Follow us on Twitter

Methodology

You are here

The selection of respondents for the survey was based on a standard random multi-stage cluster sampling procedure. We first selected Enumeration Areas (EAs), which are geographic areas developed for census canvassing by the Liberian government.[1] Researchers used the latest available information on each EA’s population size and geographic location (including GPS coordinates) to select EAs randomly in proportion to the overall population size. A total of 260 EAs were selected throughout the country. The number of EAs assigned to each county reflected the population size of that county relative to the national population, with a minimum of 12 EAs per county.[2]

At the second stage, researchers selected 16 households within each EA using a geographic method (EPI method). Interviewers identified the center of the EA and randomly selected a direction. In that direction, interviewers would select every other settlement unit. In each selected household (defined as a group of people normally sleeping under the same roof and eating together), interviewers randomly selected one adult to be interviewed from a list of all eligible respondents. Three attempts were made to contact a household or individual before replacing them with another. Due to the sensitivity of some questions, the interviewers were assigned to same-sex respondents. Thus, male interviewers were assigned to male respondents and female interviewers were assigned to female respondents.

Figure 2: Sample Distribution

Sample distribution

Within the 260 EAs, a total of 4,955 settlements (houses) were approached. Among them, interviews were conducted in 4,501 households (91%). No interviews were conducted in 454 cases (9%), most frequently because the settlement was clearly abandoned (45% of the cases), because no one was eligible (28%), or because all the eligible members of the household refused to participate (23%). From the 4,501 selected, eligible, and agreeable households, the teams interviewed 4,501 respondents out of 4,789 who were approached (94%). Most of the selected individuals who did not participate in the survey were away and could not be contacted within the survey timeline (61%), although some also refused to participate (28%).

The Committee for the Protection of Human Subjects at the University of California, Berkeley reviewed and approved the study protocol. The Liberian Ministry of Internal Affairs granted permission to implement the research. Approval to conduct interviews was also obtained from local authorities at each survey site. The interviewers obtained oral informed consent from each selected participant; neither monetary nor material incentives were offered for participation.

Research Instruments

The interviewers collected information using a standardized structured questionnaire with open-ended questions. The questionnaire included 16 sections and took an average of one hour and five minutes to administer. The sections were: (1) demographics, (2) socio-economic status, (3) priorities and services, (4) health, (5) security, (6) social and ethnic relations, (7) elections, (8) disputes (general), (9) land disputes, (10) other disputes and domestic violence, (11) the justice system, (12) peacebuilding, (13) the Truth and Reconciliation Commission, (14) exposure to violence, (15) measures to help survivors, and (16) trauma.

A team with expertise in this type of research developed the questionnaire after consultation with local experts and representatives from non-governmental organizations (NGOs), the government, and multilateral institutions. Response options based on pilot interviews were given to interviewers for coding purposes, but interviewers did not read these options to study participants except for questions employing a scaling format (e.g., Likert scale). Every question had an open-ended field to record complete responses. The researchers prepared the questionnaire and consent documents in English, and local translators then adapted them into Liberian English to ensure that the language would be appropriate for respondents with no or limited education. Finally, the team validated the instruments using independent back-translation and pilot surveys.

Once the questionnaire was finalized, it was programmed into a Personal Digital Assistant (PDA) using KoBo, our custom data collection package.[3] The use of PDAs allows enumerators to enter the data directly as they conduct interviews. The forms contain a built-in verification system that reduces the risk of skipping questions or entering erroneous values, resulting in higher quality of data. Daily synchronization with a central computer allows the lead researchers to check data for consistency and outliers during data collection.

Data Collection and Analysis

Data collection took place over six weeks, from November 1 to December 13, 2010. Ten teams of four individuals (two men and two women per team) implemented the study under the guidance of the lead researchers and field supervisors. The interviewers were Liberian university students or professionals with research experience. Prior to collecting the data they participated in a seven-day training workshop that explained the objectives and content of the study, survey and interview techniques, use of the PDA, troubleshooting, and solving technical problems. The training included mock interviews and pilot-testing with randomly selected individuals at non-sampled sites.

At the survey sites, the research plan required each interviewer to conduct four interviews per day. They conducted the interviews one-on-one, anonymously, and in a confidential setting. At the end of each day of data collection, the supervisor of each research team electronically aggregated the data and sent it to a central database. Once the data collection was completed, the database was imported into Statistical Package for Social Science (SPSS) version 16 for data analyses. All results presented here account for the complex sampling methodology and weight factors.

Limitations

The present study was developed and implemented carefully to ensure that the results would accurately represent the views and opinions of the adult Liberian population. We must, however, acknowledge some limitations. First, the sample was designed rigorously to be representative of the adult population resident of Liberia, resulting in a large sample size of 4,501 respondents. However some selected individuals could not be interviewed for various reasons (see sample section). It is uncertain how respondents who could not be interviewed differ from the sampled individuals, but we designed the sampling approach to reduce any potential selection bias. Results are valid for the Liberian population only at the time of the survey. Opinions may change over time. Second, the study relies on a self-reported method of data collection, including key informant interviews and a population-based survey. A number of factors may have affected the quality and validity of the data collected, such as inaccurate recall of past events, misunderstanding of the questions or concepts, reactivity to the interviewer due to the sensitive nature of the questions, or intentional misreporting (i.e., for socially unacceptable answers). We minimized such risks through careful development of the questionnaire to make the questions sufficiently clear and reduce potential bias (see “Research Instruments”). The interviewers conducted the interviews anonymously, and all interviewers underwent rigorous training on data collection and interview techniques.

 



[1] Available at the Liberia Institute of Statistics & Geo-Information Services, 2009.

[2] The number of EAs selected and the number of interviews per EA were computed to reach a minimum target sample size of 4,160. Additional interviews were randomly conducted to adjust for non-response and incomplete interviews. At the end of the survey, a total of 4,501 interviews had been completed.

[3] Since 2007, the authors have developed KoBo, a set of tools to facilitate electronic data collection based on Open Data Kit. The tool was first piloted in northern Uganda and bears the Acholi name, KoBo, which means “transfer.”