OPINION POLLING OVERSEAS

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP83-00856R000100070001-5
• Release Decision: RIPPUB
• Original Classification: S
• Document Page Count: 20
• Document Creation Date: December 21, 2016
• Document Release Date: October 29, 2008
• Sequence Number: 1
• Publication Date: September 1, 1982
• Content Type: REPORT

Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Directorate of Secret
Opinion Polling Overseas
ON TCli.r.
State Dept. review
completed
OR
Secret
CPAS 82-10008
September 1982
Copy 17
1
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
 Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
 Directorate of Secret
Intelligence
Opinion Polling Overseas
Office of Current Production and Analytic Support,
with contributions from Office of
Central Reference. Comments and queries are
welcome and may be directed to the Chief,
Methodology Center, Analytic Support Group,
This paper was prepared by
CPAS,
Directorate for Operations.
Secret
CPAS 82-10008
September 1982
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
 Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Opinion Polling Overseas
Summary Approximately 2,000 firms are now engaged in marketing research and
Information available opinion polling worldwide, and their number is expanding by about 10
as of 20 September 1982 percent per year. Many of these firms were founded immediately before or
was used in this report.
after World War H. They have over 30 years of polling experience and
data to contribute to assessments of current and prospective trends in
public attitudes and behavior. The vast majority of firms are active in only
one country, but substantial linkages exist among polling organizations
25X1 through interested individuals, opinion polling consortiums, and market
regularly conduct such surveys, often on a weekly basis, using random or
form of survey analysis worldwide. All of the firms
The quantitative mass survey, in which a sample is drawn according to
predetermined criteria from a national population, is the most prevalent
quota samples of the national population. 25X1
While the reliability of polling data must be carefully assessed, data from
established firms generally are accurate within the normal limits of
sampling error. Polling firms base their commercial viability on their
ability to satisfy businesses with the quality, that is. the accuracy.
timeliness, and cost, of their market research
Secret
CPAS 82-10008
September 1982
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
 Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Secret
Opinion Polling in Selected Foreign Countries
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
 Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Secret
Opinion Polling Overseas
Colombia, Ecuador, Mexico, Peru, Venezuela, and
Uruguay. These firms poll on a wide range of
Foreign Public Opinion Polling
Polling Consortiums
The largest polling consortium is the Gallup Interna-
tional Research Institutes (GIRI) founded at Lox-
wood, England, in May 1947. Initially sponsored by
the Gallup Institute at Princeton, GIRI is a loosely
affiliated group of independent opinion polling firms
in over 30 countries.' GIRI meets annually to ex-
change information on polling and to assess the
professional caliber of candidate firms. New firms are
admitted to GIRI (but never more than one per
country) only after they successfully complete a three-
year probationary period.
The individual Gallup affiliates are more responsive
to national demands for opinion polling and market
research than to directives or suggestions from GIRI.
Nevertheless, regional Gallup organizations are at-
tempting to poll simultaneously on comparable issues:
? Under the auspices of the European Community,
Gallup affiliates poll biannually on public percep-
tions of Community and national issues, life satis-
faction, political orientation, and expectations about
the future. Often these European barometers devote
much space to issues such as the European Parlia-
ment, the role of women in society, the balance
between national and community institutions in
European decisionmaking, and the status of children
in Europe.'
? Gallup affiliates in Central and Latin America have
recently organized into a consortium (Gallup Sud) to
run comparable surveys in Argentina, Brazil, Chile,
Gallup affiliates are active in Argentina, Austria, Australia,
Brazil, Canada, Chile, Colombia, Denmark, Ecuador, Finland,
France, West Germany, India, Ireland, Ivory Coast, Israel, Italy,
Japan, South Korea, Kuwait, Mexico, Netherlands, Norway, Peru,
Philippines, Portugal, Spain, Sweden, Switzerland, Uruguay, and
' The Euro-Barometer is currently run in the spring and fall of each
year by Gallup affiliates or other reputable firms in Belgium,
Denmark, France, Greece, West Germany, Ireland, Italy, Luxem-
25X1 1
25X1
domestic and international issues.
