Executive
Summary
This survey of voter knowledge and awareness was conducted for two reasons. One was to alert the voter education community to areas where, in light of our findings, education needs seemed to be greatest.
The second purpose of our study was to establish a
baseline of voter knowledge against which the impact of the voter education campaign
could be assessed, in order to improve and strengthen future voter education
campaigns in
We
reported the progress of our study and our preliminary findings to the
fortnightly meetings of the Joint Voter Education Committee throughout the
pre-election period. We reported the
highpoints of the final results of our survey in a brief Khmer and English
publication, which was distributed widely to voter educators through COMFREL,
COFFEL and KID. In that publication we
stressed the challenges implied for voter education efforts and included
recommendations for action that might be taken. (The highpoints are reproduced
in the annex to this report).
We
envisage a follow-up impact survey that would attempt to draw the lessons from
this campaign by considering the views of voters regarding their experience
with the registration and voting process.
The general aim would be to learn how successful the voter education
campaign was in helping voters understand and exercise their voting rights effectively. At the same time, we would want to learn how
successful specific educational contributions of the National Election
Committee and the NGO’s were in reaching and informing the electorate.
From
the point of view of laying the basis for an eventual impact assessment of
voter education effectiveness, the findings of this baseline survey can be
summarized under two rubrics. The first
is knowledge of the procedures for the elections and awareness of the election
process. The second is the channels and
means by which information on the election was obtained. The references in the summary below are to
sections and subsections of the main report.
A. Procedures and Process of the Elections
1. Registration.
We
found that the electorate was largely uninformed about any registration
procedures (III.A.3.Q.50), which was understandable because our survey was
conducted in March and April 1998, before the National Elections Committee was
fully formed and functional. However, we
found widespread understanding of the need to re-register for the 1998
elections (III.A.1.Q.49). Our survey
showed that 92% of respondents expressed an intention to register (III.A.2.Q.52).
2. Specific voter knowledge.
Our
survey showed that 93% of respondents expressed an intention to vote in the
1998 elections (IV.A.1.Q.69), and they seemed very well aware of the fact that
elections were going to be held soon (IV.A.2.Q.53). But there was a massive ignorance about
voting procedures (IV.B.1.Q.67, 68), with 91% of respondents saying they did
not know the voting procedures and 96% saying they did not know where to vote.
3. General voter knowledge.
Voters
also seemed quite uninformed about what the 1998 would be for—for persons,
parties or offices (IV.B.4.Q.54), and when elections would be held
(IV.B.5.Q.55). Of our respondents, 95%
did not know the name of their Member of the National Assembly, suggesting a
significant gap in understanding of how a vote translates into the election of
a candidate who is accountable to a constituency (IV.B.6.Q.77).
4. Feelings about pressure, intimidation,
danger.
Our
respondents generally did not expect either pressure or force to be applied on
them to affect their voting (V.B.1.Q.111, 112).
We found 91% of respondents said they would not obey an order from a
powerful person to vote for a particular party (V.B.2.Q.118), and 82% said
their vote would not be influenced by a gift of money (V.B.3.Q.113).
Only
19% of our respondents expressed fearfulness about the upcoming elections
compared to 62% who said they were not fearful (VI.B.1.Q.81). Only 15% thought elections were dangerous
compared to 35% who held that they were not dangerous. But, on the other hand,
50% of our respondents took a wait-and-see attitude to this question and said
they didn’t know if elections would be dangerous (VI.B.2.Q.85).
5. Attitudes about obligations to vote for a
party.
The
electorate surveyed by CAS seemed deeply uncertain about the voting obligations
entailed by a party membership commitment (like a thumbprint on a form), or a
pledge in public to support a party. We
found that 28% of respondents said party members must vote for that party,
while 24% said no, party members need not vote for their party and 48% said
they didn’t know (V.B.4.Q.71). We found
that 35% of respondents felt a person must vote in accord with his or her
pledge, while 38% said no, a pledge was not binding, and 27% said they didn’t
know (V.B.5.Q.72).
6. Attitudes about the secrecy of the ballot.
The
electorate we surveyed seemed fairly evenly divided between 47% who thought the
ballot in the forthcoming elections would be secret and 52% who said they
didn’t know if the elections would be secret or not (VI.A.Q.116). This may be compared to 78% of our respondents
who thought the UNTAC elections were secret versus 4% who thought they were not
secret (II.D.1.Q114). This division on
the question of ballot secrecy is echoed by 45% of our respondents who thought
a winning party would not punish those who had voted for another party
(assuming the winner somehow had learned who had voted against him), and the
46% of our respondents who said they didn’t know if a winning party would
punish those who voted for other parties (VI.B.4.Q.84).
