theories (what is) can be
contrasted with normative theories (what ought to be). In some political
writings, empiricism is broadly equated with the use of scientific method or
the behavioral study of politics. (see: BEHA VIORALISM, method)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
endogenous variable-a variable whose value is dependent upon and
determined by the process under study. For example, in an election model
predicting party votes from unemployment and religion, the demographic
('actors are exogenous, and the voting results are endogenous. (see:
exogenot-v variable, dependent variable)
engineering theory-often referred to as "policy science," engineering
theory in political science tries to provide answers to political problems
basically through the use of ends-means analysis. If policy makers are faced
with the question of how to deal with rising unemployment, for example,
they might develop an engineering statement that details the means for
meeting that problem. Such a statement would set forth the variables
relevant to achieving full en4nloyment and the manner in which those
variables might be manipulated in order to bring it about. (see:
ENDS-MEANS ANALYSIS, COS T-BLNEFI TA NALYfIS)
equilibrium-a state of balance ascribed to a political, economic, or
social system. One form of equilibrium analysis assumes that environmental
influences tend to affect the relationships within a system, constantly
moving them away from, then back toward, a presumed preexisting point of
stability. Equilibrium analysis sometimes assumes a dynamic
equilibrium-that is, one in which no fixed point of stability exists but the
elements of the system remain in some kind of balance with one another
while the position of equilibrium fluctuates from one point to another. (see:
SYSTEMS ANAL YSIS)
events data-information concerning the behavior of individuals or
nations. It is usually coded as input to a systematic data base or file and is
dependent upon time. (see: EVENTS ANALYSIS)
exchange theory-seeks to explain political behavior in terms of
market principles derived from economic mod,.',. Politics is, therefore,
viewed as a process of resource distribution based on rational calculations of
costs and benefits. Exchanges constitute a common form of political
interaction. On the international level, for example, where each state may
carry on interactions with more than 140 other states as well as numerous
non-state systems, analysis of exchange levels-goods for goods, or deeds
for deeds-can help bring economic insights to bear on a wide range of
political behaviors and contribute to the development of typologies of state
actions.
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
exogenous variable--a variable whose value is determined outside the
model being studied, and is independent of it but which may influence
variubks in the model. A model of energy demand in Japan, for example, is
affected by the exogenous political variable "oil decision" made by Middle
Eastern exporting countries. (see: endogenous variable, independent
variable)
expected value-a key concept in probability theory, statistics, and
game theory. The expected value of a strategy is obtained by estimating the
probability of each possible outcome, and multiplying each of these
probabilities times the value (utility) associated with that outcome. The sum
of these multiplications is the expected value of the strategy, and
risk-adver e player will normally select the course; of action with the highest
expected value. For example: Strategy I has three possible outcomes:
Outcome
Probability
Payoff
A
.6
30
B
.2
80
C
.2
-20
Strategy 11 has two possible outcomes:
D .5 60
E .5 -to
The expected value of Strategy I is:
(.6 x 30)+(.2 x 80)+(.2 x -20)=30
The expected value of Strategy 11 is:
(.5 x 60)+(.5 x - l0)=25
exponent-the power to which a number is raised, that is, the number
of times the number is multiplied by itself. In the equation, 23=2 x 2 x 2=8,
the 3 is the exponent.
exponential curve-a series of observations ordered in time, as
displayed on a graph or chart, which increases geometrically at a constant,
or approximately constant :ate. In an arithmetic progression, each
succeeding number is derived from the preceding one by the addition of a
constant quantity. In a geometric or exponential progression, on the other
hand, each succeeding number is derived from the preceding by multiplying
it by a constant quantity. If the constant is the series 2, 4, 6, 8, 10 would
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
form an arithmetic progression; the series 2, 4, 8, 16, 32 would form a
geometric progression. Assuming the numbers represented population, the
two series could be granhcd as below, Typically population growth would
follow the geometric ;rr,ionent,ial) curve.
2 3
GENERATIONS
F test-a statistical test designed to tell whether two samples are
significantly different from each other. The test measures the ratio of the
variances of the two distributions, then uses an F table to tell how often such
a difference would occur solely by chance. The process is similar to both the
chi square and t tests, which tell how confident we can be that the observed
changes are not random. (see: confidence level, significance level)
factor-in statistics, (I) a quantity under examination in an experiment
as a possible cause of variation, or (2) an item in an average or an index
number. (3) As adapted from psychology and used in multivariate analysis,
it denotes an impact of the observed variables which may be regarded as part
of those variables. (4) In FACTOR ANALYSIS, a c^?nmon condition
underlying a large number of variables or people is a factor. By studying a
set of UN votes, for example, we may identify which votes belong
together. Thus, industrial countries, former colonies, Arab states, or US
allies may emerge as voting blocs or underlying factors in UN votes. (see:
multivariate analysis, variable, FACTOR ANALYSIS)
factorial (!)-the product of successive numbers always ending with
one. For example, 7! (read "seven factorial") equals 7 x 6 x 5 x 4 x 3 x 2 x I
= 5040.
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
fair gnnte---in the theory o1' games, it game consisting of a sequence of
trials is dccmcd "fair" if the cost of each trial is equal'' to the expected value of
the gain from each trial. In a contest between two advcrsai ics with unequal
resources, or example, it is deemed "fair" that the player A with the larger
suns to stake has it better chance of ruining his opr)oncnt 13 because A can
lose it larger sum from the game. (see: GA M7'IIEORY, expected value)
fallacy of composition-the logical error of assuming that what is true
For one individual is necessarily true for the whole group, For example, one
individual at it football game can get it better view of the field by standing
up, but if' everyone tried this, they would all be likely to have the same, or
even it worse, view than before. This fallacy slips easily into cconomirs: in
group savings and in national debt, for example, what is true on it micro
level is not necessarily true nn it macro level.
fallacy of division-the logical error that occurs when one erroneously
assumes that characteristics of a group necessarily exist in each member of
that group.
false dilemma--a logical ft,liacy which omits consideration of some
available alternatives. Unfairly dichot.omiz,,ng the possibilities to the two ex-
tremes, for example, may obscure moderate solutions and make a problem
appear more perplexi,ig than it really is.
feedback loop--iii systems analysis (particularly systems dynamics) a
circular, causal chain through which the past behavior of a system affects its
future behavior. This may be contrasted with an open loop system, where
outputs depend on inputs but inputs are not influenced by past behavior.
In the positive feedback loop, the causal chain acts to reinforce change,
either up or down. Examples include the growth of deposited money via
compound interest, arms races, the extinction of a species, and industrializa-
tion. Causal chains which resist rather than reinforce change are negative
feedback loops. Given a disturbance pushing a system away from
equilibrium, the negative loop responds in the opposite direction forcing the
system back towards equilibrium. (see: SYSTEMS ANALYSIS,
SYSTEMS D YNA MI CS)
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
feedback system-u system in which the el'I'ects of the outputs are
measured and the measurements ore then reintroduced as inputs. This
allows the system to constantly monitor its own performance, and adjust to
the conditions of the environment even as it influences them. The classic
feedback mechanism is the furnace thermostat which reads the room
temperature as it adjusts it, in order to decide how r, Bch more heat is re-
quired.
field-in computer jargon, a column or set of adjacent columns on a
computer punch card which are allocated to represent one piece of data or
instruction, For example, a one column field may be punched 0 or I to in-
dicate yes or no, and the year 1974 would fit on a four column field.
In field theory, the field" is a chart upon which actors are plotted according
to their relationship to the variables under study and according to their
rclation;hip with other actors. (see: actor, FIELD TIIEORY)
Fisher's exact test-a test of significance for small samples (less than
20 observations) divided into a 2 by 2 table, which gives the exact probabil-
ity of getting the same breakdown by chance as occurs in the table. (see:
significance 'eve!, chi square)
fixed-sum game-a class of problems dealt with by game theory in
which one player's gains are exactly equal to the other player's losses. This
describes a perfectly competitive situation in which no cooperation is possi-
ble because the sums of the payoffs in each matrix cell are held to some con-
sti,nt level. (see: variable-sum rame, zero-sum game, GAME THEORY)
fluctuation-any movement up or down between consecutive items of a
series of numbers; the variation of a statistic from sample to sample.
folding-regarding all the values of a set ct' data as having the same
sign-usually plus-in order to perform certain statistical operations which
are complicated by the use of different s gns in the data.
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
forecasting-prediction, in the customary sense of assessing 1, he
magnitude it quantity will assume at some future point in time. Distinct
from "estimation," which attempts to assess the magnitude of an already ex-
isting quantity. For example, the final yield of a crop is forecast during the
year and estimated at harvest.
fractile--denotes the one value in a set of data above or below which
lies a given fraction of the distribution. For example, quartiles divide the
distribution into four parts, deciles divide it into 10 parts, and percentiles
divide it into 100 parts.
frequency-the number of occurrences of a given type of event, or the
number of members of a population fa,sing into a specified class. Where the
frequency is expressed as a proportion of the total number of occurrences or
the total number of members, it is called the relative or proportional fre-
quency. Wijen the frequency of each occurrence is added to previous fre-
quencies, the result is a cumulative frequency. (see: distribution, frequency
polygon, ogive, normal distribution)
/.Q.
Proportional
Cumulative
Scores
Frequency
Frequency
Fre
uenc
0-40
5
1/20
q
y
5
40-80
20
1/5
25
80-120
50
1/2
75
120-160
20
1/5
95
160-200
5
1/20
100
frequency distribution-(see: distribution)
frequency polygon-the outline or silhouette of a histogram. It
graphically displays a distribution of data classes and by frequency of
observation. Cumulative frequency, presented in graph form, is called an
ogive. (see: histogram, distribution, frequency, ogive)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
- 1113rocu1AM
ao
30 FNCQUENCY
POLYGON
90
10
frustration-aggression hypothesis-the assumption that frustration
always leads to some kind of aggressive reaction whether explicit or im-
plicit. (see: CONFLICT BEHAVIOR ANALYSIS)
function-in statistics, a mathematical description of the relationship
between two variables; a rule describing the value of one variable if given the
value of the others. For example, the function Y=6X+3, tells how to derive
the corresponding Y value if given any value of X.
In structural-functional analysis, structures are patterns which govern ac-
tions-including laws, constitutions or such concrete organizations as
political parties and legislatures-while functions are the results or conse-
quences of actions which satisfy the needs or demands of the system for its
own maintenance. Defense of the state, stabilization of the economy, and
execution of laws, for example, are political functions. (see: structure,
variable, STRUCTURAL-FUNCTIONAL ANALYSIS)
fundamental probability set-a set of objects or events which are basic
to a probabilistic situation, in the sense that all other objects or events under
consideration are derived from them. It follows that all probabilities are ex-
pressible by the rules of addition, multiplication, etc., in terms of the funda-
mental set.
gain-in communications theory, gain measures the effective as
opposed to the ineffective changes a system makes as a result of the
communications it receives. (see: distortion, lag, COMMUNICATIONS
THEORY)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
gambler's ruin-the name given to one of the classical topics in
probability theory, in which a gambler wins one predutc:rmincd sum of
money for every success and loses a second predetermined sum for every
failure. The play proceeds until the initiut capital ol'one player is exhausted
and he is ruined. The statistical problems involved Lire concerned with the
probability of the ruin ol' a player given the stakes, initial capittWl and chance
of success, and with such matters as the length of play. Variations of this
problem can be used in war gaming. (see: probability, GAME THEOR Y)
game-(see: GA ME THEORY)
gamma-(see: Goodinan-Kruskal gamma)
general factor-(sce: common factor)
geometric mean-(see: mean)
geometric progression-(sec: exponential curve)
Gestalt psychology-a psychological approach which basically argues
that the whole is greater than the sum of its parts. That is, the examination
of a given phenomenon in its totality yields different results or
interpretations from those yielded by examination of the different parts that
constitute it. More technically the Gestalt psychologist objects to treating
the nervous system as a static, machinelike stricture capable only of
responding piecem^al to incoming stimuli. Rather the cerebral cortex is
viewed as analogous to a force field which is in active equilibrium and in
which each incoming stimulus affects the entire field.
Similarly in political science analysis the question is whether relationships
and patterns can best be discovered by using an incremental Bayesian
approach in which the analyst reacts to each individual piece of evidence or
by a holistic approach in which the analyst gains insight about political
reality by looking at and drawing conclusions from the totality of relevant
evidence. (see: FIELD THEORY)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
(1ni coefficient--a measure of ine(luality, describing the uniformity of
the distribution of some attribute (income, power) over some population
(individuals, states), The index is equal to twice the area of difference
between a Lorenz curve and the line of perfect equality, The coefficient (G)
ranges from 0 (representing perfect equality in the distribution for everyone
in the sample), to I (representing perfect inequality ---one individual
possessing everything and the rest of the population possessing nothing).