In addition to GIRI several other multinational poll-
ing consortiums exist. Prominent among these are
International Research Associates (INRA) with 11
affiliates in Western Europe (Austria, Belgium, Den-
mark, France, West Germany, Netherlands, Norway,
Spain, Sweden, Switzerland, and the United King-
dom); the London-based Survey Research Group,
Ltd., with affiliates in most of the major nations of
Southeast Asia (Malaysia, Singapore, Hong Kong, the
Philippines, Indonesia, and Thailand); and Research
International with 18 affiliates in Western Europe
A few firms, such as Market and Opinion Research
International (MORI), poll in many countries. Al-
though based in London and heavily involved in the
United Kingdom since its establishment in 1969,
MORI has conducted polls in nearly 30 foreign
countries. Another firm, Marcomer, polls in metropol-
itan France and former French colonies including
Algeria, Gabon, and the Ivory Coast. Marcomer's
clients are many and varied. In January 1981, for
example, Marcomer assisted the Democratic Party of
Gabon in surveying an urban sample asking respond-
ents to state their confidence in various Gabonese
political leaders and institutions.
International Contacts
Opinion assessment also is internationally linked
through individuals and societies. George Gallup and
Hedley Cantril have helped expand professional atti-
tude polling from the United States to Western
Europe and then to the newly developing states.
Pioneers in this field received support beginning in the
late 1940s with the founding of GIRI and the estab-
lishment of the European Society of Market and
Opinion Research (ESOMAR) in 1948. The World
Association of Public Opinion Research (WAPOR)
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
 Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
was founded under the auspices of the UN Education-
al, Scientific, and Cultural Organization (UNESCO)
in 1947 as a companion organization to the American
Association of Public Opinion Research. UNESCO
has continued to sponsor international meetings and
conferences on public opinion research.
International conferences, periodic exchanges, and the
publication of journals, reference materials, and news-
letters on a global scale have defined opinion assess-
ment as a profession and have further enhanced the
international exchange of analytic techniques. Today,
a number of periodicals exist for transmitting public
opinion data and research approaches internationally.
Prominent among these are Public Opinion Quarter-
ly, Public Opinion, World Opinion Update, Index to
International Public Opinion, and The International
Gallup Polls.
The US Government has promoted systematic assess-
ment of opinion in foreign countries since the end of
World War 11. In part, this was a natural outgrowth
of American efforts to introduce democratic norms
and institutions in West Germany and Japan. Numer-
ous national surveys were conducted in what is now
West Germany under the auspices of the US High
Commissioner. American political and military au-
thorities were supportive of efforts by the fledgling
EMNID Institute and the Institute for Demoskopie
(Allensbach) to begin national opinion polling in the
late 1940s. Both are now the leading opinion polling
firms in West Germany.
The main impetus from the United States toward
international opinion surveying has come through the
work of the US Information Agency (USIA) and its
various parent organizations. Directed from Washing-
ton, USIA has regularly used foreign firms to conduct
parallel polls in a large number of West European and
developing nations. It uses the data to assess foreign
public attitudes about the United States and about
important issues in international affairs.'
' The comparability of USIA-sponsored polls may vary from case to
case depending on question wording, sample and questionnaire
design, and the national political context in which the surveys were
Another element of opinion research is the expansion
of international survey data archives, which document
and store data collected by individual scholars or
research organizations. In addition to the holdings on
US opinion of the Institute for Social Research at the
University of Michigan, the Inter-University Consor-
tium for Political and Social Research also maintains
a substantial and growing collection of foreign survey
and elections data.' Other leading archiving groups
are the Roper Center, the Zentralarchiv fuer
Emprirische Sozialforschung in Cologne, and the
Survey Archive of the Social Science Research Coun-
cil at the University of Essex, England, all of which
maintain regular data exchange programs and are at
the forefront of software development for manipulat-
ing and analyzing opinion polls.
Market Research and Opinion Polling
Market research differs from opinion polling only in
the nature of the clients and the types of questions
asked. Market research is the formal and quantitative
sampling of a national, regional, or local population to
evaluate the suitability and probable commercial suc-
cess of various products. Using similar sampling
techniques, opinion pollsters focus on public attitudes
and likely future behaviors on issues of political,
social, and economic importance. Typical among such
concerns are party identification, electoral choice,
support for the government, and attitudes toward
specific policies. Most opinion surveys are made pub-
lic, while the results of market research are kept
confidential. In fact, most opinion polls are conducted
with the express purpose of informing the public-
typically, through the newspapers-rather than mere-
ly enlightening the poll's sponsors.