7. Knowledge and awareness of female voters.
For
questions involving knowledge of procedures or awareness of the political
process in
In
regard to voting procedures, 95% of our female respondents said they didn’t
know what they were, compared to 85% of male respondents (IV.B.1.67, 68). In regard to obligations entailed by party
membership, 52% of our female respondents gave the “don’t know” reply, compared
to 42% of male respondents (V.B.4.Q.71).
In
regard to expectations about the secrecy of the ballot, 57% of our female
respondents replied “don’t know,” compared to 44% of male respondents
(VI.A.Q.116). In regard to the
expectation that a winning party might punish voters for other parties, 52.5%
of female respondents answered “don’t know,” compared to 36% of the males
(VI.B.4.Q.84).
In
regard to an understanding of how a multiparty system works, 64.7% of our
female respondents replied “don’t know” to a question about whether a loser in
one election could stand again in another election. Of our male respondents, 45.4% gave this
“don’t know” reply (VI.B.5.Q.73).
See
also VII.A.2.Q.122, 123 on perceptions of men's need to advise women on voting.
8. Awareness of the voting rights of vulnerable
groups and minorities.
Our
survey showed overwhelming affirmation of the voting rights of women
(VII.A.1.Q.120, 121), the very poor (VII.B.Q.132, 133), the illiterate (VII.C.Q.130,
131) and the disabled (VII.D.Q.134, 135).
However
in regard to minorities, the acknowledgement of their voting rights was
diminished by some uncertainty among our respondents. Regarding non-Buddhists
(like the Muslim Cham minority), 32% of our respondents gave the “don’t know”
reply to a question about their right to vote (VII.E.1.Q.136). Regarding the highland tribal minorities, 37% of our respondents gave the “don’t
know” reply to a question about the voting rights of these indigenous peoples
(VII.E.2.Q.137). Regarding immigrant Vietnamese, 64% of our
respondents gave the “don’t know” response to a question about the voting
rights of this group (VII.E.3.Q.138).
Regarding overseas Cambodians
or “dual passport holders,” 58% of our respondents gave the “don’t know” reply
to the question about the voting rights of these Cambodians (VII.E.4.Q139).
B. Channels for information and effectiveness of
voter education messages.
1. Registration.
Besides
having been too young, the major reason our respondents gave for not having
voted in the UNTAC elections was not having registered to vote
(II.A.2.Q.27). A follow-up should ask
why respondents did not vote in 1998. Our
findings suggest that the impact assessment should consider how effectively reliable
information about registration was provided to the electorate.
2. The elections.
Our
survey included questions on how voters had learned about the UNTAC elections
(II.B.Q37). We learned of the importance
of government officials, radio and friends and relatives as sources of
information. Our survey also asked how
voters had become aware of the forthcoming elections and by what sources
respondents obtained information about the elections (IV.A.3.Q.56, IV.B.2.Q.148). We again noted the relative importance of
electronic media, radio, television, and contact with friends, relatives and
neighbors.
A
follow-up should ask in detail what information activities of COMFREL, COFFEL,
NEC, the political parties and government officials made the most memorable
impression as far as providing useful guidance to voters in these elections.
FINAL
REPORT
BASELINE
SURVEY OF
VOTER
KNOWLEDGE AND AWARENESS
Introduction
This survey was designed with two aims in mind. First, to serve as a pre-test of the knowledge
level of the Cambodian electorate regarding various aspects of registration and
balloting. We hoped to highlight the
topics about which the electorate was particularly uninformed, and to draw
attention to sectors of the population with specific voter education
needs. The second aim is to provide a
baseline upon which a post-test of voter knowledge could be conducted after the
voter education campaign. The post-test
would assess the success of the voter education campaign and provide lessons
for future voter education campaigns in
The
first aim, to serve the needs of the voter education community in their
pre-election activities, was served by a publication of a brief paper in Khmer
and English. That paper, Highpoints of the Baseline Survey of Voter
Knowledge and Awareness, CAS Occasional Paper No. 2, (May 1998) is included
as an annex to this Final Report. We
noted the highpoints of our findings that were particularly relevant to stress
in voter education training so those gaps in voter knowledge revealed by our
findings could be recognized and remedies found. We distributed 1000 copies of this paper
widely in the voter education community, especially to COFFEL and COMFREL and
KID, which had training programs at the province, district and commune
level. We tried to assure that our
findings would be in the hands of all voter education trainers down to the
commune level, at a time when village-level volunteers were being trained.