(see: coefficient, Lorenz curve)
20 40 60 60 100
% OF POPULATION
Goodman-Kruskal gamma-a measure of association for comparing
ordinal level data in tables. Gamma varies between 0 and a maximum of I
based on the probability of predicting one quality in the table given the
other. (see: ordinal)
goodness-of-fit-in general the agreement between an observed set of
values and a second set which are derived wholly or partly on a hypothetical
basis, that is to say, derived from the "fitting" of a model to the data. In the
process of validating models, for example, goodness-of-fit tests are used to
compare the model's forecast with real world observations in order to infer
the degree to which the model comforms to reality. (see: chi square,
Kolmogorov-Smirnov test)
grade-the proportion of the total frequency that items occur in a
continuous series with values less than or equal to the particular item being
"graded." If we wish to grade item 494 in a series of 53. items, for example,
its grade is 494i53 1, or 93%.
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
grid sampling-u form of cluster sampling, the clusters in this case
being individual areas of a grid and hence consisting of groups of basic cells
arranged in some standard geometric pattern. (see: SUR VE 3'
RESEARCH, cluster sampling)
group-a set of'elements, individuals, or observations, distinguished by
some common attribute or shared relationship. In group theory, the
collectivity of individuals may be categorized in many ways. A formal or
organized group has recognized goals and structures affecting group
interaction. An informal group, such as like-minded politicians meeting at
lunch to share views, lacks such expl;cit goals and organization structure.
A primary group is a relatively small collectivity that engages in frequent
and direct interactions (families, work groups, etc.) while a secondary group
interacts infrequently and communicates mainly by indirect and impersonal
means. A categoric group consists of individuals who do not necessarily
interact but share some common attribute, such as income level, religion,
etc.
The label interest or pressure group is applied to collectivity that seeks to
influence governmental policy in favor of some goal or shared concern of
the group. (see: GROUP THEORY, interest group)
group factor-(see: common factor)
harmonic analysis-a regression analysis of data which, when graphed,
has periodic, winding components. The harmonic analysis finds that straiL.,`it
line which can be drawn through and will best fit all the data. (sec-,
REGRESSION)
harmonic mean-(see: mean)
a
3
2
1
0
ORIGINAL DATA GRAPHED
4
3
2
0
HARMONIC ANALYSTS GRAPH
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
heuristic-intellectual aid used in the search for new ideas and insights.
The use of models as heuristic devices, for example, contributes to an
understanding of politics because it permits the analyst to compare the
model with real situations. (see: model, SIMULATION)
high-low graph-a form of graph used to depict ranges of variation in
successive intervals of time. For example, daily price variation might be
represented by taking time intervals of one month on the horizontal axis
and, at each monthly point, showing the maximum and minimum price
attained during the previous month. The higher' prices can then be joinea y
a line drawn from high point to high point; similarly the low prices. Or, for
each month, the high and low points may be joined by a vertical bar.
histogram-a graphic form of the frequency distribution in which the
number of cases within each class is represented by a vertical bar whose
height is proportional to tI. number of cases within the class. For example,
a histogram. describing how many cars passed a certain street on each day of
a typical week, -..;ght look like this:
M T W T
GAYS
pproved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
historical approsch-the traditional approach to political analysis in
which information i organized and interpreted chronologically.
Continuities with earlier epochs may be determined and linkages
established. The cumulative development of ideas during one tine span, for
example, may be studied to determine its impact on the institutions and
societal changes that occurred during a subsequent period.
Hollerith card-sometimes called it key punch card, it is divided into 80
columns and information is punched on to it to be read by the computer.
(see: Hollerith code)
Hollerith code-a pdnch card code in which the top three positions in a
column are called zone" punches, (labeled R, X, and 0, from the top down)
and are combined with the remaining 9 digit punches to represent
alphabetic, numeric, and special characters. The letter A, for example, is a
combination of an R and a I punch in Hollerith code.
holism-the theory that group properties or characteristics are distinct
from those of the individuals who comprise them. In political science, the
division in approaches has generally been between those; who regard the
group as the proper level for the study of politics, and those who focus on
the individual and his behavior as the basic unit of analysis. (see: GROUP
THEORY)
homeostasis-the tendency toward maintenance of stability in a system
through self-adjustments that provide compensating responses to disruptive
or destabilizing influences. It has been argued that a system which loses the
capacity to deal effectively with disruptive inputs faces the problem of
disintegration or the possibility of being transformed into another kind of
system. (see: SYSTEMS ANALYSIS, feedback loop, feedback system)
homogeneity-in statistics, a term used to indicate different
populations, or samples from different populations, that are identical in at
least one respect. For example, if two populations have the same mean
(average) they are homogeneous even if they have different dispersions. (see:
population, sample)
hypergeometric distribution-the distribution of a discrete variable (one
which can take only whole number values) with a finite numi;ai of cases. As
the number of cases tends to infinity, the distribution tends to the normal
distribution form. (see: discrete variable, normal distribution)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
hypothesis-a statement of an expected relationship between variables
that may be tested empirically to determine its validity. A hypothesis may
be derived from observation, deduced from a larger body of theory, or based
simply on a hunch that the analyst is willing to use provisionally.
hypothesis testing-the attempt to confirm or disconfirm a factual
proposition, or hypothesis, by gathering and analyzing relevant evidence.
Testing refers both to the use of certain 'ypes of measuring instruments,
such as intelligence tests or attitude scales, and to statistical tests of the level
at which observed relationships in data are sufficiently different from
chance relationships (level of significance). In the latter case, a null
hypothesis, or opposite of the original hypothesis, is iol med with the hope
that it can be disproved. If the null hypothesis can be rejected, confidence
in the original hypothesis is increased. (see: null hypothesis, significance
level, validation)
hypothetical population-a ;;tatistical population which has no real
existence but is postulated to be generated by rep^titions of a given event
(such as throws of a die, aircraft failures, elections, etc.), The hypothetical
population is then used as a control group for comparisons with the actual
population data.
ideal type-a concept whose characteristics are represented in so pure a
form that examples of the concept are rarely if ever found in reality. Ideal
,types are used by political analysts in GAME THEORY, model building,
SIMULATION, and the development of general theory and are used as a
base for judging, explaining, or investigating reality. An ideal type of a
supply-demand capitalist economy can be created, for example, in order to
determine how far an actual system departs from the model. (see: GAME
THEORY, model, SIMULATION)
idiographic explanation-an explanation based on specific propositions
that explain unique or individual social or political phenomena as opposed
to a nornothetic explanati,)n which is based on general propositions
applicable to classes of events or objects. History i, often regarded as
essentially an idiographic discipline, while political and other social sciences
that emphasize explanation by theoretical generalization are more
nomological.
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
illusory association or hiusory eorr' Ltion--an association between
attributes which is statistically significant but involves no direct causal
connection. If in Europe, for example, the possession of blonde hair was
found to be positively correlated with an ability to ski, the association would
be illusory in the sense that either attribute does not cause the other. While
such an association might be statistically valid, it is due not to causality but
to the accidental circumstances that the blonde nationalities inhabit the
northern European countries where skiing is common. (see: attribute,
association, CORRELATION)
image theory-concentrates on the analysis of a nation's perceptions of
foreign societies as transmitted from generation to generation through the
cducation system, folklore, news media, etc., to ascertain likely reactions to
different cultures. The theory emphasizes that impressions of foreign
nationalities are . -med in childhood and are not imposed upon people by
their leaders, although leaders can and do manipulate impressions,
stereo-types, and prejudices.
impact panel-in panel studies, an experimental technique which
interviews individuals before and after a planned event in order to ascertain
the impact of the event upon attitudes.
independent events-events which do not influence the probability of
occurrence of each other. Fcr example, the probability of obtaining heads
on the second toss of a coin is :?;uependent of the result of the first toss. (see:
P.' utually exclusive events, conditional probability)
independent variable-a variable whose cha;. es are used to predict or
explain changes in some other variable called the dependent variable. (see:
variable, dependent variable)
index number-a figure which discloses the relative change, if any,
of prices, costs, or similar statistical phenomenon from one period
of time selec:-ed as the base period. The base period usually is assigned
the index number 100 and any changes from it represent percentages.
For example, if an index of cost of living with the year 1969=100 as a base
rises to 150, the cost of living has increased by 50% since 1969.
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
indicator-an event, observation, or value used to measure an abstract
c-)ncept. All abstractions must be measured indirectly-by indicators. For
example, the number of riots or the crime rate might be used as indices of
the abstract concept-civil order. (see: validity, variable)
indifference curve-(!so-preference line) a graphic description of
various combinations of two goods which will leave the consumer equally
satisfied. Used in economics, or in any analysis involving trade-offs: guns vs.
butter, unemployment vs. infla+:ior, or new car vs. home improvement;
where limited resources mean O?i~!t more of one type of goods can only come
at the cxpensf-, of less of the other type. An indifference curve shows how
much more of one type of goods a customer would need in order to give up
one unit of the other and still feel equally well off. In a set of indifference
curves, each one represents a different level of overall utility or well-being,
and individuals seek to attain the highest one, subject to budget constraints.
Along any given curve, the individual is indifferent-he feels he would just
about break even with any combination, and he has no preference between
them. Here, points (a,, bi) and, (a2, b2) are on the same indiffE:rence curve:
the consumer will sacrifice some of goods B (measured by the quantity
b,-b2) in order to attain more of goods A (measured by the quantity
(a,-a2) or vice versa-he is indifferent between the two combinations. Point
(a3i b3), however, like all points on indifference curve II, is preferred to any
point on curve I
INCREASING
DESIRABILITY
indirect sampling-sampling from documents, or some record of the
characteristics of a population, rather than the recording of information
obtained first hand. A political analyst may find this method of sampling for
survey research techniques best suited to his needs since he is usually unable
to survey in the country he is studying. (see: SURVEY RESEARCH,
sample, population)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Induction-u process of reasoning which attempts to reach logically
valid generalizations from individual facts, cases, or events. Inductively
reached generalizations are never quite certain to be true or valid; they may
always be corroborated or disprovcn by subsequent observation. Induction
contrasts with deduction, which is a process of logically reasoning from
general statements or premises to conclusions about individual cases, (see:
deduction)
inductive statistics-(sce: inferential statistics)
inference-drawing a conclusion from facts and premises. In statistics,
inference is the process of drawing causal implications from statistical data.
An important distinction must be made between causality and correlation.
Statistical techniques alone can establish that two variables are closely
related and that they move together, but this does not necessarily
prove a cause and effect relationship. If we find, for instance, that
an increase in the number of teachers in an area has been paralleled by an
increase in the number of liquor stores, we can not "prove" that one has
caused the other. Rather, both may have been caused by a third factor:
rising population. While statistical techniques alone can not firmly prove
causality, the absence of any significant correlation between two variables
can be powerful it disproving a false hypothesis of causality. (see:
CORRELA TION)
inferential statistics-inferential statistics (inductive statistics) the
science of drawing causal implications from numerical data, and of making
inferences about a large group of individuals when only a small sample has
actually been observed. This is distinct from descriptive statistics, which
provides methods to collect, organize, and present information on an entire
group withow drawing implications. (see: descriptive statistics)
in.luence-the capability of a political actor to affect the behavior of
others in a manner favored by the actor. (see: actor, power, authority)
information system-in systems analysis, a set of interrelated structures
that receives an input of new data, processes this data by comparing it to
memory and previously established values, and then makes a decision about
the data. This decision leads to storage of the data in memory and, if
appropriate, to the implementation of actions which may in turn cause a
feedback into the information system. (see: SYSTEMS ANALYSIS, data,
distortion, feedback system)
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
input-any influence that affects the functioning of it system. The use
of "input" as it descriptive and analytical term in political science culls
attention to the dynamic linkage between the system and its environment
and between different elements of the system, Inputs rclatc directly to the
making of decisions and to the outputs that follow. Thus, in it political
system, inputs result in demands upon and support for the system which are
converted into outputs in the form of authoritative policies and
implementing actions.
Instrumental variable-u generic term to refer to tiny predetermined
variable that can be used to help identify and estimate the parameters of
a large system of equations. Instrumental variables are most frequently
used in large econometric models in which the number of variables is
greater than the number of available observations.
integer-a value which can only be expressed as a whole number, as
opposed to real values which can have fractions. The number of cabinet
members in government- 15, for example-is an integer, while the average
age of the cabinet mcmbers-57.3 years-is a real number (see: real number,
variable)
integration-the process by which .wo or more political units inc.-case
their cooperative contact with one another; the political merger of two
previously separate units. The condition of integration is also called political
community.
interactions-in events analysis and communications theory,
interactions are activities and events between individuals or groups, usually
across national boundaries, which have immediate political significance.
Official visits, military forays, or aid and trade flows arc examples of
interactions. The routine exchanges across borders such as mail flows or
tourists are termed transactions, and have less political significance. (see:
EVENTS ANALYSIS, COMMUNICATIONS THEORY)
intercoder reliability-a method for checking the reliability of the
coding process in content analysis by having several coders analyze the same
messages and by comparing the results. The amount of agreement is then
computed and expressed numerically as the index of intercoder reliability.