All opinion polling firms conduct market research,
but most market research firms do not conduct opin-
ion polls, preferring instead to work confidentially for
commercial clients. Among firms that do both, reve-
nues from commercial surveying largely underwrite
political polling. Market research enables the firm to
develop the interviewing staffs, sampling frame, and
data processing capabilities that can then be used by
other clients for political polling.
25X1
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
 Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Secret
In West Germany, for example, where firms like the
EMNID Institute and the Institute for Demoskopie
(Allensbach) have been active for over 30 years,
opinion polling accounts for only 3 to 5 percent of
their annual revenues. Even in France, which has a
reputation for being "poll-happy," formal opinion
assessment accounts for a small proportion of total
revenues. With the exception of Louis Harris-France
and the Institute Francais de Recherche Economiques
et Sociales (IFRES), political opinion polling accounts
for small proportions of total research activity
25X1 (see table 1).
Funding Public Opinion Polls
In almost all instances, survey organizations conduct
polls on behalf of specific clients. Few polling firms
can afford to poll public opinion on political issues,
whether out of curiosity or to support scholarly in-
quiry. The public press and political parties are
usually the major clients for opinion polling firms
Firms that conduct market research surveys establish
the infrastructure and custom of direct attitude as-
sessment, and their data may be used in charting
social and economic trends. Market research firms
regularly collect demographic data about the respond-
ents, including age, income, education, residence, and
occupation. Systematically collected, these surveys
can tell much about socioeconomic trends that may
affect political behavior.
Many firms are willing to sell data, publications, and
their polling service to the US Government. Others
make their data routinely available to survey archives
and in a limited number of cases to commercial
distributors. The West German firm Demoskopie
(Allensbach), for example, commonly provides its
surveys to the Bundesarchiv (federal archives) in
This ad hoc collection, although large, lacks timeliness. Time lags
between initial data collection and availability to users is more
often measured in years than in weeks or months. Of course,
archived data can provide a historical context for current informa-
tion on public opinion and may also be useful for establishing
Table I
France: Opinion Polling as a Proportion
of Total Revenue of Polling Firms
Institute Francais de Recherche
Economiques et Sociales (IFRES)
Public 12
Institute Francais d'Opinion 10
Publique (IFOP)
Societe Francais de'Etudes
par Sondages (SOFRES)
Indice Opinion
Publiemetrie
public distribution.
Koblenz, although these data will not be publicly
available for 30 years. Irish Marketing Surveys depos-
its its data with the Survey Archive in Essex for
Many polling firms are tied closely to particular
parties. An example is MORI, which has established
close links with the British Labor Party. Before the
general election in May 1979, MORI conducted 10
"instant" polls that were generally discussed the
following morning at 10 Downing Street, reflecting
the timeliness with which polling data can be made
available and the access that pollsters have to the top
political leadership.
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
 Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Utility of Opinion Polls
Foreign leaders use public opinion polls to describe,
explain, and predict public attitudes and behavior
across a wide range of issues. Intensive analysis of poll
data provides insights into why people think and act
the way they do.
The predictions of pollsters are part of the decision
processes of all aspiring political figures, particularly
in parliamentary systems. Governments often adjust
Opposition parties also poll extensively on their
Secret 25X1 4
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
 Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Secret
Political leaders use opinion polls to assess their
standing with the populace and their own party.
survey analysis worldwide.
ber of clients may place their questions.
Government leaders frequently commission polls to
assess the effects of their words as well as their deeds.
national and international significance.
Opinion Sampling Techniques
Polling firms assess public opinion qualitatively
through in-depth interviews and quantitatively with
national samples.
Qualitative, in-depth interviews focus on a relatively
small number of people (usually about 100), take two
to three hours to complete, and are relatively expen-
sive (about $150 to $200 per interview depending on
country and complexity of issues). They are often
undertaken to examine in more detail important
findings of larger sample surveys or to better prepare
for such large-scale surveys. Qualitative surveys fre-
quently focus on a small group within the population
that is of special interest to the client.
25X1 25X1 25X1
The quantitative mass survey, in which a sample is
drawn according to predetermined criteria from a
national population, is the most prevalent form of
The polling firm has already developed an interview-
ing infrastructure-interviewing staff and sampling
points, which are villages, city blocks, and so forth-
and sells space on the questionnaire to commercial
clients. Clients then place their own "rider" questions
on the firm's omnibus poll. Thus, respondents face a
set of questions that range from consumer issues, such
as the proverbial toothpaste question, to issues of
Interested parties can commission ad hoc surveys, but
these are more expensive than the omnibus method.