The
second aim, to establish a baseline against which an impact study of voter
education effectiveness could be conducted is provided in this detailed report
of our findings.
The
survey instrument we used was devised in English in the course of meetings with
voter education curriculum developers in Cambodian NGOs, in light of the
materials they were developing and using.
After a pre-test of the Khmer translation of the instrument, the
instrument was streamlined and simplified.
Methodological
Considerations
Instrument Construction
The
pre-test of the Khmer version of the questionnaire revealed an aspect of Khmer
culture that may be useful to note for other survey-based studies. We discovered that the Khmer language of the
pre-test instrument had been written in a high literary register. This reflected the academic and intellectual
qualities of our researchers who translated the questions from English. As excellent translations, the Khmer
questions had a hypercorrect, formal, literary character.
But
the questions in Khmer were also formal for another reason, which was to assure
proper courtesy to whatever reader might encounter the document. During the pre-test, I noticed that the
interviewers first read the question, and then “translated” the question into
colloquial Khmer of the oral register.
This translation subtly took into consideration who the informant was,
and how the informant reacted during the reading and then the translation of
the questions.
We
soon realized that each interviewer was going to add his or her own
interpretations in their translations from the literary register to the oral
register, depending on the researcher’s assessment of the respondent’s reaction
to the questions.
In
order to minimize this unwanted element of diversity introduced by the
interviewers themselves, the questions had to be recast in an acceptable oral
form, but in writing. This proved to be
very awkward for my researchers. They
worried that if a person of high status or power should see the written
questions in such an oral form, that person might get angry with the researcher
because the obligatory marks of formality (encoding deference) expected in a
message addressed to him, were absent.
Our
efforts to simplify what could be said in the Khmer question fed back to an
adjustment of what the English question could ask. Questions that attempted to ask for scaled
responses and to probe nuances in attitude that were part of the original
questionnaire design gave way to more straight forward questions that could be
asked and answered simply in Khmer. The
English and Khmer versions of our instrument are reproduced in an annex to this
report.
Sampling
Lacking
a census, and within a limited budget and timeframe, we were unable to design a
probability sample. The guidelines or quotas
we used in selecting categories of informants reflected the aims of the
research. We wanted to assess voter
knowledge and awareness especially in remote rural locations. We wanted to be certain to include voter-aged
women, disabled and minorities. These
are the kinds of groups usually underrepresented in a non-probability sample.
Our
aim was to replicate the general characteristics of the Cambodian population,
at least as far as gender, rural/urban, Khmer/non-Khmer characteristics were
concerned. Our guide here was the Demographic Survey of
Our
scope covered a considerable diversity of Cambodian provinces. The sixteen provinces (or municipalities) and
forty-seven districts (or sangkat) we
visited are listed in an annex. One
criterion we considered was to survey different ecological zones, where the
basis of livelihood might be a little different, for instance, flood plain
villages near the lake and upland rain dependent rice areas. This was not carried out according to a
systematic plan however. The main
practical criterion of our selection of villages to survey was that they be
accessible to a four-wheel drive vehicle during the dry months March and
April. Mindful of our mission to
highlight voter education needs, our team probably biased the selection in
favor of communities far from surfaced roads, but on laterite roads, away from
urban areas and into
At
the end of a series of weeklong surveys to Northwest, South, Southeast, and
Survey Execution
Our
team consisted of eight persons. There
were two expatriates, an American academic who was project director, and a
Dutch UNV who worked on the data with SPSS 7.5 for Windows. The Khmer interviewers consisted of a senior
male academic who was the field coordinator, another male and female academic
and a young man and two young women who were graduates. These interviewers would usually arrive in a
village and each work singly to find men and women, old and young, at home or
at work, who would agree to be interviewed.
The team usually worked in at least two different locales each day,
separated by a lunch break. The afternoon
village visits were usually the most productive because by
Our
researchers found that they had to overcome a typical initial assumption by the
interviewees that the questionnaire was being conducted on behalf of a
political party. On these grounds a
number of people we approached refused to participate in the survey. Many of our pre-test respondents had balked
at questions which had originally been designed to be answered "yes"
or "no." They forced us to
include a "don’t know" reply to many of these questions. As our survey proceeded, we were surprised at
how large the "don’t know" response was. We sensed that many of these responses
reflected a genuine lack of knowledge.