(see: CONTENT ANALYSIS)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
lntercorrelatlon- .ti, relationship of a number of variables among
themselves, as distinct from their relationships with an "outside" or
dependent variable. (see: variable, CORRELA77ON)
Interest group-(pressurc group) a collectivity that seeks to influence
governmental policy in favor of some goal or shared concern of the group.
Interest groups can be subclassified as associational (organized interest
groups), institutional (groups bused on major social institutions such as
bureaucracies, education, church), non-atisociational (groups lacking
continuity in structure or regular procedures for articulating interests such
as ethnic, regional, or class groups), and anomie (largely spontaneous,
temporary groups that express discontent through direct actions such as
demonstrations or riots). (see: group, CROUP THEORY)
interface-any mechanism or physical location where two different
objects or kinds of objects meet. In international relations, the boundary
between two countries is an interface. The term is often used to represent a
border or boundary that affects the flow of communications between the
entities involved. In computer terminology, the man-machine interface is
the computer terminal.
interpenetrating samples-two or more samples taken from the same
population by the same process of selection. The samples may or may not be
drawn independently of each other. (see: sample, population, SUR VEY
RESEARCH)
interval scale data-(see: data-interval)
intervening variable-an event or condition that affects the operation of
the independent variable on the dependent variable. If, for example, we are
interested in examining the effect of poetical propaganda on public
attitudes, time may be introduced as the intervening variable by allowing
some interval between the interviewing of respondents to ascertain the
extent to which their attitudes are shaped by propaganda. (see: dependent
variable, independent variable)
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
interviewer bias-bias in the responses or recorded information which
is the direct result of the actions of the interviewer. This bias may be due,
among other things, to failure to contact the right persons for survey; failure
of the interviewer to establish proper rapport with the informant, with the
result that imperfect or inaccurate information is offered; or to systematic
errors in recording the answers received. (see: SURVEY RESEARCH,
sample, bias)
inverse probability-the probability approach which endeavors to
reason from observed events to the probabilities of the hypotheses which
may explain them, as distinct from direct probability, which reasons
deductively from given probabilities to the probabilities of contingent
events. Bayes' theorem is an example of such reasoning from observation to
probabilities. (see: probability. hypothesis, BA YESIAN ANALYSIS)
isometric chart-a chart which attempts to depict three-dimensional
material on a plane, The distances on the three axes are measured on an
equal scale.
isomorphism-similarity in form or structure. In political analysis it
generally refers to the similarity between a conceptual model and the
institutional or behavioral attributes that it represents. If a model is to be
useful it must be in some degree isomorphic to the system it represents. (see:
model, attribute, heuristic device)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
iso-preference line-(sec: indifference curve)
of-curve--a strongly skewed frequency distribution curve in which the
mode (the value observed must often) occurs at one extreme of the range.
For example, in shr)wing how many speeding tickets a number of individuals
had been given during the past five years, a curve would show some
dispersion, but most of the individuals would have reeordcd zero violations.
(see: skewed distribution, normal distribution)
1 2 3 4
NUMDER OF TICKETS
joint probability-the probability that both events A and B will occur.
If A and B are independent, it is simply the product of their separate
probabilities. For example, the probability of a flipped coin landing heads is
.5; the joint probability of two heads in a row is .5 x .5=.25. (see: conditional
probability)
4
NUMDER OF 3
INDIVIDUALS
On thousnnds) 2
Kendall's tau-a statistic for measuring the strength of association
between ordinal level variables-that is, variables which lack a true zero
point and equal intervals. It is most often used to measure the association
between two variables as represented on a two-by-two table. When
computed, usually by computer, the tau ranges from 0, indicating no
association, to +1, indicating complete association. (see: data-ordinal,
measure of association, variable)
Riot Soverity
1 2
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
key punch card--a card, sometimes called a Hollerith card, divided into
80 columns for cii: ruing information from some source to a form that the
computer can user--.making the information "machine readable." (see:
Hollerith code)
Kolmogorov-Smirnov D-a statistic which measures the significance; (or
probability that the observed data would occur solely by chance) of ordinal
level variables (those which lack a true zero point and equal intervals). The
D value is the largest absolute difference between the cumulative observed
frequency of a variable, and the cumulative expected frequency if the
variable were governed entirely by chance. Once the D value is computed, a
table is consulted to find the corresponding level of significance of this
particular data. (see: data-ordinal, significance level)
kurtosis-a measure of how much of the total area under a frequency
distribution curve lies within plus or minus one, two and three standard
deviations from the mean. When a distribution is perfectly normal, it is said
to be mesokurtic, when it is peaked, it is leptokurric, and when it is flat, it is
platykurtic.
0
MEAN
laboratory approach-t.. conduct of scientific experiments or tests
under conditions arranges or controlled by the experimenter. In social
science research, the analyst aims at holding certain variables constant so
that he may systematically manipulate, observe, describe, and measure
the remaining variables. Simulations and small group experiments are the
principal applications of the laboratory approach in political analysis. (see:
variable, SIMULATION, GROUP THEORY)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
lag-in communications theory, the time interval between receipt of
the communication and the reaction to it, (see: COMMUNICATIONS
THEOR Y)
lambda-a statistic for measuring the strength of association between
two variables by giving the "probable reduction of error" between the two. A
lambda of .60, for example, means that knowing the independent variable
will reduce by 60% the number of errors made in predicting the dependent
variable, For example, knowing the size of a city where riots occurred might
reduce by 60% the number, of errors in predicting riot severity. (see:
measure of association, significance level)
latent function-the relevant but unintended consequence of activities
initiated to satisfy the needs or demands arising from individuals and groups
within a political system. The manifest function of a political machine, for
example, is to maintain and exercise political power, but its latent functions
may include meeting welfare needs and providing channels for social
mobility. (see: STRUCTURAL-FUNCTIONAL ANALYSIS)
latent variable-a variable which is unobservable but is supposed to
enter the structure of a system under study, such as power in politics, or
demand in economics. (see: variable)
law of large numbers-(se' : Bernoulli theorem)
least squares-the most common rul:; for drawing a REGRESSION
line. The least squares method specifies that the regression line should be
drawn as the "best fit" of he data on a scatterplot. It minimizes the
sum of the squared vertical distances from each point in the group to the
regression line.
. I J A a I a
LEVEL OF UNEMPLOYMENT
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
I
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
If the line shown in the chart is the proper least squares regression line, then
tho., sum of the squares of the distances represented by the dotted lines will be
smaller than that achieved by any other possible line on the graph. The lcast
squares method instructs a computer to derive this line, which describes the
observed data better than any other line, and can be used to predict the
dependent variable (number of riots, for example) if the independent
variable (in this case, level of unemployment) is known.
legitimacy-the quality of being justified or willingly accepted by
subordinates that converts political power into "rightful" authority. The
consensus that provides the legitimizing factor in the exercise of power may
be cultivated through the sanctity of tradition, by the devotion of people
to a particular leader, or by the acceptance of the supremacy of law. (see:
power, authority)
leptokurtic-a tail, thin frequency c!istribution curve. (see: kurtosis)
0
MEAN
Likert scale-a simple scale for scoring the attitudes of respondents
(called judges) according to their agreement or disagreement with
statements which measure a test variable-such as hostility toward a
neighboring country. Ordinarily there are five choices for each statement:
strongly agree, agree, uncertain, disagree, strongly disagree. Each statement
is then given a score, 1 to 5, equal to the average response of the judges. The
judges are also scored according to their average response and then ranked
according to their position on the test variable. (see: SCALING, scale,
variable)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
line sumpling-a method of sampling in a geographical area. Lines
are drawn across the area and all members of the population falling on
the line, or intersected by it, are included in the sample. (see: SURVEY
RESEARCH)
linear-involving a straight line; an equation without exponents.
2X+3Y=6 is a linear equation; X2+3Y=14 is non-linear. (see: curvilinear,
exponent)
linear model-a model in which the equations connecting the variables
are in linear form, that is, without exponents. (see: model)
linkage theory-studies the convergence of national and international
systems, the relationships that link the stability, functioning, institutions,
and goals of national political systems to variables in their extended
environment. (see: system, SYSTEMS ANALYSIS, variable)
load-in communications theory, demands which are communicated to
the communications system which challenge the efficacy of the system. The
demand for more rapid transmission of ideas from the elite in the political
system down to all other levels, for example, constitutes a load. (see:
COMM UNICA TIONS THEOR Y)
loading-in factor analysis, factors are made up of interrelated
variables. Loading is the degree to which any given variable correlates
with a factor as expressed on a scale from 0, indicating no correlation to
+1 for positive and -I for negative correlation. Factors are artificial con-
structs which have no inherent meaning; the analyst must infer what the
factors represent from knowledge of which variables an!. most heavily
loaded on them. If 3 of ten variables under consideration-age of party
member, years in party, and highest office held-are "loaded" on a factor
at the .75 or better le~ql, the analyst may surmise from the variables in-
cludcd that the new factor represents "seniority." (see: variable, factor,
FACTOR ANALYSIS)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
logarithm-an exponent, used in a system especially helpful in
simp;ifying computations involving very large or very small numbers. For
example, if B is a positive number (other than one) and BL = A, then the
exponent L is the logarithm of A to the base B. In symbols, if B" = A, then
L = Ing?A. For example, 23 = 8, so 10928 = 3. Also, 10939 = 2 because 32 =
9. instead of carrying out lengthy multiplications (12,650,000,000 x 683,000,
for example) a quick and close approximation can be gained with the use of
a logarithm table, found in the appendix of most statistics or mathematics
books. The log to the base 10 of the first numbar is 7.1021 and the log of the
second is 5.8357. We add these logarithms to multiply the bases, and their
sum, 12.9378, corresponds to 8,667,000,000,000, which is close to the actual
answer of 8,665,250,000,000.
long run-in economics, a time period long enough to permit a firm to
adjust its fixed plant facilities to conditions of changing demand. Such time
periods are conceptual rather than calendar measures-a small firm may
adjust its plant size fairly quickly, but in heavy industries or utilities, lags of
several years may occur before all the necessary adjustments may be made.
(see: short run)
longitudinal analysis-(see: TIME SERIES ANALYSIS)
Lorenz curve-a graphic representation of the distribution of some
attribute (income, power) over some population (people, states).
20 40 60 80 100
% POPULATICN
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
For example, Lorenz curve A shows that the lowest 20 percent of the
population in this income group receives 5 percent of the total income of the
society. Similarly, the lowest 40 percent gets 20 percent, and the highest
paid 10 percent of'the population gets 20 percent of the total income. Curve
"3 represents a somewhat more equal distribution, with the bottom 20 per-
cent earning 10 percent of the income, and the bottom 50 percent getting
35 percent. Line C is the line of perfect equality. We measure the area be-
tween this line and the Lorenz curve, then double it to derive the Gini index.
This ranges from 0 to I and is a simple measure of inequality. (see: Gini
index)
loss matrix-in decision theory a matrix specifying the economic loss
or gain incurred according to the various decisions which can be taken and
the various situations which can exist. (sec: matrix, DECISION THEO12 Y)
DECISIONS
A B C D
1
-5 +8 +3 +9
U)
Z
02
-3 -4 +4 +3
3
-2 +5 +1 -1
U)
4
+6 +5 +7 0
machine readable-ir.formation that can be processed directly by a
computer with no transfer of information from one medium to another.
Information on key punched cards or transmitted electrically and stored on
tapes or discs is machine readable. (see: computer, key punched card)
macro/micro analysis-alternative levels of analysis, the micro level
consisting of subunits of the higher level macro unit. In macro analysis the
researcher directs his attention to the collective, systemic, institutional,
composite, or group level. In micro analysis, he focuses attention on the
parts, subsystems, Components, or individuals that comprise the collectivity.
A study made of a legislature might constitute macro-level analysis
compared to the study of its committee system, but micro-level analysis
compared to the study of the larger governmental system of which it is a
part.
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
management by objective (MBO)-a system for establishing both the
output requirements and measurement criteria of each managerial position
and the periodic conversion of these requirements and criteria into
measurable time-bounded objectives which are linked to future planning.