Ad hoc polls are justified if the client wants a
different sample than that available through the
omnibus instrument. Thus, if one wanted to examine
the position of selected groups in the West German
population on theater nuclear forces, it would be
possible at some cost to stratify the sample to include
more young people, more Bavarians, more blue-collar
workers, or more from other socioeconomic groups. =
Polling firms prefer quota sampling, or a blend of 25X1
random and quota sampling, to a purely random
method for several reasons. First, random sampling is
a more complex process. To be genuinely random, a
A quota sample is a set of individuals chosen from a target list
(such as a census register or electoral records) on the basis of their
standing in particular population groups. Most commonly, these
individuals are selected on the basis of their sex, age, socioeconomic
class, or the region of the country in which they live. Interviewers
work to meet their quotas for types of respondents within randomly
selected sampling points rather than specific individuals. Quota
samples are virtually equivalent to purely random samples in a
statistical sense. A random sample results from the selection of
individuals one at a time from a list of the national population.
Such lists are either the census files or electoral registers. In a
random sample, each individual has an equal probability of being
selected without regard to geographic location. Quota samples are
preferred for commercial polling because of their relatively low cost
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
 Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
sample must be drawn person by person from the
population until a sufficient number has been select-
ed. Second, random sampling immeasurably compli-
cates the interviewing process since interviewers must
contact specific individuals and this can necessitate
several return visits. Third, randomly selected individ-
uals are geographically dispersed in a national poll-a
requirement that imposes substantial travel costs in
medium-sized societies (West Germany, France, and
the United Kingdom) and in larger ones (the United
States, Australia, and Canada). Hence, most sampling
points or districts are randomly selected from national
lists. Within each of those districts interviewers are
Figure 1
Sample Size and Sampling Error
charged with finding quotas of individuals to meet 4
specified criteria-age, sex, and socioeconomic class.
Actual inteviewing is geographically concentrated 3
with interviewers free to choose respondents who meet 2
the quotas.
Increasing sample size can enhance the statistical
reliability of the data and allow for more discrete
breakdowns of the sample for comparative analysis.
Sampling error decreases as sample size increases, but
with decreasing marginal returns. As the sample size
expands, the gain in precision shrinks, particularly
when the sample exceeds 1,200 (see figure 1).
The quality of the sample should be evaluated by
comparing the survey results to information provided
by a national census and be evaluated with reference
to actual behavior that, for instance, the poll may
cover. Only in countries with reliable data on the
socioeconomic distribution of their own populations
can one evaluate the quality of the sampling frame. In
a number of important developing nations, such as
Lebanon and Nigeria, the collection and publication
of national census data have proved troublesome
political issues that are not likely to be resolved to the
satisfaction of the pollsters.
One can evaluate the merits of random and quota
samples by comparing the results of the survey with
actual behavior subsequent to the taking of the poll.
The behavioral test is clearly more relevant to election
forecasting than to assessing foreign policy beliefs.
With elections at least, there is subsequent behavior
0 0 0 0 0 0 0 0 0 0 0 0 8 8 8 8 88 0 0 0 0
o'n o'noo 000000 0000 00000
.- .- N N M V Vt ' r W o N lD 00 O ~/1 O o 0 0 0 0
,--,--.-. .-. -? N N M 7 Vl ~D 1~ 00
that can be measured. For example, in May 1979, 25X1
MORI forecast from quota samples that the Conserv-
atives would receive 44.4 percent of the vote in the
British general election. Working with a random
sample, Marplan projected a 45-percent vote share.
The actual result was 44.9 percent for the Conserva-
tive Party. The difference in the accuracy of the
forecast was trivial and did not justify the greater cost
of the Marplan effort.
How Reliable Are Opinion Polls?
A crucial issue for those who commission and those
who use public opinion surveys is the quality of the
data. Do sample surveys accurately represent public
views? The evidence suggests that, on balance, opin-
ion polls do portray the views of the mass populace
reasonably well at the time that the survey is taken.
Whenever possible, a current poll should be reviewed
in the light of previous findings. Once analyzed, it
should become part of a data base that can display
trends over time.
