Methodology

Overview

The Arab Barometer conducts high quality public opinion surveys in the Middle East and North Africa (MENA) since 2006. All country surveys are based on probability samples representative of citizens aged 18 or above. Each country surveyed includes approx. 1,200 respondents, on average. The error margin of our surveys is ± 3 percent. Interviews are conducted face-to-face in the respondent’s residence.

The Arab Barometer is the largest repository of publicly available data on the views of men and women in the MENA region. Our findings give a voice to the needs and concerns of Arab publics. To date, the Arab Barometer has conducted 36 national surveys over four waves including more than 45,000 interviews in 15 Arab countries.

Respondents

The surveys provide national coverage of citizens aged 18 and above in all counties. Thus, the universe is comprised of the population by governorate living in both urban and rural areas. The survey results are representative at the national level as well as the governorate level. See our Technical Reports by wave for detailed information by country.

The survey is designed include all non-institutionalized citizens. By necessity, it does not include some citizens who are inaccessible, including those who are in hospitals, care homes, or other collective living arrangements such as halls or live at military bases, or are inmates in the country’s prisons. In some cases, small populations living in remote areas of the country are excluded from the sample. This typically concerns areas with less than 2 percent of the total population. Depending on the context, some countries may also exclude other specific populations, such as internally displaced citizens. See our Technical Reports by wave for detailed information by country.

The unit of observation is an individual from a randomly selected household. Only one respondent per household is eligible to participate in the Arab Barometer survey.

Sampling Frame

Due to the absence of address registers in the region, which typically serve as sampling frames, the Arab Barometer countries rely on maps paired with the latest population estimates from the statistics authorities in the respective country covering all citizens ages 18 and above.

Sampling Strategy

Sampling strategies vary by country, but it is a requirement that probability samples are drawn. In general, the sampling strategies in each country ensure that every eligible respondent in the country has a known and calculable probability of being included in the sample that is unequal to zero.

Most countries apply multistage sampling methods based on stratification. Stratified sampling aims to better represent certain homogenous groups within a population, such as regions. The Arab Barometer surveys are generally stratified by governorate (or similar regional characteristic) and sub-stratified by type of settlement (urban/rural). In some countries, alternative substrata apply. For example, instead of stratifying by settlement type, Lebanon stratifies by religious sect.

Interviews are allocated to each stratum using probability proportional to size (PPS), i.e., each stratum allocates interviews relative to its population. Within each stratum, primary sampling units (PSUs) are randomly selected. The precise definition of the PSU unit differs across countries. In most cases, it is a pre-defined geographic unit that has been demarcated by the country’s central statistical authority. However, in some cases, it is a standard unit that has been demarcated by our local partner. Within each PSU, blocks are defined as the secondary sampling unit, normally containing between 100-250 households. Random starting points for random walks are then selected within these blocks to determine the household. Kish grids or first/last birthday methods randomize the selection of the respondent at the doorstep. See our Technical Reports by wave for further information on the precise sampling strategies in each country.

Mode of Data Collection

Interviews are conducted face-to-face. The first three waves were conducted using Paper and Pencil Interviewing (PAPI). The fourth wave employed a multiple mode design. While some countries remained in the PAPI mode, other countries shifted to Computer Assisted Personal Interviewing (PAPI). The fifth wave, currently in the field, will almost exclusively rely on CAPI. See our Technical Reports for each wave for specific country information.

Fieldwork

One key element of ensuring high data quality is the implementation of fieldwork quality assurance measures in each country. The Arab Barometer has instituted a rigorous process for interviewer briefings prior to fieldwork, including a multiday training led by members from our regional hub institutions attended by a member of the core research team. Pretesting and piloting are then carried out by our local teams and the results are carefully evaluated by the local team, Arab Barometer hub institutions, and the core research team. Fieldwork is regularly monitored by the local team leaders, regional hubs, and the core research team. While interviews are ongoing, supervisors oversee interviewers to ensure that they are carrying out the sampling plan. They also sit in on some interviews to ensure the instrument is being implemented correctly. At least 20 percent of respondents are also re-contacted and are administered an abbreviated version of the instrument.

Post-Processing

The Arab Barometer employs a number of quality control checks after fieldwork. All data set are checked using a number of post-assessment techniques. These include a number of sophisticated tests that look for unusual patterns in the data or paradata. Additionally, the team uses specialized software (Percent Match) that flags near duplicate observations. This technique was developed by the Director of the Arab Barometer to detect cases of likely data fabrication and has become a gold standard in the field of survey methodology. The fifth wave will also trial a live data checking tool applying these techniques as the data come in on a daily basis.