The goal of M130 is to improve managerial effectiveness-that is, the extent
to which a manager achieves the output requirements of his position.
management science-the use of scientific and particularly
mathematical methods to assess the effectiveness of organizations from the
perspectives of managers. It is of greatest interest to political scientists in
the fields of public administration or urban affairs. (see: SYSTEMS
ANAL YSIS)
manifest functions;-in structural-functional analysis, the purposeful or
intended consequences of activities that r;atisfy needs or demands arising
from individuals anyu groups within a political system. (see: STRUCTURAL-
FUNCTIONAL ANALYSIS, latent function)
manifold classification-the final grouping of a population which has
been divided into a number of mutually exclusive classes according to some
characteristic and then each class subdivided by reference to some second,
third, etc. characteristic. (see: population)
marginal (marginal cost, marginal revenue, marginal utility)-the term
"marginal" refers to the increment added to the total cost, revenue, or
utility by the last extra unit. This is "marginal" in the sense of "last," not in
the sense of "unimportant." Marginal cost is the amount it costs to produce
the last unit of' a good-a perfectly competitive firm will produce until
marginal cost equals price. Marginal revenue is the change in total revenue
due to the last sale-a monopolistic firm will produce until marginal
revenue equals marginal cost (usually at lower production and higher sales
price than a perfectly competitive firm). Marginal utility is a psychological
concept, not directly measurable. It refers to the change in total satisfaction
or happiness resulting from the last unit of consumption. "The law of
diminishing marginal returns" refers to the fact that marginal utility
generally declines after a certain level of satisfaction has been .eached. A
second car, for example, may be valuable to a family, but not as valuable as
the first, without which they would be immobile. A third car is somewhat
less valuable and succeeding cars are of even less utility. "The law of
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
diminishing total returns" says that at some point, additional units may
actually be bad, and subtract from total utility. For example, if a family gets
too many cars, its driveway is always jammed, license fees rise, and they get
in the way of each other. The family may then NO itself to be better
off-i.e., to have a higher total utility-without these last cars, which are
then said to be of negative marginal utility.
Markov chain-(see: MARKO V ANALYSIS)
matched samples-a pair or set of samples in which each member of a
sample is matched with a corresponding member in every other sample by
reference to qualities other than those immediately under investigation. The
object of matching samples is to obtain better estimates of differences by
removing the possible effects of extraneous variables. A survey of political
attitudes, for example, might match individuals of several different samples
according to age, income, education, race, and sex, but not residence, ire
order to ascertain regional differences in political beliefs. (see: SURVEY
RESEARCH, sample, variable)
matrix-a table of numbers in which the row and column positions
have some meaning, as well as the number entered in that row and column
position. For example, a matrix might be a table whose rows are nations in
d;scending order of wealth (as judged by GNP per capita) and whose
columns are GNP, population, and amount of US aid
COUNTRY
GNP
POP
US AID
Argentina
$25,420 m.
24 m.
$22 in.
Chile
$ 7,385 m.
9 m.
$10 m.
Brazil
$34,600 m.
98 m.
$96 m.
Dominican Rep.
$ 1,395 m.
4 m.
$19 m.
Ecuador
$ 1,602 m.
6 m.
$lG m.
maximin, minimax-strategies derived from game theory in which
players try either to maximize their minimum gain (maximin) or mi:.imize
their maximum loss (minimax).
If several different courses of action each offer varying arrays of payoffs,
the maximin player will select the strategy which has the best minimum
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
guarantee-to make sure that he has a chance for at least some success. In
the matrix below, player A is selecting his attack strategy in bombing player
B's supply depots. The numbers in the cells tell how much of B's supplies A
can destroy. If A is a conservative or a pessimist, he will select strategy III,
his maximin, because the lowest value in that column (8) is still higher than
the lowest value in any other column (6, 0, 2).
Player A's Attack
I II III IV
Player B's
Defense Strategy 2
22
8
8
11
14
21
16
2
6
0
12
10
The minimax player, by comparison, will select the strategy which has the
smallest worst value-to guarantee that if things go wrong it will still not be
too bad. If B is designing defenses to protect his supply depots from A's
attacks, his strategy 3 is a minimax because its highest possible value, 12,
still hurts B less than the highest possible values from the other strategies
(22, 21). Minimax is thus the defender's version of a maximin strategy.
(see: saddle point)
McNemar test-a statistical test used to determine whether an
observed difference between proportions, or percentages, derived from two
related samples is significant or whether that difference could be a random
accident. The McNemar method might be used, for example, to test
whether a propaganda broadcast affected a gr, oup's political views. The two
samples would be taken before and after the broadcast and tested to see if
any observed differences in political views were significant. (see: confidence
level, significance level)
mean-the arithmetic average, written x and pronounced "x-bar."
The mean is equal to the sum of all observations divided by the total
number of observations. For example, the mean of the series: 2, 4, 4, 8,
9, 10, and 12 is 7. The mean is one of three common measures of central
tendency which try to use one number to describe a series of numbers.
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
nl
The geometric mean is a special case of the mean found by taking the Nth
root of the product of N numbers. For example, the geometric mean of 2, 5,
10, and 15 is the fourth root of 2 x 5 x 10 x 15 6.22. The geometric mean is
used to average rates of change, percentages, and ratios, and for highly
skewed distributions such as income distribution.
The harmonic mean is the reciprocal of the arithmetic mean of the numbers
being averaged. For exarm i)ic, the harmonic mean of 2, 5, 10, and 15 is 4.62
since 4 (1/2 + 1 /5 + 1/10 + 1115) = 4 x 30/26 = `.o2. (see: measure of
central tendency, mediae mode)
mean deviation-the average difference between the values in a set of
values and their mean or median. If 3, 4, 5, 8, and 10 are values in a set, for
example, their mean is 6 (3 + 4 + 5 + 8+ 10 = 30 divided by 5 = 6). When
the differences between each value and this mean (3, 2, I, 2, 4) are added
and divided by the number of values (5), the mean deviation of 2.4 is found.
(see: mean, median)
mean range-the arithmetic mean of the ranges of a set of samples of
the same size. The mean range in repeated sampling may be used as an
estimator for the population: standard deviation. If the ages of 20 people in
three different samples are from 20 to 30 for the first sample, 20 to 45 for
the second, and 20 to 60 for the third, the ranges of the three samples are 10,
25, and 40 respectively. The mean range for the three samples is 25, since 10
+ 25 + 40 = 75/3 = 25. (see: mean, range, population, sample, standard
deviation)
measure of association-a statistical test which shows the degree of
dependence or independence which exists between two or more variables.
Some of the more commonly used measures of association include:
Cramer's V, Goodman and Kruskal's gamma, Kendall's tau, Lambda,
Pearson's C, Phi-square, Somer's D, Tschuprow's "f, and Yule's Q.
measure of central tendency-any of several statistics which
summarizes the "middle" value of a set of data. The mean is the most
common measure of central tendency, particularly for interval or ratio level
data. If the data are ordinal, the median is used, and the mode is most often
used to best summarize nominal level data. (see: data, mean, median, mode)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
measure of loci Lion-(see: measure of central tendency)
measures of variability-any of several statistics which summarize the
spread or variation of data around its central value. The most common are
range, variance, and standard deviation.
median-the middle value of a population, denoted "M." The value
which has hall' of the group above it and hall' the group below it, For
example, in the series: 2, 4, 4, 8, 9, 10, 12 the median is 8. (see: mean, mode)
mesokurtic-a normal frequency distribution curve. (see: kurtosis)
LEPTOKURTIC
MESOKURTIC
PLATYKURTIC
0
MEAN
methodology-a body of knowledge and technique relating to the
process and assumptions of scholrtc y inquiry within a discipline. (see:
approach, technique)
minimax-(see: maximin)
modal personality-a description of an entire group as revealed by the
dominant personality traits of its adult members. Modal personality is
essentially an average, and any discussion of it should include an estimate
of the degree of variation from the mean. Complex industrial societies may
be multi-modal, and personality patterns may cluster around two or more
dominant types. Developing states are often uni-modal-that is the modal
personality is also the "national character." (see: personality trait)
mode-the value occurring most often in a population. In the series:
2, 4, 4, 8, 9, 10, 12 the mode is 4. (see: mean, median)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
model--u simplified description of reality which abstracts and
generalizes about the complexity of the real world in order to accurately and
clearly isolate the relatively few variables of particular interest to the
analyst. (see: variab!-,. SIMULATION)
moment-in statistics, the first moment is the mean of the distribution,
the second moment is the variance, the third is the measure of skewness of a
distribution, and the fourth is the kurtosis.
monotonic--a relationship which tends in the same direction when
shown in a matrix. Both diagrams A and B indicate monotonic
relationships. Diagram C does not.
City Size
Small Mod Largo
r
?Y~ Low
Mod
High
5
1 0
Low
Mod
High
City Size
Small Mod Largo
Low
rn Mod
Hlgh
6
3
5
3
4
0
6
2
6
morphogenesis-in systems analysis, a set of interrelated processes that
tend to preserve a system's values in the face of a changing environment by
evolving its structures. The international system after World War I and II
attempted to preserve the system value-peace-in a changing environment
by setting up first the League of Nations and then the UN. In contrast,
homeostatic systems tend to preserve structures. Morphogenesis is said to
be a better socizl model because a society in fact does much more than
preserve its central structures; it is continually in the process of elaborating
upon and changing those structures so that institutions will still serve the
society's values even when conditions change. (see: SYSTEMS
ANALYSIS, homeostasis, model, structure)
City Size
Small Mod Largo
2
0
0
0
4
0
0
1
6
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
multicollinearity-ii regression analysis, a situation in which the
independent variables ar; more highly correlated among themselves then
with the dependent variable. This situation makes the regression unreliable,
(see: REGRESSION, independent variable, dependent variable)
multiple correlation and regression-techniques which measure how
much of the variation in a dependent variable can be explained by two or
more independent variables acting together. (see: CORRELATION,
REGRESSION, variable)
multiple correlation coefficient-(scc: R')
multivariate analysis-a generic term for any analysis which takes into
account more than two variables. (see: -variable)
mutually exclusive events-,?~,ents which cannot occur simultaneously,
e.g., heads and tails are mutuaiiy exclusive on a single flip of a coin. (see:
independent events)
N, n-the number of items in a series or population, or the last one of
those items. Usually capitalized when referring to the entire population,
lower case when referring to a c;;.mple. (see: population, sample)
national character-a modal personality which appears to be dominant
for an entire state. (see: personality trait, modal personality, NA TIONA L
CHARACTER STUDIES)
natural experiment-research utilizing observation of human behavior
or some other phenomenon in the normal field setting without injecting
artificial stimuli. When a researcher studies the political statements made
by businessmen at lunch without introducing questions to elicit answers, for
example, he is conducting a natural experiment.
negative relationship (or inverse relationship)-a relationship between
two variables such that when one rises in value, the other falls, and vice
versa. (see: positive relationship)
nested hypotheses-a sequences of hypotheses in which the hypothesis
at any stage is contained in all hypotheses later in the sequence. (see:
hypothesis)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
noise-in communications theory, a series of random disturbances
which increase the possibility that the information received will be different
from the information sent, (see: COMMUNICATIONS THEORY)
nominal scale data-(see: data)
nomothetic explanation-explanation based on general propositions
applicable to classes of events or objects rather than specific propositions
that explain unique or individual social or poli,ical phenomena. The latter is
called an idiographic explanation. Political And other social sciences that
emphasize explanation by t-.;coretical generalization are considered
nonlogical, while history is often regarded as an essentially idiographic
discipline.
nonlinear correlation-a correlation between two variables which
cannot be graphed as a straight line. (see: CORRELATION)
X
Non-Linear Correlation
nonparametric statistics--(see: statistics)
non-zero-sum games-(see: GAME THEORY)
normal distribution-a sample in which the cases tend to cluster about
their average and progressively decrease in number toward either extreme of
measurement. When graphed, the normal distribution appears as a
symmetric bell shaped frequently distribution curve in which the mean,
median and mode are identical. (see: distribution, frequency, frequency
polygon, mean, median, mode, statistics)
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
null hypothesis-the opposite of an original hypothesis, o;ie which an
analyst formulates with the expectation that it will be discred,ed by the data
gathered. One using the scientific approach do c not -rdally try to prove
directly that a hypothesis is true but seeks to discredit its opposite with a
given degree of probability. For example, it may be hard to prove directly
that a certair strain of virus causes a disease. The theory may be supported,
however, if diet, weather, insects, other viruses, etc., can be discarded from
consideration. If the test of the null hypothesis shows it can be rejected
with 99% confidence, the original hypothesis is strengthened, but not
proven true. (see: hypothesis testing)
objective function-the criterion by which we select the best solution
from a realm of alternatives. For example, in trying to build the cheapest
battieworthy tank, the objective function would be cost, and the constraints
would be speed, firepower, strength, etc. (see: constraining equation)
observation-an element of a population or sample, a specific case.
