TASK FORCE 1-A - STATISTICAL METHODS IN D/S/TF

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP75-00662R000300060001-5
• Release Decision: RIPPUB
• Original Classification: K
• Document Page Count: 2
• Document Creation Date: November 9, 2016
• Document Release Date: July 9, 1998
• Sequence Number: 1
• Publication Date: April 12, 1951
• Content Type: MEMO

Sanitized - Approved For Release
25X1A9a 25X1A9a
: CIA-RDP75-00662R000300060001-5
111111 12 April 1951
Task Force 1-a -- Statistical Methods in D/S/TF
1. Certain economic information to be supplied by DA/TF hna related
branches of aRR is ultimately a number or a serios of numbers. Preeentation of
these numbers, as bell as their compilation, is neither in accore aith, nor
exhausts the poasibiaities of, stanuard statistical practices.
2. TheL;e practices include curtain standard tools, among which might
be listed averages, correlation anelysis, etc. Matoviel available to allow the
use of Lhese tools in most ca4,e8 eaists for the are-4orlu Aar II perioe an. in
storm ceses for a limited part of the post-dorld aar II period. Use of these
methods soulo :Allow the aetermineLion of certain basic osta on which eutimaLes
of the ra4anituae of current economic phenomena coulu DU -lore firmly founaeu.
Firmer foundations would result in firmer estimates.
3. These procsaaea, called statiatics, also poasesa certain rules,
aaequateiy set forth in any basic teat on Lhe subject. Theae rules, waneraliy,
are directoa at. unambiguous communication between the producer ane the consu-Jer
of the "number". They are well standardized, for experience in the use of taoae
numbers has pointed out certain man-.raps which await the careless user. Gen-
eral4 speakine, in WiE publications and in the TF part of Task Force 1, the
rules have not been observed. Hence, unanbiguous coartuuication between TF and
the consumars of its products ie impossible.
4. Reeommenaations
a. Re Rules: The basic aria steneard rules of eaatisticul pre-
sentation should be followed by all portions of VRR concerned *ith such processes.
Tais obeervance of stanaera rules should oe followed ey analysts at all levels
of prouuction and shoula ultimaLely os Impocied ank. enforced by the eeitorial
aectIons of URR.
b. i.e Methods:
1. The following shoule be absolutely determined:
a) Extent of data evailaOle for determination of Livee
basic measuree waich are or may oe of use in forecaating; i.e., averaaes of
various types, trends, if existent, etc.
b) If aaLa available be sufficient, the measures should
be calculatee ane current estimates eauminee in their light (if any).
C) Deciaien shoule be made az to the actual value of the
proceaL to both the reporter and. the estimator.
a) If expected value is founa to actually exist, pru-
Lr,.al for full utilization of this type statistical analysis should co leie uown
for the offices so concerned.
Sanitized - Approved For Release : CIA-RDP75-00662R000300060001-5
Sanitized - Approved For Release : CIA-RDP75-00662R000300060001-5
- 2 ?
2. Correlation anulysist Correlation analysis presumes u
fairly extensive and comprehensive set of tim series. If in the above process
they be found to exist, cork should be done to determine the practicability of
adding, to the foundations of estimates throuch tnis procesu. The necessary
work to so oetermihe would ue considerable and the chunces of satisfactory
results highly conjectural. If Ligh a4;rees of corAiltion be found, however,
tho conjecture of the estimator may 4e ?reatly reduceo.
?
Sanitized - Approved For Release : CIA-RDP75-00662R000300060001-5