World Values Survey 2006, Wave 5

For the purpose of this research, the population frame was defined as all adult citizens (18 years and more) who lived in Iraqi urban and rural areas. The sample consisted of 2880 respondents in 18 governorates. The number of cases to be included for each governorate was determined by considering:

1. each governorates population as a percentage of the overall population;

2. the number of cases necessary to make valid regional comparisons; and

3. the rural area population as a percentage of the governorate population. A multi-stage probability-based sample was drawn utilizing residential listings from Iraqs 1997 Population Census. Six sampling stages were deployed. First, the number of interviews was distributed among census districts (Qada) proportionally. Second, each Qada consisted of a number of census sub districts called (Nahia) which received its share of interviews proportionally. Nahias, in turn, consist of many blocks. At the third stage, blocks were regarded as the primary sampling units (PSU) in the urban areas, with 96 PSUs being selected using probability-proportional-to-size procedures. Thirty, twenty, or ten interviews were conducted in each block. Because more than seven years had passed since the census, the residential list in each street was not necessarily accurate, so each interviewer was asked to draw a map for the households located in the street he\she was supposed to work in. In the fifth stage, each interviewer selected, randomly, five households in the street using random tables. Finally, within each selected household, one respondent was randomly selected using the last birthday method. Because many elderly Iraqi citizens do not know their exact birthday, a list of random birthdays was used by interviewers when necessary. Due to the inaccuracy of the addresses in the rural areas, interviewers were trained on how to use simple rules in choosing the household which should be interviewed. This was also practiced on the urban Kurdistan cities which were not covered by the 1997 national census. Standard rules have been followed to choose blocks, streets and households. The rule of choosing the respondent is the same in all 2880 interviews. The margin of error was calculated in order to

4. incorporate the clustering effects of multistage sample design, and

5. to provide 95% confidence level. Although the overall margin of error is estimated at ±3.017% (n=2701), the margin of error varies across the 18 governorates included in the sample.

The sample size for Iraq is N=2701 and includes the national population aged 18 years and over for both sexes.