But we also sensed a degree of reluctance to commit to a "yes"
or "no" to us, unknown strangers.
Although we tried to assure respondents of the neutrality and
non-political stance of the Center for Advanced Study, the reader should keep
in mind the background of uncertainty about our motives that was probably never
entirely dispelled.
A
large proportion of those who did agree to complete the questionnaire was
reluctant to reply to the open ended questions.
We lacked the time needed to gain rapport and overcome the natural
wariness of peasants toward urban investigators. This probably accounts for the rarity of
responses to the open-ended questions, and the brevity of the responses we were
given.
On
the other hand, by the end of the interview, when we reached last question (“If
someone came here to answer one question about the elections, what would you
ask?”), the closure of the interview often gave rise to intense requests for
more information about politics, voting and the elections. In general we sensed a great hunger for more
specific information about the elections, which we were in no position to
provide. Our impression overall was that
a voter education campaign was eagerly awaited by the electorate in the
villages we surveyed.
I. Background Demographics
The aim of these demographic descriptions of
our respondents is to indicate that the diverse elements of the Cambodian
population were taken into account and to show how the proportions in our
sample replicate the demographics of
A. General Characteristics
1. Q. 8. Rural-Urban distribution
This
dichotomy proved to be a thorny issue among our Khmer interviewers, as we
discovered in our pre-test analysis.
Long, inconclusive arguments about how the rural boundary should be defined
as one left a town were finally settled in practice by driving far enough along
dirt roads to communities that everyone agreed were “rural.” In the end, our distribution closely mirrors
the findings of The Demographic Survey,
which finds a 85.6% rural and 14.4% urban distribution, based on administrative
jurisdictions and boundaries.
2. Q. 9. Gender
The
gender distribution in our sample shows an overrepresentation of female
respondents, if compared to the Demographic
Survey, which finds 47.84% male and 52.16% female. Part of the explanation for this collection
error was due to our special concern to include the voice of women in our
survey, especially from remote rural areas.
But another reason for the discrepancy is also that the people our
interviewers were likely to meet during the day in a village were women.
3. Q. 12. Age
Our
sampling quotas encouraged the interviewers to attempt to replicate the actual
age distribution in the Cambodian population.
A chief criterion for selection as a respondent to our survey was,
obviously, that the candidate be a potential voter, 18 years old or older.
If
the Demographic Survey findings on
age distribution were recalculated to eliminate those under voting age, we
would find that the 18-24 group would be 20.18% of the voter aged population;
the 25-30 group would be 19.55%; the 31-40 group would be 24.34%; the 41-50
group would be 16.02%; and the 51+ group would be 19.97%. (An accumulation of rounding off errors
affects the precision of these figures).
The
age distribution of our respondents mirrors the general age distribution of
voter aged persons, but slightly over-represents the younger voters, 18-24
years old, and under-represents the oldest voters, 51+ years old. This again is due to our field methods. In a village during the day young women
householders with the time to answer a questionnaire are more likely to be
encountered than other age groups. Very
old people are also likely to be encountered in the village, but many of them
declined to be interviewed. They often
pleaded a lack of understanding or hardness of hearing. If they did agree to be interviewed, many of
these potential older informants were distracted after answering a few
questions and did not complete the questionnaire.
B. Family characteristics
1. Q. 15. Marital Status
The
marital status of our sample replicates the findings of the Demographic Survey fairly closely. The Demographic
Survey finds not yet married 28.3%; married 60.9%; separated/divorced 1.4%;
widowed 9.4%.
All
the charts above suggest that a fairly representative sample of the Cambodian
population was obtained, although we did not conduct the kind of probabilistic
sampling that would enable us to estimate with some level of confidence the
extent to which the findings in our survey were likely to differ from what we
would have found by studying the whole population.
C. Social characteristics
1. Q. 11. Ethnicity of Respondents
Non-probabilistic
sampling is especially likely to under-represent minorities. Accordingly we made special efforts to set
quotas for our sample that would include members of non-Khmer communities. Most of the CAS researchers on this baseline
survey project had worked on an earlier CAS project to study the ethnic
minorities of
While
voting age was stated explicitly as a criterion for participation in the
survey, we did not raise the issue of whether the respondent was actually
legally entitled to vote. This is an
issue for voter registration officials to decide. We were interested in the level of knowledge
and awareness of the elections in these communities and wanted to get enough
respondents from minority peoples to determine if they had distinctive voter
education needs. Our proportion of 86%
Khmer to 14% minority may be slightly disproportionate in favor of minority
groups.