(see: population, sample)
odds-the chance that a given event will occur, expressed as a ratio
rather than a percentage as in probability. If the probability of an event
occurring is 55%, for example, the odds for its occurrence are 55 to 45 (100
minus 55 = 45) and the odds against its occurrence are 45 to 55, or 11 to 9
for and 9 tc 11 against, respectively. (see: probability)
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
olive-a special case of a frequency polygon which graphica;ly
represents the cumulative frequency of a set of data. The ogive below, for
example, shows that 20 out of a sample of 100 had IQ scores under 80 and 80
had scores less than ! 20. (see: distribution, frequency, frequency polygon)
I I '0
40 60 120 160 200
I.Q. SCORES
OGIVE
opportunity cost-the foregone benefit which could be reached by
reassigning resources from their current use. The opportunity cost of
investing in a ballistic missile system, for example, is measured by the
amount of benefit that could have been gained by spending for some other
program. Opportunity costs involve no cash outlay, but they are real in the
sense of representing benefits or income which we deny ourselves by
selecting the present policy. Opportunity cost, therefore is useful as a way of
measuring the effectiveness or desirability of some investment or resource
allocation problem, because it asks how much could be gained by alterna-
tive programs.
ordinal scale data-(see: data)
ordinate-the left vertical axis of a graph, designated Y. (see: abscissa)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
organization-a goal-seeking collectivity of individuals having some
kind of structure designed to help achieve the collective goals. In formal
organizations, role relationships among the members are explicitly stated,
while informal organizations evolve from repeated contacts among people
without an express statement of goals or defining of roles. (see: ORGAN-
IZA TION THEORY, GROUP THEORY)
origin-the intersection of the x and y axes of a graph, where both
equal zero. (see: abscissa, ordinate)
y
4r
0 1 2 3 4
ABCISSA
output-activities carried on by a political system in response to
demands or stresses placed upon the system in the form of inputs. Outputs
take the form of governmental policies, programs, decisions, and
implementing actions. (see: COMMUNICATIONS THEORY,
SYSTEMS ANALYSIS, input)
paradigm-a model, pattern, or example that helps organize thought
and give direction to research. (see: model)
parameter-a conciiucn or value that is given or held constant or
assumed to be a limiting condition in a research problem. Parameters serve
as boundaries or constraints upon the operation of the variables. Parameter
values may be arbitrarily assigned for research purposes. In a
decision-making study, for example, perfect information may not be avail-
able in fact but is assumed as a parameter of the situation.
In statistics, ?parameter is a numerical characteristic of a population,
such as a mean, median, upper limit, or lower limit. This is contrasted with
a statistic, which is a numerical characteristic of a sample of the popula-
tion.
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Pareto optimal-a weak optimality condition which describes
situations in the distribution of some desired good from which no one
individual can be made better off without making someone else worse off.
Essentially it is an efficiency criterion, eliminating waste, but not concerned
about the equality of the distribution. 'Thus, if we are dividing $1000 among 3
people, a $700/$200/$100 split is a Pareto optimal combination, because all
the money has been used up, and no one can be made better off (i.e., get
more money) without making one or both of the others worse off. There are
obviously several different possible distributions which satisfy the
;'equirements of a Pareto optimality: $500/$5001$0; $1000/$0/$0; or
$333.33/$333.3/$333.33. Pareto optimality is referred to as a weak
optimality condition, because in some cases it may be that equality of
distribution is more important than eliminating all waste. Thus, for a true
socially optimal distribution, $300/$300/$300 may in some cases be
preferable to $600/$300/$100, although the latter is a Pareto optimality and
the former is not. But $333.33 each might be preferable to both.
partial correlation-(see: CORRELATION)
payoff matrix--a table used in game theory to list possible strategy
options for the players, and to describe the outcomes for each.
12
Player A
The player picking strategies on horizontal rows (player B) gets the return in
the lower corner of the intersection box, and the column player receives the
payoff in the upper right. Thus if player A picks his strategy I and player B
selects strategy 4, the outcome will be worth -6 to player A and + 12 to
player B.
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Pearsonian product-moment correlation (r)--u statistic which measures
the degree of association between two variables which, if plotted on a scatter
diagram, would form a straight rather than a curved line. The notation "r" is
simpiy a shorthand way of expressing the coefficient which can range from 0
to + I for positive correlation and 0 to - I for negative correlation.
The correlation coefficient can be translated into a percentage by squaring
it and multiplying the result by 100 (100 r2). If' the correlation between
the percentage of the vote for a party and the percentage of blue collar
workers is +.42, for example, this means that 18% of the. total vote for the
party is accounted for by the percentage of blue-collar workers in the
voting population (100 X .422 = 18%). (see: measure of association,
CORRELA TION)
Pearson's C-also called the contingency coefficient, Pearson's r tests
the strength of relationship (measure of association) between variables as
represented in tables of 2-by-2 size or larger. The C value ranges from .71 for
2-by-2 tables to 1.0 as the number of rows and columns increases. Some
social scientists recommend computing C only for tables of at leas" 5-by-5?
since only then does the maximw,, value of C approach 1.0. (see: n,E
of association)
perception-the act of becoming aware of things by means of the
senses. It involves two related operations-receiving impressions through
the senses and the assignment of meaning to the sensory impressions.
Studies of social perception deal with differences in people's perceptions
arising from differences in their cultural and social backgrounds, values,
and beliefs. (see: attitude, personality trail, PERCEPTION ANALYSIS,
PSYCHOLINGUISTICS)
perfect competition-a th,,;oretical model of an economy, used by
economists to characterize broad guidelines. This usually descriptive model
includes the presence of a large number of independent, well-informed sellers
and buyers, none of whom are large enough to affect the prices in a
well-organized market. The perfect competition model is without
advertising, product differentiation, or barriers to entry. No national
economy has ever closely approximated this model, but some
industries-notably agriculture and national stock exchanges-seem
nearest, and the model is a valuLyie learning tool.
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
performance evaluation. review technique (PERT)-a system aiding in
the efficient control of research and development efforts by isolating job
components, bottlenecks, uncertainties, etc., which are involved ; achieving
the desufed output subject to relevant constraints. The technique requics
three different estimates of the time needed to perform a particular task.
One is the "most likely" time, the others are "optimistic" and "pessimistic"
estimates. An expected value is derived from this probability distribution
in order to efficiently allocate resources.
permutation-any one of the sequences or changes in position possib-
with a group of data. The permutations of A, B, and C are:
ABC BAC CAB
ACB BCA OBA
The formula for finding the number of permutations of N things taken Rat
a time is:
P= (NNR)i (NI Is read "N factorial")
Thus the number of permutations of 5 things taken 3 at a time is:
5! _5= x4x3x2x1 120
P__ (5-3)! 2x1 - 2 = 60
(see: factorial)
personality trait-the persisting, organized disposition in a person that
leads him to respond in characteristic ways to his environment. Personality
traits are usually equated with the underlying dispositions that determine
behavior rather than the behavior itself, but are sometimes regarded as acts,
or characteristic modes of behaving. Personality traits are interrelated,
although psychologists differ on how traits are organized within a
personality structure. Psychologists generally hold that personality traits
precced attitudes and therefore teed to shape attitudes. The traits that shape
responses to political st',nuli are called political personality. Modal
personality refers to personality features most widely shared by a group of
people, or the most prevalent personality type. (see: attitude, behavior
pattern, model personality, NATIONAL CH.-iRACTER STUDIES)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Phi-square-a statistic which measures the strength of an association
between two dichotomous variables, such as whether sex makes any
difference in rn individual's voting Democrat or Republican. 1'hi-square is
equal to chi-square divided by the number of observations and usually varies
between 0, indicating no association and I, indicating complete association.
(see: chi square, dichotomous)
Dmnocrat
Ropubllcnn
12 (Phl?squaro) -0 (no association)
Maio Fomalo
0 (PhI.squaro)-
(comploto association)
planning, programming, and budgeting system (PPBS)-a management
system introduced to the executive branch in the early 1960's which includes
three elements: planning-the,study of objectives, alternative ways to achieve
them, and contingencies and possible responses; programming-a system of
evaluating the objectives of an activity and relating the costs or inputs needed
to the benefits or outputs produced; budgeting-the process of distributing
scarce available resources among competing claims. (see: DECISION
THEOR Y)
platykurtic-a flat frequency distribution curve. (see: kurtosis)
0
MEAN
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
pproved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Poisson distribution-a special type of probability distribution which
often applies when there are a large number of independent events and for
each of which there is only a small probability that a certain outcome will
occur. The Poisson probability distribution has been applied the analysis
of the number of accidents likely to befall a large number of workers per
day, or the number of machines out of a large number which may break
down in any one day. Its graphed plot is a single peaked curve with a steep
ascent to the top and a more gradual descent to zero. (see: normal dis-
tribution)
POISSON / / NORMAL
political action-any observable, overt behavior of an individual or
social group carried on within a political system. It may be unplanned,
random behavior or part of a coherent decision process. A response to
political actions undertaken by others is a political reaction. (see: A CTION
THEOR Y)
political anthropolieg -the study of governmental institutions and
practices among ethnic communities, particularly in primitive and tribal
societies. Such study probes the relationship of political behavior to the
broader group culture and examines the ways in which political institutions
t,nd practices evolve.
political communication-the transmission of meaning having
relevance to the functioning of a political system. Such communication
could include symbolic acts (burning a draft card), assassination, voting,
speeches, communiques, or the simple sending and receiving of spoken
messages. (see: COMMUNICATION THEOR11, CONTENT ANALY-
SIS, cybernetics)
political community-(see: integration)
66
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
political culture-the pattern of orientations toward government and
politics within a society. Political culture generally connotes the
psychological dimension of politic al behavior-beliefs, feelings, and
orientations--and is the product of the historical experience of the whole
society as well as the personal experiences that contribute to the
socialization of each individual. Within rat-Ok, gal political culture, elite and
mass subcultures are often distinguishable. (see: NATIONAL
CHA RA CTER STUDIES, socialization)
political development-growth and change within political systems, or
change from one system to another, generally toward greater governmental
capacity to cope with the demands made upon it. Commonly associated
with increasing complexity, specialization, and differentiation of political
institutions in a society, regardless of their democratic or authoritarian
character. The terms traditional, transitional, and modern are commonly
used to designate societies in different stages of political, economic and
social development. (see: development theory, political socialization)
political ecology-the entire physical, cultural and social environment
of a political system excluding other political systems. (see: ECOLOGICAL
ANAL YSIS)
political philosophy-the branch of intellectual inquiry dealing with
ideas about politics, particularly ideas relating to political values, the nature
of political reality, and the intellectual assumptions of political analysis. As
normative theory, it seeks to clarify political values and to define what is
desirable and moral. As speculation about what is rather than what ought to
be, political philosophy is concerned with the basic nature of reality rather
Uan particular observable manifestations of it. While not rejecting
observation and experience, the political philosopher looks for rational
interconnections between things as a means of describing and explaining
political reality while the empirical scholar seeks scientifically verifiable
hypothesis as the bases of explanation. As analytic philosophy, political
philosophy deals with the meaning of words and concepts, the logical
consistency of arguments, ways of knowing truth, and the grounds on which
a proposition may be taken as true or false.
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
political pluralkm-a society in which power is widely distributed
among numerous groups arrayed in shilling patterns of conflict, compe-
tition and cooperation with one another.
political scienci--the systematic study of government and politics. The
discipline has moved in the past century from an emphasis upon formal
institutions and legal relationships to a concern for processes, the behavior
of individuals and groups, and informal relationships. Methodologically,
political science has supplemented the predominantly legal, historical and
descriptive analysis of an earlier period with the methods and perspectives
of modern behavioral science. (see: BEHA VIORALISM)
political socialization-the learning process by which individuals
acquire beliefs, feelings, and values about government and politics. From the
perspective of society, political socialization is the means by which political
culture, that is the pattern of political orientations within the society, is
maintained or changed. The family and educational system are generally
regarded as the most important socializing agents, although political
orientations may also be shaped by exposure to mass media, organized
groups, informal groups, or any experience having political relevance.
Political socialization is a key determinant of political behavior. (see:
attitude, political culture)
political socicliogy-the study of political institutions and processes
in their social setting. It is concerned with the ways in which political
phenomena influence, and are influefced by, other aspects of society. The
whole society or macro approach deals with such questions as the social
foundations of power, the impact of social and economic class conflict on
politics, and the reciprocal influence of political institutions upon social class.
The micro approach to political sociology focuses on particular political
institutions as social organizations, including their formal and informal
structures, leadership patterns, methods of managing conflict and
relationships with other organizations. (see: GROUP THEORY, ELITE
ANALYSIS, ORGANIZATION THEORY)
political system-the pattern of human relationships through which
authoritative decisions are made and carried out for a society. A political
system is distinguished from other social systems by four characteristics: it
is universal in its reach, extending to all members of a society; it claims
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
ultimate control over the use of physical coercion, it has the right to make
binding decisions; and its decisions are authoritative, bearing the force of
legitimacy and a substantial probability of compliance, (see: legitimacy,
system, SYSTEMS ANALYSIS, SYSTEMS D YNA MICS)
political theory-a body of thought that seeks to evaluate, explain, and
predict political phenomena. Political theory is also a subfield of political
science concerned with political ideas, values, concepts and prediction of
political behavior. In this broad sense it has two major branches-political
philosophy with its value, analytic, historical, and moral concerns, and
empirical theory devoted to an effort to explain, predict, guide research, and
organize knowledge through the formulation of abstract models and
scientifically testable propositions.
politics- -human activity concerned with making and implementing
decisions vested with the authority of the society for which the decisions are
made. Politics connotes activity or process, whereas political system implies
the existence of structures or patterned relationships.
polity-the political organization of a society. Polity may refer to the
citizenry of a particular country or it may refer to the institutional forms
and processes through which the country is governed. The concept may
apply to highly organized states as well as very primitive societies i.n which
recognized political authority has barely emerged. (see: political system)
population-any group of individuals or things which have at least one
characteristic in common. (see: sample, parameter)
positive relationship-a relationship between two variables such that
the values of each rise and fall together. (see: negative relationship)
post hoc ergo propter hoc-a logical fallacy which may occur when
attempting to establish causality. It results from mistakenly assuming that
merely because something occurred after some other event, it must have
occurred because of it. Temporal sequence can disprove causality, but can
not prove it.
