(Sanitized)FROM: A. W. MARSHALL

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP80M01133A000800110021-6
• Release Decision: RIPPUB
• Original Classification: K
• Document Page Count: 4
• Document Creation Date: December 20, 2016
• Document Release Date: November 16, 2006
• Sequence Number: 21
• Publication Date: November 10, 1972
• Content Type: MEMO

Approved For Re, 2007/03/08: CIA-RDP80M01133AQ800110021-6
NATION.-),L SECURITY COUNCIL
V\SHiNGTON,D.C. Z0506
November 10, 19l'
FROM: A. W. MARSHALL
After our October 26 meeting, Chip and I discussed your
plan to review NIEs and to develop techniques to poll con-
sumers. Both ideas are good and I would like to be kept
informed as your plans progress. Chip wrote me a short
paper presenting our initial ideas on polling consumers.
NSC review(s) completed.
25X1
Approved For Release 2007/03/08 : CIA-RDP80M01133A000800110021-6
 Approved FQ?Release 2007/03/08 : CIA-RDP80M0Ii A000800110021-6
MEMORANDUM
November 2, 1 72
MEMORANDUM FOR A. W. MARSHALL f'ROM: GEORGE E. PICKETT, JR.
SUBJECT: Quality Control and Feedback
I would like to comment on your discussion with I
October 26 concerning feedback and quality control.
There is a definite need for a quality control effort within the intel-
ligence community. The danger, however, is that the results of
such movements in the past have been to sterilize the product rather
than make it more useful. Quality control is frequently misinter-
preted as being standardization. Hence, it results in creating a
standard set of qualifying adjectives and adverbs, a standard writing
style (e. g. , the ubiquitous editor), a set of rules and procedures
about who has to clear what in order for it to be published, and a
limitation on experimentation with new techniques of analysis and
presentation. The fundamentals which are most frequently overlooked
are that ultimately real quality control is keyed to the consumer of a
product and that the diversity of consumers requires a quality control
system which allows diversity of output.
An effective system must start from the consumer perspective and
work back into the organization, rather than the usual reverse process.
Consumers have to be categorized and examined within the framework
of a fundamental question: What does each consumer need? Feedback
is definitely a problem in making such an assessment, but there are
techniques which can be used. In addition it is not critical to talk face
to face with each consumer. One can learn a great deal about a rra 's
needs by talking to those around him and by simply speculating abou"
the type of work he does and what he needs to do it.
Having collected data on what are the needs of various consumers (and
maintaining a continuous effort to update and expand this information)
one can begin to guide the efforts of intelligence producers. The major
problem of a standard product can be alleviated by classifying products
by the consumers who read them and then outlining the criteria whi -.h
]XI
25X1
must beiproved For Feease
e 2lOg~03/6:t8lA-&( i4s0~'OSD~``v2 td
 Y~ - Approved For Rase 2007/03/08 : CIA-RDP80M01133A,(0800110021-6
Editors can be given more explicit instructions. Several key mina~:e rs
can conduct sample reviews, picking a product at random and evalua'- n?9
its performance against need. Experimental products can be tested
on consumers as intelligence develops a better feel for their needs
than they themselves may have. Training material can be modi_'ied
to increase student awareness that he is providing a service and must
be sensitive to his consumers needs. (e. g. , It would be interesting
to ask students to hypothesize on Kissinger's needs, his time to read.
and other sources of information he has; then ask them to evaluate the
appropriateness of the material he receives.)
The only bureaucratic aspect of quality control must be the require-
ment to continuously poll the consumer for his views. This feecibac.<
can then be translated into specific guidance keyed to each producer
and into criteria to evaluate the performance of analysts and manage-s.
Marketing people (especially those in the publications industry) and
business school instructors could provide thoughts on how to poll
consumers. They could probably be consulted without requiring major
security clearances. I can suggest the following tools in any case:
1. Extremely brief "yes-no" or "scale" questionnaires coul3
be used to reach busy people. Essential information could be derived
such as: has he read the document; did it add to his knowledge; were the
points clear; did he find it valuable enough to store, etc.?
2. Questionnaires of varying lengths could be used for other
consumers. What is important is to construct questionnaires which
are brief and keyed to the individual. For example, one type of
questionnaire would be used for analysts, one for staffs supporting
major consumers, and one for unfamiliar third parties (e. g., consultants).
3. Interviews of varying types can be conducted. Several readers
could be interviewed in depth; many could be called on the phone and
asked a few unclassified quick questions (simply to know if a documen
was read would be helpful); sampling could be used to reduce volume
of work.
Flexibility is critical. Questionnaires should add or delete items as
issues or problems seem to gain or lose importance. Polling should
Approved For Release 2007/03/08 : CIA-RDP80M01133A000800110021-6
 Approved For Rei"se 2007/03/08: CIA-RDP80M01133Ao800110021-6
initially sample the general views of people, then find key- nrob=em 4
and pursue these in greater detail. Short and long questionnaires a ni
different styles of interviewing should be intermixed in order to kee.
the effort alive. Different approaches should be used for the different
categories of publications: current products (e. g, CIB), ez5tim.tes
(e. g. , NIEs) and special products (e. g. , weapons studies).
I do not think its necessary for the polling of views by intelligence to
be a statistically intense, scientifically oriented effort. People car :oo
easily become embroiled in statistical analyses and discussions of the
psychological impact of questionnaires. Since the present efforts of
the community are so underdeveloped, I think it's far more important
that an aggressive, common sense approach toward eliciting consur-per
reaction be taken. A well managed effort will easily surface major
issues and problems in consumer demand.
Approved For Release 2007/03/08 : CIA-RDP80M01133A000800110021-6