This study was undertaken to assess the population’s knowledge, perceptions and attitudes toward social reconstruction and the ECCC, and related changes between 2008 and 2010. Specifically, the goals of the 2010 study were:
To achieve this objective, a survey of 1,000 adult Cambodians was conducted over the first 20 days of December 2010. The research followed the methodology used for a survey implemented by the research team in 2008. The 2008 survey serves as a baseline for comparison with the 2010 survey.
In 2008 the research team randomly selected respondents from among all adult (18 years of age or older) residents of Cambodia. As a sampling frame, the team used the Cambodia General Population Census 2008 Village’s database prepared by the National Institute of Statistics, Ministry of Planning. A multi-stage cluster sampling strategy was then designed to ensure that results would be representative of this population. The first stage involved researchers randomly selecting 125 communes out of 1,621 using systematic random sampling proportionate to population size. They selected at least one commune in each of the 24 provinces of Cambodia. Second, researchers randomly selected two villages from each commune using a simple random sampling procedure proportionate to population size, resulting in a sample of 250 villages. Then, within each village, the team randomly selected four households using linear systematic sampling with equal probability of selection. Finally, within each household, one respondent was selected for an interview at random using a Kish grid. 
In 2010, researchers returned to the same 250 villages selected in 2008. Within each village, four respondents were selected. All efforts were made to interview at least one participant of the 2008 survey. For the remaining respondents, up to three households were randomly selected and, within each household, one respondent was selected using the same sampling strategy as in 2008. The total sample size was 1,000 individuals with 300 people (30%) of the sample overlapping with the 2008 sample.
Interviewers attempted to contact randomly selected respondents three times over the course of a day. If a selected respondent could not be interviewed or refused to participate, the next randomly selected respondent was approached. One-on-one interviews were conducted anonymously in a confidential setting at or near respondents’ dwellings. The interviewer read to each of the selected respondents a standard consent form that presented the project (i.e., university-based research not affiliated with the government), stressed the confidentiality of the responses, and informed respondents of the voluntary nature of their participation (i.e., no compensation was provided and they were given the option to refuse to answer or stop the interview at any time). Verbal consent was obtained before the interview began.
Interviewers used a structured questionnaire similar to the one used in 2008. It covers the same nine topics and questions but adds questions related to the Duch trial and Civil Party participation. The topics include: (1) demographics; (2) needs and priorities; (3) justice and rule of law; (4) the Khmer Rouge regime; (5) the ECCC’s outreach efforts; (6) knowledge, attitudes, and perceptions of the ECCC; (7) establishment of a historical record, truth-telling, and reparations; (8) exposure to violence; and (9) mental health. Exposure to violence and mental health results are not discussed in this report. The instrument was first developed in English and then translated into Khmer. Back-translation and consultation with local experts ensured the quality of the translation. The questionnaire was further pre-tested with 67 randomly selected respondents in non-study sites to ensure that the instrument was culturally adapted and easily understandable to respondents.
The research team conducted the data collection in partnership with the Center for Advanced Study (CAS), a Cambodian-based, nonprofit survey research organization. A total of 25 experienced Cambodian interviewers and supervisors organized into five teams conducted the interviews. The interviewers and supervisors participated in a five-day training to familiarize themselves with the questionnaire, interview techniques, including non-suggestive probing, and selection of respondents. The training included discussion of the study objectives and concepts, mock interviews, and a survey pilot. This procedure enabled trainers to improve the survey instrument and to identify and improve weaknesses in interview techniques. In the field, supervisors provided oversight to ensure proper execution of household sampling procedures, as well as uniform application of the research protocol. To reduce data collection errors, interviewers checked the survey instrument for completeness and entry errors before leaving the selected household. Each team supervisor then performed a second review. The interview time per questionnaire ranged between 90 minutes to two hours. Finally, researchers analyzed data using the Statistical Package for Social Science (SPSS) version 18.0.
Limitations of the Study
While the study was conducted as rigorously as possible, some limitations must be acknowledged. First, the survey contained questions related to events that took place over 30 years ago. For respondents, the passage of time could have resulted in recall errors or created certain biases. However, we developed several questions to test the validity of their responses. These questions are discussed in further detail in the results and the discussion sections below.
While three attempts were made to contact selected respondents, not all selected individuals could be interviewed due to the limited time spent in each village. In the end, 397 households (23% of all households) had to be replaced because no one was home (90%), no one was eligible (2%), all members refused to be interviewed (5%), or no one was available (3%). In addition, within selected households, 347 individuals (20%) had to be replaced by the next randomly selected individual within the same household because the respondent was not home (97%), or refused to participate in the interview (3%).
Because views and attitudes are influenced by local and contemporaneous events (such as media coverage), this report represents only a snapshot of views as they were at the time of the survey. The data collection for the survey started on December 1, 2010, fewer than two weeks after the stampede at Diamond Island (Koh Pich).  Although not related to the ECCC, the tragedy may have affected some respondents’ state of mind and thus some of their responses to the survey. Finally, while the survey questions were finalized following piloting, respondents were free to interpret the questions according to their own understandings of the terms used. Careful choice of phrasing and translation and in-depth consultations with local experts about terminology, however, reduced the risk of misinterpretation of the questions.
 Communes (Khum) are an administrative division regrouping villages.
 The Kish grid is a selection method in which all eligible participants are ordered by sex and age and assigned a number. Interviewers use a selection table to randomly select a number and the corresponding eligible participant is interviewed.
 The study protocol was reviewed and approved by the Committee for the Protection of Human Subjects of the University of California, Berkeley. The Cambodian Ministry of Interior and local authorities in sampled villages approved the study but were not involved in the design of the study or in the analysis and presentation of the results.
 On November 22, 2010, in Phnom Penh, several hundred people were killed or injured while being caught in a stampede on a bridge between Diamond Island and the mainland. For details, see “Cambodia Stampede at Koh Pich on CNN”, YouTube, available at http://www.youtube.com/watch?v=GZFANh4bGYI