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
 Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Secret
Although the reliability of polling data must be
carefully assessed, data from established firms gener-
ally are accurate within the normal limits of sampling
25X1 error. Polling firms base their commercial viability on
their ability to satisfy businesses with the quality-
that is, accuracy, timeliness, and cost-of their mar-
From Reliable Polls to Election Forecasts
The public reputation of polling firms in the United
States and overseas is based on their success in
electoral forecasting, in part because that is their most
visible activity. Although hardly infallible, survey
25X1 firms in major industrial countries have generally
forecast elections accurately.
In the May 1979 British general election, the average
rate of error between final forecasts and the actual
result for each of the three major parties (Conserva-
tive, Labor, and Liberal) was 1 percent. These projec-
tions were made independently by MORI, Marplan,
Social Surveys (Gallup), and the National Opinion
Poll. Social Surveys correctly called the outcome of
nine of the last 11 national elections in the United
Kingdom (see tables 3 and 4):
25X1 French polling firms have reached a similar level of
professionalism, particularly in presidential elec-
tions (see table 5).'
? The Roy Morgan Research Centre (Gallup) correct-
ly predicted the winner in eight Australian national
elections between 1958 and 1975. Moreover, the
average error in predicting the gap in vote share
between the two main parties was only 3.6 percent.
'To be accurate, almost everyone, including most pollsters, missed
the 1978 French legislative elections. The projected leftist landslide
did not materialize, and this was caught only in a last minute
SOFRES poll, which was too late to affect much of the public
? The Swedish Institute of Public Opinion (SIFO) was
also accurate in forecasting the outcome of four
national elections held during the 1970s (see table
6).
Although data are not available, we believe similar
firms in developed, open societies are of equal profes-
sional caliber. Because opinion polling is an interna-
tional business, there has been a widespread transfer
of expertise, survey and interview methodology, and
data processing technology. The growth of profession-
al associations has helped to improve the quality of
international opinion polling. Competition in both the
national and international markets for opinion data
winnows out firms whose work is below known and
measurable professional standards. Although these
observations are most relevant to polling firms in
Western Europe, as developing societies modernize we
expect the quality and quantity of their opinion
research will increase as well.
At the same time, any electoral forecast based on
polling data should be treated cautiously-no matter
how reliable the polling firm-for several reasons.F-
Polls report only attitudes or personal assessments of
probable future behavior.' Actual behavior can be
affected by circumstances beyond an individual's
control, and one's attitudes may change between the
time the polls are conducted and the actual balloting
takes place.
Every survey contains a statistically normal level of
sampling error. Although this error margin can be
reduced, at a cost (see figure 1), analysis of survey
data should be done with this in mind.
Finally, the reliable collection of opinion data should
be distinguished from its analysis and presentation.
Improper or biased analysis and presentation of opin-
There are, of course, "exit polls," which ask voters how they voted
as they leave the polling stations. This is one method for gaining
insights into the results a few hours before the official results are
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
 Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Table 3
United Kingdom:
Election Forecasts, May 1979
Date of
Poll
Conservative
Labor Liberal
Average
Error
Marplan
1 May
45
1-2 May
46
Gallup
1-2 May
43
search International (MORI)
Result
Source: David Butler and Dennis Kavanagh, The British General
Election o/'1979 (London: Macmillan Press, 1980): 264-5.
dence that the originally collected data are unreli-
Opinion Polling in Selected
Foreign Countries
able.'
Several variables-including the distribution across
parties of the total vote-are of interest besides which
party wins." Especially in multiparty, parliamentary
systems, the seat distribution effectively determines
political power in the legislature. Even in systems
based on proportional representation, a 1-to-1 rela-
tionship between vote and seat share does not neces-
sarily exist. One might also be interested in changes in
support for political parties among various groups in
the population. A perennial question in France is how
voters from the French Communist Party (PCF) will
cast their votes in the second round of the presidential
election.
turnout and then the distribution of the vote by party among those
that come to the polls. Forecasts are improved to the extent that
turnout is required (Australia, Belgium, and Italy, for instance) or is
highly stable as it is in West Germany at a consistent 90 percent of
registered voters. Moreover, the accuracy of forecasts improves as
the elections draw near, as more people make up their minds and
fewer opportunities exist for events to alter those intentions. C
United Kingdom
Polling is a well-established research tradition in
Britain. Public opinion assessment began in the
United Kingdom with the founding of the British
Institute of Public Opinion in 1937. Founded two
years after its American parent, British Gallup (now
Social Surveys, Ltd.), was the first firm of its kind in
Britain. Until 1961, British Gallup published its poll
results in the News Chronicle; since 1961, it has
regularly published findings in the Daily Telegraph.