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
post behaviorallsm-an intellectual movement in political science
dating from the late 1960's, that asserts the obligation of political scientists
to become more "relevant" and concerned with values and to use their special
knowledge of politics and political systems to improve society. It represents
an attitude toward the discip ne and the profession of political science held
by some political scientists rather than a well d'weloped doctrh . or school
of thought.
power--the capacity to affect the behavior of others in some desired
way deriving from the threat of punishment, force, or coercion, as distinct
from authority in which comp!i-tnce is voluntary.
power function-in the Theory of hypothesis testing, the probability that
the null hypothesis will be rejected when in fact the alternate hypothesis is
true. (see: hypothesis testing, null hypothesis)
premise-an idea, belief, or assumption one holds to be true. It is the
starting point of a syllogism used in deductive reasoning. (see: deductive
reasoning, syllogism)
oresent value-the current worth of future income. The formula for
present value is given by the amount of money to be received divided by
the sum of one plus the interest rate. Thus, if the interest rate is five per-
;ent, the present value of an expected $100 to be gained a year from now
s only $100 :-(1+.05) or about $95.24. This is the amount of money you
fould have to invest now at 5 percent in order to get back $100 one year
'om now.
prisoner's dilemma-a class of variable sum matrix problems dealt with
in game theory. This describes the situation in which the best alternative for
both players is to cooperate, but due to a lack of communication or trust,
they each have incentives to defect, or break their coalition, bringing results
neither side wants. The game takes its name from a situation in which the
police have apprehended two criminals suspected of a major crime. The
police already have enough evidence to convict both for a related, less
important crime, but unless one confesses, the major charge will not stick.
Prisoner A
Keep
silent Confess
Keep
silent
C
0
0
it Confess
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
The police separate the two criminals, explain the situation, and offer each a
reduced sentence in exchange for a confession implicating the accomplice in
the major crime. The matrix below displays the choices. If each keeps silent,
both get a light sentence for the minor charge (worth -5 to each). If one
confesses and the other d^";s not, the confessor goes free (worth 0) and the
silent one gets the heavy sentence (worth -15). If both confess, both get a
reduced sentence on the major crime (worth -10). In this situation, the best
strategy for both is silence. But with no communication between them, and
little trust, each may perceive it to be in his own interest to confess.
Moreover, each knows that the other is facing a similarly tempting
situation, and the game has its saddle point in the lower right, with mutual
ruin. This type of game is often used by analogy to describe situations like
superpower diplomacy, retail sales competition, etc. (see: variable sum
game, saddle point)
probability-the chance that a given event will occur, expressed as a
fraction from zero (signifying an impossible event) to one (signifying an
event that will definitely occur), or the corresponding percentage from 0 to
100.
probability diagram-a graphical method for breaking a problem into
its component parts, assessing the probability of each part and thereby
finding the overall likelihood of an occurrence. If the analyst were assessing
the probability of a coup in a foreign country, for example, and he felt the
possibility of a foreign threat to be the major determining factor as to whether
a coup would occur or not, he might construct the following probability
diagram to assist his analysis. (see: probability, DECISION THEORY)
PC COUP
Foreign threat
N=CO P
d
No Pe COUP
foreign threat
P NO COUP
f
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
probability distribution-a prediction of the relative frequencies with
which possible values of a variable will occur. For example, in predicting the
number of Redskin football victories this season, we could pick just one
number, as our prediction, but we might do better by defining a probability
distribution such as: .1 chance of winning 10 or fewer games; .2 chance of
winning I I or 12 games; .3 chance of winning 13 or 14 games; .3 for 15 or 16
games; and .1 of winning more than 16 games-totali g 1.0 or 100%
probability.
process-a sequence of related actions or operations. As a general
concept, political process embraces all the activities by which people
attempt to gain and wield legitimate influence within a society. (see:
political system, politics)
program-a set of instructions or steps that tell a computer how to
handle a complete problem.
projective methods-a research technique used to measure
psychological variables by inducing human subjects to project their inner
attitudes, feelings, values, motives, and needs upon external objects. The
most common projective methods are association, constructions,
completion, and expression. In the use of association techniques, the subject
is asked to respond to words or pictures with the first idea that occurs to
him. The response will presumably be a projection of his attitudes and
feelings. Construction requires the subject to produce a story or picture
which is then interpreted. With the completion technique, the subject is
asked to finish a sentence, story, or incomplete situation. Expressive
techniques permit the attitudes and feelings of the subject to be deduced
from his behavior in some observed activity-talking, playing games, and so
on.
property space-one of two basic graphic forms of data representation.
The tabular property space cross tabulates two variables, each divided inte
any number of subvariables, in a table. A scatter diagram represents the
two variables as the two axes of a graph, and pinpoints each observation on
the graph.
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
? I ? I I I I I
CONSERVATIVESNESS
psychological scaling-the quantification and measurement of either
certain personality traits or entire personality types. The analyst ranks
individuals in particular categories based on his knowledge of their
behavior. The categories are given values which can be statistically
manipulated. "Miller's Scale Battery of International Patterns and Norms,"
for example, contains 15 rating scales to help ascertain important attitudes
and beliefs within national cultures.
A number of scales and ini 1e.xes have been designed to quantify and analyze
individual behavior. A VIP 6chavior and performance study, for example,
identifies five measurable factors-task orientation and functioning,
emotional adjustment, interpersonal style, leadership style, and
influencibility. (see: SCALING, personality trait, behavior pattern)
psychology-the science of human and animal behavior; the study of an
organism in all its variety and complexity as it responds to the flux and flow
of the physical and social events which make up the environment. These
responses tend to coalesce into behavior patterns, which can be either overt
or covert. Psychologists employ many methods for discovering and
analyzing behaviov patterns beyond simple observation. For example, when
an individual is angry, subtle changes occ_~ r in his digestive tract, his
vascular system, and breathing, all of which can be discovered through the
use of special instruments. (see: personality trait, behavior pattern)
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
psychoanalysis-a branch of psychology directed toward the
understanding, cure, and prevention of mental disorders. Generally the
roots of human behavior are sought in unconscious motivation and conflict,
public opinion-beliefs and attitudes of people generally on a socia; or
political issue. Public opinion may relate to matters of fact or of preference.
(see: SURVEY RESEARCH, attitude, behavior pattern)
quartile or Q value-the three values in a set of dat.. below which lie the
bottom 25% of the data (Q1), below which lie the bottom 50% of the data
(Q2), and below which lie the bottom 75% of,the data (Q3). Quartiles are a
class of arbitrary divisions called fractiles which can be changed to suit
particular analysis needs. Deciles, for example, divide the data into
increments of 10%. (see: fractile)
NUMBER OF
RIOTS
CUMULATIVE
PERCENT
1
0
.0
2
3
2.5
3
2 (Q I =riot 30)
12.5
4
(Q 2 - riot 60)
30.0
5
60.0
6
24 (Q3 = riot 90)
80.0
7
12
90.0
8
9
97.5
9
3
100.0
10
0
100.0
R-squared (R2)-a measure of the "goodness-of-fit" in regression
analysis. If R2=0.78, then 78% of the variation of the dependent variable is
explained by or determined by the independent variables. R2 is also called
the coefficient of determination. The square root, r, is known as the multiple
or Pearsonian product-moment correlation coefficient. (see:
REGRESSION)
random-governed by chance. A random variable is one whose values
are assigned by chance; a random sample is one whose elements are selected
by chance. This means that each number or element in the population must
have an identical probability of being selected.
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
range-the difference between the largest and smallest values in a set of
data. (see: variable, domain)
rank order correlation-a technique of correlation which involves the
ranking of subjects (for example, countries) in descending order of
magnitude on each of two variables (for example, gross national product
and size of armed forces) and determining the extent to which rank order on
one variable is similar to the ranking on the other. This form of corgi elation
takes only the difference in rank into account. (see: CORRELATION,
product moment correlation)
rational actor model-traditionally the most widely used model in
political analysis, it sees the nation as a unitary and rational actor. A single
entity, such as the Soviet Union, Peking, or the Third World, is assumed to
be capable, like an individual, of having ideals, perceiving interests, reacting
to events, laying plans, having purposes, making decisions, taking actions,
and pursuing policies. In his book, Essence of Decision, Graham Allison
contrasts the rational actor model (which he calls Model I) with the
organization (Model II) and bureaucratic politics (Model III) models. The
organization model assumes that the actions of a government are not the
actions of a national entity but those of the semi-autonomous organizations
making up the government. The bureaucratic politics model conceives of
government as an arena of competition among individuals within that
government. (see: BUREAUCRATIC POLITICS, ORGANIZATION
THEORY)
ratio scale data-(see: data-ratio)
raw score-the actual original numerical value of an item or individual
according to one test or measurement. This value may be transformed into a
standardized z-score to allow the comparison of values from entirely
different sorts of tests. (see: z score)
real number-a value which can be expressed either as a whole number
or a fraction-age or height, for example-as opposed to an integer, such as
number of persons or weapons, which must be a whole number. (see:
integer, variable)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
recruitment-the selection of people to rill roles in a social or political
system. Recruitment refers both to the filling of formal, legal positions, such
as president, legislator, or civil servant, as well as less formal roles, such as
lobbyist, party activist, or media representative. Specific techniques of
recruitment may include co-optation, appointment, election, rotation,
apprenticeship, examination, and purchase. Since recruitment is a function
performed within every political system, structural-functional analysis
treats it as a major category of inputs. (see: STRUCTURAL-FUNCTIONAL
ANAL YSIS, POLITICAL SYSTEM)
reductionism-a method of explanation in which the attributes or
behaviors of the whole are reduced to, or explained by reference to, the
attributes or behaviors of its constituent parts. A reductive approach, for
example, might explain the functioning of a political system in terms of
personality variables, or by reference to particular components of the
system, such as actor characteristics, processes of political socialization and
recruitment, political values, z.nd so on. Reductive explanation may be
contrasted with holism, which focuses on the functioning of the whole
system rather than its parts. (see: holism)
reference group-any group of people from which a person adopts
attitudes, beliefs, or values that affect his behavior. His behavior is therefore
determined by reference to theirs. A comparative reference group provides a
standard for comparative self-evaluation-for example, a group of
co-workers, classmates, or persons of one's own social background. A
normative reference group helps to shape the values, attitudes, and norms
by which the individual evaluates the world around him. (see: attitude,
behavior pattern, personality trait)
regressor-any independent variable in a regression relationship with
another variable. The variable "unemployment," for example, would be a
regressor in a regression study of the causes of riots. (see: REGRESSION,
independent variable)
regret-in decision theory, the difference between the total loss as a
result of a decision and the unavoidable loss regardless of the decision. It is
this excess loss that has to be minimized. (see: DECISION THEORY)
rejection region-(see: critical region)
76
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
relative frequency (or proportional frequency)-a frequency which
is exprt ssed as a proportion of the total, number of occurrences or the
total number of members. (sec: frequency)
FREQUENCY
! RELATIVE
I FREQUENCY
0- 40
5
1/20
41- 80
20
1/5
81- 120
50
1/2
121- 160
20
1/5
161- 200
5
1/20
J
relaxed oscillation-in a time series model, when the value of a
phenomenon, although oscillating, generally increases during a given period
of time through the operation of its internal forces. These forces precipitate
a "crisis" or "bursting" followed by a return to zero after which the process
begins again. Models of some economic phenomena, such as inflation and
depression, follow this pattern. (see: model, TIME SERIES ANALYSIS)
reliability--the extent to which a measurement technique produces the
same results time after time, assuming the relevant conditions to be held
constant. It implies constancy, dependabilit , and accuracy of
measurement, but it does not imply that the measurement technique will
accuratel,:' measure the variable the researcher is concerned with, in the way
he wants it measured. That is a question of validity. (see: validity)
reproducibility-an experiment or survey is said to be reproducible if,
on repetition of the experiment or survey under similar conditions, it gives
the same results.
research design-the overall plan for a research project, specifying
what the researcher proposes to do and how he proposes to do it.
residual-the difference between the actual value of an observation and
the estimated value found from the regression formula. If the observed
values were plotted on a graph with the best fitting regression line drawn
through them, for example, the residual would be the sum of the differences
between the points and the regression line.