There
is much controversy about the statistics on minorities in
2. Q. 14. Years of Schooling
Educational
level was included in the survey to enable us to see if there were any trends
in the cross-tabulations of schooling with the knowledge, awareness and
attitudes assessed in the survey. Those
trends will be mentioned below. We included
schooling in a Wat (or other religious school) as well as in a government
school in this variable.
3.
Q. 16. Occupation
No
specific quotas for occupation or profession were set for the
interviewing. This chart confirms the
mainly rural focus of our research, but also indicates by the diversity of occupations
represented that non-farming respondents were not neglected. The category “other” mainly includes people
who marked two or more occupations on the questionnaire (like farmer and
trader), but also includes people whose occupation was other that any of the
categories listed on the questionnaire.
4.
Q. 25. Income
We
included questions on roof type, cooking fuel, possession of rice fields, oxen
and some durable consumer goods, which enabled us to discern fine differentiation
in socio-economic level in the rural, farming population. However, from the point of view of attempting
to identify simple associations of socio-economic status and voter attitudes,
the most useful variable was income. Our category “low” corresponds to a stated
income of 0-50$/month. “High”
corresponds to $51 and above/month.
It
must be noted that most farmers actually handle little cash over a year, giving
rise to a response of very low income.
What the income category chart probably really indicates is a division
between those in a semi-subsistence economy and those in a cash economy.
II. Voter Perceptions of UNTAC elections
We
included questions on the UNTAC election to develop some sense of a comparison
with attitudes toward the forthcoming elections. From the point of view of the implications
for the voter education campaign that we wished to serve, we were especially
interested in why respondents had not voted in the UNTAC elections. From the point of view of assessing voter education
impact we were particularly interested in how the voters had obtained
information about the election procedures.
A. Participation in UNTAC elections
1. Q. 26. Experience with voting in UNTAC elections
If
the respondents to our survey who were too young to have voted in the UNTAC
elections are excluded, then the incidence of voting in the UNTAC elections
among our respondents who were of voting age in 1993 is 89%. Compare the estimate of 89.5% of voting
incidence given, for example, in Trevor Findlay, Cambodia: the Legacy and Lessons of UNTAC, SIPRI Research Report
No. 9 (OUP 1995, 82).
2. Q. 27. Reasons for failure to participate in UNTAC elections
The
large number of cases responding that they were too young (under 18) in 1993,
and who would now be between 18 and 23, is consistent with the age distribution
in our sample.
The
next most frequent reason given for not voting in UNTAC elections was not having
registered. These findings, taken with
the responses “not having understood what to do” and “did not know there was an
election,” present a challenge to voter educators to attempt to reach this
largely remote, rural electorate that missed the UNTAC election, with messages
that explain the need to register and clarify what the elections are about.
B. Voter Knowledge of UNTAC Elections
One
of the main aims of an impact survey of the voter education campaign will be to
identify the media and the messages that were most effective in reaching and
informing the electorate.
This
chart highlights the importance of local government officials, mainly of the
SoC, in informing the voters about the UNTAC elections. Those local authorities are, by and large,
still in their positions in the CPP dominated local government, so we would
expect an impact survey to show a similar strong response for local government
officials as a major source for election information.
The
importance of friends and relatives as sources of information is consistent
with the predominantly oral culture of Cambodian peasantry. Voter educators who would take advantage of
the natural gossip networks of rural communities would construct their messages
to assure that they could be transmitted effectively along these channels.
C. Influence on Voting
1. Q. 101. Using gifts to influence voters in UNTAC
Vote
buying is such a highly developed art in neighboring countries in the region that
we might expect the practice to appear in
It
will be interesting to follow-up this question in future surveys to see if a
vote-buying trend emerges.
2.
Q. 104. Using promises to
influence voters in UNTAC
An
election campaign is almost by definition characterized by promises to the
electorate. The promises we listed in
our questionnaire were developed during the pre-test with villagers and include
only fairly concrete or visible promised benefits. Future surveys, reflecting more Cambodian
experience with election campaigns, should probably include the incidence of
promises of more abstract benefits, like security, peace, welfare and
development.
D. UNTAC Election Milieu
1. Q. 114. Perceptions of the preservation of secrecy
The
electorate was extremely confident that the 1993 UNTAC balloting was
secret. This can be compared with the
prospects expected for a secret ballot by our respondents in the months
preceding the 1998 elections. (See
Section VI. A. Q.116 below).
Perceptions
of the secrecy of the balloting should obviously be followed up in a post-election
survey.