Through the mid-1950s, Gallup conducted monthly
omnibus surveys. Since then it has conducted weekly
Index, which has been published since 1960."
polls of at least 1,000 individuals on major national
and international issues. Many of these results are
disseminated through the monthly Gallup Political
" The results of many Gallup polls have been collected for publica-
tion by George H. Gallup in The Gallup International Public
Opinion Polls: Great Britain 1937-1975, 2 vols. (New York:
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
 Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Secret
Britain's role in world affairs, and its relationship to
the European Community.
Each Social Surveys poll also collects information on
the characteristics of the respondent. To assess the
sources of political loyalty and support, pollsters
profile the Conservative voter, the ecologist, the new
Social Democrat, and other groups by regularly col-
lecting respondent age, income, trade union member-
ship, occupation, education, religion, social class and
standard-of-living indicators. These data can comple-
ment the explicitly political data or can be directly
analyzed to indicate socioeconomic changes between
censuses.
Beginning in the 1960s, Social Surveys began to face
competition from other polling firms, notably Market
and Opinion Research International (MORI) and the
National Opinion Poll (NOP). MORI polls a national
sample of the United Kingdom, a sample of Scotland,
and samples of specific constituencies to support
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Table 4
United Kingdom:
Gallup's Forecasting Record, 1945-79
1945
47.0 Labor
49.0 Labor
-2.0 Labor
1950
45.0 Labor
46.8 Labor
-1.8 Labor
1951
49.5 Conservative
49.3 Labor
-2.2 Labor
1955
51.0 Conservative
49.3 Conservative
1.7 Conservative
1959
49.5 Conservative
48.8 Conservative
0.7 Conservative
1964
46.5 Labor
44.8 Labor
1.7 Labor
1966
51.0 Labor
48.9 Labor
2.1 Labor
1970
49.0 Labor
46.2 Conservative
-4.2 Conservative
1974 (February)
39.5 Conservative
38.6 Conservative
0.9 Conservative
1974 (October)
41.5 Labor
40.2 Labor
1.3 Labor
1979
43.0 Conservative
44.9 Conservative
- 1.9 Conservative
a Error rates are presented for the party receiving the largest share of
the vote in a given election. It is not always the party that
subsequently formed the government.
Social Surveys has conducted over 2,000 surveys, with
over 1,000 respondents in each survey answering
approximately 25 questions on diverse topics. A
standard core of political questions on satisfaction
with the performance of the prime minister and the
government, political party preference, and appraisal
of the performance of the opposition leader(s) are
asked on a weekly basis. These questions are often
phrased in different ways or with additional supple-
mentary questions as circumstances warrant. Satis-
faction with the performance of the prime minister is
often joined with questions about the ability of the
prime minister to lead on a particular issue. Approval
of government performance may be disaggregated to
25X1 probe for variations on specific policies.
Basic political questions are supplemented by ad hoc
queries on specific issues or by rotating questions on a
monthly rather than a weekly basis. Social Surveys
routinely adds questions about specific policy con-
cerns facing the United Kingdom. Recent examples of
this include theater nuclear weapons, the budget,
social issues such as race relations and housing,
 Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Table 5
France: Reliability of
Electoral Forecasts
Date Number of Average Forecasting
Parties Error (percent)
1973 Legislative (First Round)
URP
38.84
37
36
1.84-2.84
PS (Socialist Party)
19.16
21
23
1.84-3.84
PCF (French
Communist Party)
21.29
20
19
1.29-2.29
1974 Presidential (Second Round)
Giscard
50
50.9
50.6
0.1-0.2
Mitterrand
49
49.4
49.1
0.1-0.2
1978 Legislative (First Round)
UDP (Union of
Democrats for the
Republic)
21.5
20
20
1.50
RPR (Rally for the
Republic)
22.6
22
22
0.60
UDF (Union for
French Democracy)
27.6
30
33
2.40-5.40
PR (Radical Party)
16.3
16
14
0.30-2.30
PS
24.7
28
27
2.30-3.30
PCF
20.5
20
21
0.50
analysis of elections or to assess public opinion on
local issues. On request, it will also conduct ad hoc
polls in Northern Ireland. Its findings are regularly
published in The Times, The Sunday Times, The
Standard, The Scotsman, and a monthly bulletin,
Public Opinion in Great Britain. This bulletin con.
tains detailed poll results, time-series data on regular-
ly asked questions, commentary, and technical discus-
sion.