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
pproved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
y REGRESSION LINE
A
In regression analysis, the residual represents the portion of the observed
variation in a dependent variable which is not explained by the independent
variables. For example, if we are attempting to predict government
instability, we may find that a knowledge of demographic factors such as
unemployment, urbanization, size of the armed forces, and literacy allows
us to correctly predict instability 80% of the time. The residual in this case,
20%, tells how often we will be wrong because of the insufficient explanatory
power of the data. If the residual is large, we have not specified all the
important independent variables which influence the variable under
examination. (see: REGRESSION)
return period-in TIME SERIES ANALYSIS, the interval of time
taken by the series to rvIurn to some assigned value, usually an extreme
value. For example, the "return period of flooding" in a river is the time it
takes the river to recede to normal levels between floods. (see: TIME
SERIES ANAL YSIS)
Rice-Beyle cluster bloc-(see: CLUSTER ANALYSIS)
risk-generally synonymous with uncertainty. Sometimes risk refers to
situations where the probabilities are known, and uncertainty to situations
where they are not.
role-the set of behaviors expected of one who occupies a given
position in society. Political roles are the expected behaviors associated with
positions as legislator, party leader, voter, or revolutionary, which relate to
making and implementing authoritative decision.:, for society. The
occupant's actual behavior is called role enactment, which is affected by his
perceptions of the role expectations of others, his own interpretation of his
role, his sensitivity to demands for specific enactments generated by the
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
situation, and his skills and capacity to respond. A group of roles occupied
by a single individual is a role set. Roles within a set that make
contradictory demands upon the individual produce role conflict. (see:
behavior pattern)
routine-a computer program or set of instructions arranged in proper
sequence to cause a computer to perform a desired task.
run-a single cony;-vous performance of a computer routine.
sample-a small part r,,' a population, selected to represent the entire
population. (see: population, statistic)
saddle point-a solution to a game theory matrix which represents both
a maximin to one player and a minimax to the other. Any unilateral change
from the saddle point results in a reduced payoff for the player making the
change, thus the saddle point represents a stable situation. Not all matrices
have saddle points, and in those which do the point is not always reached
during the course of the game.
An example of a saddle point is in the matrix below. Player A is a maximin
player, attacking the storage areas of player B, the minimax player. The
number in each cell tells how much material is destroyed-A wants a high
figure, B a 'tow one. Player A selects strategy III because the lowest value in
the column (70) is higher than the lowest value in either column I (30) or
column If (50). Player B selects strategy 3 because the highest value in row 3
(70) is still lower than the highest value in either row 1 (90) or row 2 (80). The
solution, the intersection of row 3 and column III is a saddle point,
representing a stable equilibrium. It is stable because if either player
suddenly tries to disrupt it by switching to another strategy, the outcome is
worse for him. (see: maximin, minimax)
Player A (MAXIMIN)
ATTACK STRATEGIES
t 11 ttt
Player B
(MINIMAX)
DEFENSE
STRATEGIES
30
90
80
80
50
80
50
60
70
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
sample space-a set of points plotted on a graph corresponding to all
possible values of all possible samples.
r ON (2 3) (3,3)
(12) (2,2) (3,2)
(1*1) (271) (3 1)
1
i I _x
0
(0,0) 1 2 3
If in the two-dimensional sample space the X axis represents three issues in
the UN and the Y axis represents votes (yes, no, and abstain), the sample
space would show all the possibilities for those votes. If a third variable
were added--level of development of the voting country (underdeveloped,
developing, and modern)-as axis Z, a three-dimensional sample space
v?""uld be needed. The point 2, 3, 2 whould then represent an abstention
by a developing country on the second issue. The point 1, 1, 1 would rep-
a resent a yes vote by an underdeveloped country on issue 1.
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
scale-a means of classifying, ranking, or otherwise measuring objects
according to their possession of a given characteristic. A scale may be
constructed to measure one of four levels of data: nominal, ordinal, interval,
or ratio. A scale of political instability, for example, might rank instability
in a country each month from 0 to 5, ranging from no overt acts of violence
(0), through assassination (3), to civil war (5). (see: data, SCALING,
GUTTMAN SCALE)
scatter diagram-a graph plotting observed points, used to visually
inspect the relationship between the variables represented on the two axis.
For example, a scatter plot with height on the vertical axis and age on the
horizontal axis may show 19 observations like this:
6.0
5.5
5.0
4.5
= 4.0
C7 3.5
= 3.0
2.5
2.0
LJ ' I ' JJ
2 4 6 8 10 12 14 16 18
set-any defined collection, category, group, or class of things. A set
must be defined by listing all of the objects in it, or more commonly, by a
rule stating the crit?ria for membership. A part or sample of a set is known
as a subset. The universal set consists of all elements of whatever is being
examined, and a set without any members in it is an empty set or null set.
Set theory is a branch of mathematics devoted to the manipulation of
numerical sets.
short-circuiting-(see: distortion)
short ran-in economics, a period of time in which a firm may vary its
utilization of some but not all resources. In the short run, output may be
varied by hiring and firing, by purchasing more or f:mer inputs, etc., but not
by major changes in fixed plant facilities. Changes in physical plant take
longer in most cases, and these adjustments are made only in the long run.
The dividing line between the long and short run is flexible and will vary
from industry to industry. (see: long run)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
sigma or sum--the Greek letter sigma is the symbol used to indicate the
operation of successive additions. The numbers i' )ve and below the
indicate the upper and lower limits of the operation. For example,
.1
E (n+ 1) -9 because (1+1)+(2?I.1)+(3+1)-=9
t
0
Z (n1 )=~86 because 3 2 ?1.41 +52 ?1-61 t-86
3
significance level-(also called confidence level) the statistical
probability, derived from a significance test, that the observed data would
occur solely by chance. (scX significance test)
significance test-a statistical test designed to determine the
probability that the distribution of observed data could occur entirely by
chance. These tests generally specify some low probability level, such as I or
5 percent, as the threshold for consideration of the data. This means that
unless we are 99 or 95 percent certain that the observed distribution could
not occur by chance, the data is considered insignificai,t. Some of the more
common t%sis of significance include: chi square, F tes:, Fisher exact t ;st,
Kolmogorov-Smirnov D test, McNemar test, t test, and z test. (see:
confidence level)
skewed distribution -a lopsided frequency distribution, in which the
mean, median, and mode do not coincide as they would in a normal
distribution. A statistic measuring skewness will equal zero when there is
left-to-right symmetry; a positive value indicates a distribution skewed to
the right; and a negative value indicates a distribution skewed to the left. (A
distribution is said to be skewed to the direction in which the excess tail lies.)
(s:ee: kurtosis)
SKEW RIGHT
(Positive)
NORMAL CURVE
(Zero skew)
SKEW LEFT
(Negative)
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
slope-a measure of the steepness of a line. It is computed by
measuring the number of units the line rises (or falls) per unit of movement
to the right. That is, if moving 12 units to the right involves a 6 unit rise in
the graph of the line, then the slope is 6/ 12 or 1/2. A negative slope
indicates that the line falls from the upper left of the graph to the lower
right, and the absolute value of the slope then tells how sharp the fall is.
smoothing-any procedure that reduces the fluctuations or random
scatter in data, or imposes a pattern upon seemingly random data.
Regression is a type of smoothing. (see: random, REGRESSION)
socialization-(see: political socialization)
social matri;:-the web of social relationships within which political
behavior occurs. Political roles are examined in relation to other social roles
and to the group and class structure of the society. An analyst studying the
social matrix of an African government, for example, would begin by
analyzing the social aspects of the society-tribal organization, class
structure, clan system, traditions and mores-and would study the political
leaders in this context. (see: role, society)
social mobility-(see: social stratification)
social science-the applications of scientific methods to the study of
human I haviur, the body of knowledge produced, or the disciplines
engaged in such inquiry. Behavioral science is a newer term for essentially
the same approach. Anthropology, sociology, psychology, economics, and
political science, are generally regarded as the social sciences. History,
geography, law, and public administration, are frequently added to the list.
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
social stratification-the division of society into classes or strata,
hierarchically rankcJ according to perceived differences in wealth, prestige,
or other social characteristics, Movement of individuals from one level to
another or the shift in relative position or a whole stratum or class is known
as social ,nobility. Political analysts are inte., ~tcd in stratification as part of
the broad area of political behavior, particularly in the study of political
power, influence, and authority. (see: political sociology)
social system-an aggregation of two or more persons who interact
with one another in some patterned wily. A social system, consisting of
interacting people, may be distinguishf.d from a cultural system, which
refers to learned modes of behavior characteristic of a society, and from a
personality system, which refers to the organized dispositions within a single
individual that lead to characteristic responses.
society-an aggregation of people who have certain common attributes
that distinguish them as a group and who interact with one another in some
characteristic way. In political science, society is treated as the larger
system, usually a nation, within which political actors, or interactions, may
constitute a subsystem.
sociogram-the graphic representation of relationships within small
groups by means of circles and connecting lines, or son; other geometric
figures. (see: SOCIOMETRY)
Sociometrically-mapped diagram
Moderates Radicals
2 7
4 5
Key
1. President
Ministers of
2. Foreign Affairs
3. Security
4. Defense
5. Finance
6. Education
7. Development
8. State
9. Culture
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Sonier's d-a statistic which measures the degree of association
betweer, two variables. The statistic varies between 0 to indicate 11o
relationship and + 1.0 to indicate a monotonic relationship (one in which the
relationship tends in the same direction when graphed on a matrix). (see:
r,easure of association)
spillover-an integrative process by which cooperation between
political units in one issue area leads to further cooperation in other areas.
Spillover typically arises from the attempt by states to resolve problems
created by their previous cooperative acts. The expansion of the European
Coal and Steel Community into a wider European Economic Community is
a classic case of spillover. (see: INTEGRATION THEORY,
FUNCTIONALISM)
spurious-a false causal relationship. Correlation analysis, for
example, may indicate a high association between variables A and B, but
causal inferences would be unfounded if in reality variable C causes the
observed movement in both. (see: inference)
stability-a condition of a system whose components tend to remain in,
or return to, some constant relationship with one another. Stability is
identified with the absence of basic or disruptive change in a political
system, or the confining of change to acceptable or specified limits. (see:
system, SYSTEMS ANALYSIS)
standard deviation-the most common measure of the dispersion
around the central value of a group of numbers. Standard deviation,
denoted "s," is equal to the square root of the variance, which itself is
equal to the sum of all the differences between the mean value of a set of
values, divided by the number of'values. For example, the two sets of' data-
5, 6, 7, 8, 9 and 2, 3, 7, 8, 15-have the same mean and median-7-brit
very different standard deviations-l.4 and 4.6, respectively-reflecting that
the second set is much more widely dispersed about its mean than is the
first. The standard deviation of the second set is determined by the process:
VARIANCE= (2-7)2+(3-7)2+(7-7)2+(8-7)2(15-7)2
106/5 = 21.2
STANDARD DEVIATION =V21.2=4.6
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
In a normal distribution, about 68% of the observations will occur within
one standard deviation abo,.: -. or below the mean. About 95% of the cases
will be within 2 standard deviations of the mean, and about 99% will fall
within 3 standard deviations above or below the mean value of the
distribution. (see: dispersion, normal distribution)
-3 -2 -1 +1
MEAN
standard error of estimate-the standard deviation of the distribution
of points around a regression line. This tells how much of the total
variability observed in the data remains unexplained by the regression
analysis. (see: standard deviation, REGRESSION)
statistics-the science of dealing with the collection, classification, and
presentation of numerical data. As applied to political research and
analysis, statistical methods are used to summarize data and reduce it to
manageable form, to locate and evaluate patterns in the data, and to assist
the analyst in making reliable inferences from the data. Statistical
procedures are parametric if they are designed to deal with samples of data
having a normal distribution-that is, the cases making up the sample tend
to cluster about their average and progressively decrease in number toward
either extreme of measurement. Nonparametric statistical
procedures-including Kendall's tau and Goodman and Kruskal's
gamma-do not require normal distribution.
A statistic is also the name given to a numeric. 1 characteristic-such as the
mean, median, or mode-of a sample as contrasted with a parameter, which
is a characteristic of the larger population from which the sample is drawn.
(see: parameter)
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
pproved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
stochastic-a variable or process involving chance. A variable is
stochastic if the values it assumes are governed by chance. The results from
repeatedly flipping a coin, for example, would be a stochastic variable, as
opposed to a variable such as a pricetag, which is calculated and assigned.
(see: random)
Stouffer's H-a technique to give greater stability to Guttman scales by
basing each scale item on three or more subitems, thus reducing the effect
of random errors. Item 3 on a riot severity scale might be composed of
subitems (a) vandalism, (b) looting, (c) arson. If at least two out of three
of the subitems occurred, item 3 would be listed as occurring. (see:
G UTTMA N SCA LING)
strategy-a plan of action to defeat an opponent or to achieve some
other goal. Associated with game theory where it is used to mean the plans
of the players. (see: GAME THEORY)
structure-a pattern of related roles or established relationships among
people. A political party organization is a formalized pattern of related
roles, and is therefore a structure. The established relationship between the
party organization and the voting public, although less formalized, could
also be regarded as a structure. In STRUCTURAL-FUNCTIONAL
ANALYSIS, structures are the patterns which govern actions, while func-
tions are the results or consequences of actions. (see: STRUCTURAL-
FUNCTIONA L ANAL YSIS, function)
student's t-(see: t test)
suppily-a measure of the willingness and ability of an individual or a
society to produce and sell a good at a given price. (see: demand)
survey data-data which provides information on individual cases
under observation but does not summarize the characteristics of a number
of cases as does aggregate data. Import and export figures by transaction,
or specific counts of people per household are examples of survey data. (see:
SUR VEY RESEARCH)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
pproved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
syllogism-an urguirwot consisting of a major premise, a minor
premise, and a conclusion. A syllogism is the major tool of deductive
reasoning, which moves from general classes to specific cases. If the
premises of a syllogism are true and the reasoning is logical, the conclusions
must also be true.