NOP, which evolved from the Media Research De-
partment of Associated Newspapers, Ltd., began poll-
ing regularly in 1959 and accurately forecast the 1959
general election. Since then, it has conducted polls on
a regular monthly basis to assess the central political
questions of the day, including questions on unem-
ployment, the Social Democrats, Britain and the
Community, and international security issues. The
results of NOP surveys are regularly reported in its
Political, Social and Economic Review.
France
Polling in France is similar in scale and sophistication
to that in the United Kingdom or the United States.
This is due, in large part, to the reputation of the two
leading firms-the French Institute of Public Opinion
(IFOP) founded in 1938 and the French Society for
Public Opinion Research (SOFRES), which has been
active since 1962. Smaller, less well-known firms
include Louis-Harris France (started in 1977), Public
(1976), Indice Opinion (1980), and Publiemetrie 25X1
(1970). In addition, the Ministry of the Interior's
Reseignments Generaus are charged with periodic
reporting on public attitudes and behavior in each of
the 100 prefectures.12
Opinion polling took off in France at the time of the
1965 Presidential election, which was the first since
World War II with universal suffrage. The frenetic
pace of polling since then, and the complexities of the
two-round electoral system, occasioned the establish-
ment in 1977 of a national polling commission (Com- 25X1
mission des Sondages) within the Ministry of Justice.
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
 Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Secret
25X1
This polling commission is charged with monitoring
the objectivity and the quality of polls taken and
published during election campaigns. As a further
check on the undue influence of public polls on the
democratic process, French law prohibits the publica-
tion of poll results during the week prior to both the
first and second rounds of legislative and presidential
elections.
Two types of polls dominate French surveys, or at
least published materials about polling-popularity or
image polls, and voter-intention polls. Popularity or
image polls ask respondents to identify their most
admired political leaders, the qualities they associate
with particular candidates, and to express their degree
of confidence in particular leaders. Voter-intention
polls ask for a respondent's choice on political party if
a legislative election is imminent or on an individual
candidate if the upcoming election is presidential. In
addition, these polls present hypothetical election
contests to assess alternative races in both the first
25X1 and seconds rounds.
eign policy issues.
West Germany
Public opinion polling in West Germany is rooted in
the US Government's efforts to build democratic
institutions in the Allied zones after World War II.
Although the Nazi security service (Sicherheitsdienst)
maintained a sophisticated network of informers regu-
larly reporting to Berlin, formal opinion polling began
with the establishment of the Emnid Institute in 1945
and the Institut fuer Demoskopie (Allensbach) in
1947-two firms that still dominate opinion polling in
West Germany. EMNID, with almost 40 years of
polling on political topics as a member of the interna-
tional Gallup consortium publishes Aktueller Poli-
tischer Dienst and Emnid Informationen on alterna-
tive months. These periodicals analyze opinion trends
and provide time-series data on party identification,
support for the government, satisfaction with the
performance of the chancellor, and attitudes on for-
More than 100 professional market research and
opinion polling organizations are now active through-
out West Germany. As in most other countries, these
organizations devote 90 to 95 percent of their re-
sources to market research. A number of research
centers provide the professional and technical infra-
structure for their polling activity. Polls are routinely
analyzed by academic experts at the Zentralarchiv
fuer Emprische Sozialforschung (Cologne) and the
Zentrum fuer Umfragen, Methoden and Analysen
(Mannheim), as well as several universities.
University-sponsored panel studies on West German
elections and analysis based on data collected by
others provide a quality check on materials collected
and published by the major polling firms.
India
Opinion polling in India is largely the preserve of the
Indian Institute of Public Opinion (IIPO), a member
of the international Gallup affiliates which has con-
ducted periodic opinion surveys since 1955. The re-
sults of these surveys are regularly published in
IIPO's own bulletin, Public Opinion. Other institutes
active in opinion assessment and market research-
such as the Institute of Marketing and Management,
International Research Associates, and the Indian
Market Research Bureau-lack IIPO's experience in
attitudinal surveying that focuses on political and
social issues.
IIPO regularly conducts surveys on a wide variety of
topics, including voter intentions at each Lok Saliba
(lower house) election since 1961, midterm polls in
constituencies with byelections, international issues
for USIA, and Gallup's end-of-year survey on popular
expectations. The firm conducts several standard ur-
ban surveys each year (Bombay, Calcutta, Madras,
and New Delhi) and conducts two all-India surveys
(with 4,000 respondents each) twice yearly.
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
 Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Spain
Opinion polling in Spain-conducted by the Centro
de Investigaciones Sociologicas (CIS), Data S. A., and
ICSA Gallup-is a fledgling institution in a new
democracy:
and the family. The bulk of formal opinion research in
the socialist countries is concerned with either demo-
graphic analysis or market research-not political
? CIS, which began polling in the early 1960s, mainly
concentrated on international issues (for example,
Gibraltar and the EC), religion, and tourism until
the mid-1970s. Since then it has increasingly polled
on domestic political issues-such as party identifi-
cation, satisfaction with government performance,
and popular views on policy alternatives. In June
1979, a political barometer was developed by CIS
consisting of a national poll of 1,200 Spanish adults
that is run six times a year. The firm also under-
takes monthly surveys, which it summarizes in the
Revista Espanola de Investigaciones Sociologicas.
and political surveys.
most of its political data are available in a
volume entitled Evolucion Politica Espanola
1975-80. The company will conduct omnibus polls
for any client as well as vend existing data on social
and political topics.
? ICSA Gallup polls on a weekly basis and publishes
the results in a monthly bulletin, using a quota
sample of about 2,000 adults drawn from the penin-
sula and the Balearic Islands. Interviews are con-
ducted in person in the home of the respondent, a
procedure that is used by the the best polling firms
overseas. Polling data on political parties, available
since March 1980, have increased in depth and
breadth.
Polling in Communist Countries
Opinion assessment in Eastern Europe and the Soviet
Union is monopolized by state research institutes.
Most surveys in Eastern Europe have not dealt with
such basic issues as the legitimacy of the regime, the
validity of Marxism-Leninism as a guide for public
life, or the appropriateness of the Soviet military
presence abroad. They have focused on such issues as
consumer satisfaction and expectations, attitudes of
the youth, workers and the workplace, and marriage
polling as it is understood in the West.
Although opinion polling had been undertaken by
East European research institutes for some time,
interest in opinion assessment as an aid in policymak-
ing grew in the 1970s following contacts with Western
scholars and the acquisition of the computer hardware
and software needed for intensive data analysis. In
1976 the Institute for Public Opinion Research was
founded as part of East Germany's agitation and
propaganda apparatus and was tasked with keeping
the leadership of the Socialist Unity Party fully
informed about public attitudes. Scattered high-level
support exists in the Soviet Union for research
through opinion surveys on "what the people think
and what they want." A Pravda article in late 1982
cited public opinion as "a sensitive barometer whose
readings, if properly analyzed, can tell about deep
social processes that at times seem hardly noticed."
Polls have been completed on both the elites and the 25X1
mass publics in Eastern Europe and the Soviet Union:
? In April 1972 East Germany polled a sample of
administrative and several scientific personnel on 25X1
key political issues. These elite respondents, who
benefit most from the status quo, were asked wheth-
er East Germany was a democratic state, whether
individual views were heeded, whether friendship
with the USSR was beneficial, and to make com-
parisons between East Germany and West
Germany.
? The USSR Academy of Sciences surveyed 1,500
individuals in the Moscow region on the sources of
information on which the respondents rely, the
problems and concerns of daily life, and their
assessment of the most important problems facing
the Soviet Union. This poll was published in the
Journal of Sociological Research.
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
 Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Secret
Public opinion assessment exploded in Poland during
the two years prior to the December 1981 declaration
of martial law. Most surveys have been sponsored by
Solidarity or by an external actor such as Paris
Match. Polls by the former focused on the attitudes of
union members while those conducted for the foreign
press have attempted to obtain a national sample. In
1981 a Solidarity poll of its own members showed
extensive support for the union and the Catholic
Church and much less trust in the party and the state.
Another survey in the fall of 1981 indicated that 75
percent of the Polish people supported Solidarity,
although there was a small increase in the number of
people willing to assign responsibility, in part, for the
25X1 current crisis to the labor union (table 7).
Table 7
Trust of Solidarity Members
in Polish Institutions
Police
Government
Polish Socialist Workers Party
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
 Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Iq
Next 7 Page(s) In Document Denied
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
 Secret
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5
Secret
Approved For Release 2008/10/29: CIA-RDP83-00856R000100070001-5