Major premise: All Washington Redskins are overpaid.
Minor premise: Larry Brown is a Washington Redskin.
Conclusion: Larry Brown is overpaid.
(see: deduction)
system-any set of elements that exist in some patterned relationship
with one another. The elements of a system may be specified by the analyst
to meet his needs, for example, a group of people might be specified as the
elements making up a political or social system, a the elements might be
specified as political interactions rather than particular individuals.
The way in which the analyst defines the elements of the system and the
relevant relationships among them will determine the boundaries of the
system-the analytical line separating the system from its environment. An
open system is one whose functioning is affected by inputs from the
environment. A closed system is unaffected by its environment. (see:
SYSTEMS Al!'AL YSIS)
t test (or student's t)- a statistical test, designed to tell whether two
samples are significantly different. The test compares the means of the two
distributions to tell how often such a difference could be accounted for
simply by chance. The t test is a special case of the F test, with wide uses in
statistics to tell how much confidence we can place in the hypothesis that the
two samples are significantly different. (see: confidence level, significance
test)
tautology-a statement that is logically true by definition but can
neither be confirmed or contradicted by facts. For example the statement,
"The Soviet Union is or is not a dictatorship," is a tautology since it includes
all logical possibilities and cannot be controverted. It also imparts no
information about the subject. Propositions of mathematics or i,Jgic are also
tautologies because they are necessarily true within their own logical
systems.
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
taxonomy-a systematic classification scheme I'or the orderly arrange-
ment of subject matter into categories or classes according to perceived
similarities and differences, The classification of all information on a
country into political, econosnic, military, social, biographic, geographic,
and historic is a simple taxonomy.
technique-any systematic method of analyzing data in an attempt to
answer theoretical questions about it. Technique is contrasted with an
approach in that the latter is a strategy of analysis that provides a unique
viewpoint and includes a set of tools or techniques for examining data.
CONTENT A NA '. YSIS, SIMULATION, and REGRESSION ire
examples of techi, ies, while the use of social psychology in politioal
analysis is an approach. (see: methodology)
tendency statement--the positing of a relationship b;;iween variables
as a probability rather than a certainty. It usually suggests the direction
but not the magnitude of the relationship and is subject to verification or
disconfirmation, for example, the statement: Political participation tends to
increase with the citizen's level of education." (see: hypothesis, probability,
variable)
terminal-a point in a system or communication network at which data
can either enter or leave.
Tetrachoric correlation-a form of correlation used when both
variables have been condensed into dichotomies (high-low, large-sma.li,
etc.). The computation of the Tetrachoric coefficient is complicated, so
tables have been developed as an aid. (see: CORRELATION, coefficient,
variable)
theory-i'n idea or body of thought that purports to explain, predict, or
prescribe in any field of inquiry. A single theoretical proposition or
generalization is sometimes referred to as a theory, but the term usually
implies the linking together of a number of related propositions in some
coherent structure, such as a theory of power elites or voting behavior in
international organizations.
T ourstone scale-a s,.-ale which ranks individuals or groups of
individuals at approximately equal intervals according to some
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
criterion-UN members, for example, ranked according to the proportion
of time they voted with the US. The equal intervals between the rankings
permit the use of statistical techniques involving arithmetic manipulation of
the data. (see: SCALING)
time-sharing-the alternate use of a device, such as a computer, by two
or more users.
traditional approach-an approach to political research and analysis
that emphasizes humanist, legal, and philosophical perspectiy;~s in contrast
to the predominantly scientific and empirical approach called
bthavioralism.
transactions-a type of events data consisting of the routine exchanges
across national boundaries such as mail flows or tourists. Interactions are
the public, political flows of events such as official visits or government
communiques. (see:PVENTS ANALYSIS, COMMUNICATIONS
THEOR Y)
transformation-any mathematical operation performed upon a set of
values to convert or transform these values into numbers more easily
manipulated in statistical tests. The most common types of transformation
include squaring the value, taking its square root, dividing the value into 1,
and finding its log. If a set of data involves some positive and some negative
values, for example, the normal procedure for finding their average
(dividing the sum of the values by the number) would be inadequate since
the negative values would be subtracted from the positive ones. In such a
case, the values are first squared (each multiplied by itself) so that tine
resulting values are all positive, and then averaged.
triserial correlation--an extension of biserial correlation (correlation of
a dichotomous variable such as "male-female" with a measurable variable)
to be used when .a trichotomous variable (such as low, medium, and high
levels of riot seve' Sty) is correlated with a continuous interval-level variable
(such as unemployment rates). (see: CORRELATION)
trL a values-mathematical values such as coefficients or predictions,
based upon all possible observations (the whole population) rather than on a
sample set of observations. (see: population, sample)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
pproved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
truth table-used in symbolic logic, a list of the truth values of certain
logical statements. For example, if A represents any statement and "not A"
represents its opposite, then one or the other, but not both, may be true. The
table below shows another possibility: if statement A is true and statement B
is false, then anything that is included in both A and B (read "A intersection
B") is false, and anything in A or B but rot both (read "A union B")
is true.
Not A A and B A or B
A (A) B (AnB) (AUB)
T F F F T
Truth tables are used in symbolic logic for analysis and also have
applications in statistical inference and computer work. Truth in symbolic
logic is not necessarily truth in fact, however, if the hypothesis, in fact, is not
true. (see: Boolean algebra)
Tschuprow's T-a measure of association which is used to show the
deorree of dependence or independence between two or more variables when
these are arranged in table form. The T value varies from 0, indicating
no association, to 1 indicating complete association. (see: measure of
association, chi-;square)
Type I error'-(see: alpha error)
Type II error-(see: beta error)
uncertainty-generally synonomous with risk. Sometimes risk refers to
situations in which the probabiiiiies of relevant events are known, and
uncertainty to situations in which they are nt;t.
universe-in survey research, a population- -chat is, all members of any
class of things, including people, that might be studied. A sample, by
contrast, is a smaller group of things taken from the larger universe or
population for more detailed scrutiny. (see: SURVEY RESEARCH)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
pproved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
up cross-the point where a time series, measured about its mean,
changes in sign from negative to positive. Conversely, the point where it
changes from positive to negative is called a down cross. (see: TIME
SERIES ANALYSIS)
util-in the study of policy options, an artificial measure of utility.
utility-psychic satisfaction derived from consumption or gain. Utility
is a subjective, personal value. It is not the same as "usefulness" in the
functional sense-a painting, for example, may not be useful, but it does
clearly afford satisfaction. The concept of utility in economics is key to the
theory of consumer demand. In political analysis, utility is important to the
study of policy options and decision-making. (see: DECISION-MAKING
ANAL YSIS)
validatian-in hypothesis testing, a means of establishing the accuracy
of the measuring instrument, or the validity of the test, in measuring the
relevant variables. An instrument is considered valid if it measures the
characteristics it is designed to measure. (see: hypothesis testing, variable,
validity)
validity-the accuracy with which a measurement technique measures
the variable an analyst is investigating in the way he wants it r,,easured.
Validity contrasts with reliability which is the extent to which a
measurement technique produces the same results time after time. A public
opinion poll that consistently finds 75% of its respondents favoring
monarchy in the US might. be very reliable, in that only monarchists were
samp::d, but would not be a valid measure of the actual state of opinion in
the country.
value-a concept of what is desirable or good or, in measurement
terminology, a difference in amount or kind of a variable. The variable
I.Q., for example, may have values ranging from 0 to 200.
variable-any characteristic or an object or class of objects that may
vary in quantity or quality for different members of the class or for the same
individual at different points in time. In political science, variables do not
usually refer to changes in the behavior of the ~ame individual but to
differences between individuals in the same class. Legislators, for example,
may be compared with one another on the basis of such variables as party
affiliation, age, geographic region, or education.
Approved For Release 1999/09/25: CIA-RDP85T00875R001100020001-3
pproved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
The event or condition the analyst wishes to explain or predict is called the
dependent variable. The factors or conditions which may have explanatory
or predictive power are the independent variables. The relationship between
dependent and independent variables is not necessarily causal, although
causality is often inferred.
variable sum game-a class of problems dealt with by game theory
which includes elements of both competition and cooperation. The sums of
the payoffs to the two players in the different matrix cells are not equal,
meaning that some combinations of payoffs are better for each player than
are other payoff combinations. Unlike a fixed sum game, in which o,-,e
player always loses as much as the other gains, a variable sum game also
includes regions of cooperation. Some, but not all, interests an, held in
common by the players in a variable sum game. For example, in the game
below, both players seek to avoid the outcome resulting from the
combination of strategies Il and 3 (both lose heavily in this situation).
Beyond that, however, there is competition-A's best outcome (the
intersection of strategies 11 and 1) is not good for B, and B's best outcome
(strategies II and 2) is not good for A. (see: fixed-sum game, zero-sum
game)
Player A's Strategies
I II III
Player B's
Strategies 2
37
40
21
?2
6
25
-1
-8
12
100
21
35
-25
16
2
-40
23
variance-a measure of the dispersion of a set of numbers around their
central value. Denoted s2, variance is the sum of the squares of the
differences between each value and the average of all the values, all divided
by the total number of values. In the sample, 5, 8, 11, 16, 20, for example, the
mean is 12 and the variance is (12-5)2+(12-8)2+(12_ 11)2+
(12-16)2+(12-20)2=146/5=29.2=s2. (see: standard deviation)
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
pproved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
vector-a quantity which has magnitude and direction and is most
often represented by an arrow on a sample space or scattergram. The head
of the arrow indicates the direction, and the length of the arrow represents
the magnitude of the quantity.
Venn diagram-a drawing, used in Boolean algebra, to represent sets,
their elements, and operations concerning them. For example, in the
universe represented by the box below, there are some elements that fall in
set A, some in set B, some in both, and some in neither.
Neithor A nor
verification-(see: hypothesis testing)
weighting-a ranking of observations according to their importance or
some other criteria, before computation is performed. In multiple
regression, for example, we may wish to compensate for possible errors in
the data by weighting. If an observation has error E, then weight W=1/E is
used so that the larger the error, the less weight the observation will have in
the total regression equation.
Yate's cqrrection-an adjustment made in the computation of a chi
square when the data base is small. If one or more of the expected cell
frequencies in a 2 x. 2 chi square matrix i,. less than 5, the normal ui:i square
computation will overstate the true relationship. Yate's correction is
designed to deflate this. (see: chi square)
Yule's Q-a quick statistical test which tells the approximate degree
and direction of the association between two variables. It is a relatively
unsophistict?.ted test, confined to 2 x 2 contingency tables (arrays which
display two dichotomous variables and the relationships between them). It
ranges from, --1 to +1, with 0 indicating no significant associations. For
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
pproved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3
example, the calculation of Yule's Q for sets of votes show political
affiliation to be strongly associated with a yes or no vote on issues I and 3.
On issue 2, however, there is no such marked relationship.
ISSUE I
Llb Cons
Yes
No
1
5
6
]
Yes
No
ISSUE II
Llb Cons
21
18
16
ISSUE III
Lib Cons
In Guttman scaling, items having a Yule's Q value of .80 or higher with each
other are scalable. (see: measure of association, GUTTMAN SCALING)
z score--a standardizedi measurement unit equal to one standard
deviation of a normal distribution, which is used to compare the results of
entirely different tests or measures. The z score is obtained by subtracting
the original raw score from its mean value, and dividing that result by the
standard deviation. Rict frequencies can be compared by unemployment
levels, for example, by converting both to z scores. (see: normal distribution,
standard deviation)
Normal Curve / 1 \ One Z=
0 +1z +2z +3z
MEAN
z test-a statistical test of the significance of data (or the probability
that the same data would occur solely by chance) found by changing the
data into standardized z scores and consulting a table of the probabilities of
various z values occurring by chance. If an analyst was checking the
hypothesis that the average age of world leaders was 55, for example, he
might sample 20 leaders and find the average of the sample to be 61. But is 61
significantly different from 55 or could the difference be due to chance in
selecting the sample? Changing both averages into z scores and comparing
them on the table, he finds there is a 2.5% probability of getting a sample
mean age of 61 or higher when the real mean age is 55. This chance is
Approved For Release 1999/09/25: CIA-RDP85T00875RO01100020001-3