The poll field work was conducted by Nielsen Bangladesh based in Dhaka, Bangladesh. Oversight and analysis was provided by David Williams of Williams & Associates of Salem, Massachusetts.

The population under study is representative for the adult population of Bangladesh and of all six administrative divisions of Bangladesh.

The urban sample accounts for 30 percent of respondents; the rural sample accounts for 70 percent of respondents.

The male-to-female ratio was 50 percent male to 50 percent female.

• 18-24 (20%)
• 25-29 (19%)
• 30-39 (26%)
• 40–49 (19%)
• 50-59 (10%)
• 60-69 (5%)
• 70+ (1%) 

The sample was distributed at the administrative division level, rural and urban, in all six administrative divisions of Bangladesh.  The sample was then post-weighted to make it proportionate to national representation by administrative division.

In rural areas, the survey team visited the central part of each selected cluster and randomly chose a direction from there by randomly selected starting points from a number of starting points or by spinning a bottle.  The first sample household was selected from among the households located along the chosen direction.  From each starting point, field team followed the left/right hand rule (depending on the availability of additional households).  Any one from the first five households was selected randomly. After that, the team intercepted every second household.

In urban areas, the same methodology was followed, except if the urban cluster area was divided into different portions by roads, lanes, or by-lanes, with each part containing about an equal population.  All such portions were then listed serially.

Within each selected household, the names of all household members were listed in descending order of their age.  Serial numbers were provided to all members aged 18 years and older in the descending order of age.  One respondent (voter) was selected randomly from each household using the last birthday method.

Sample size: 5,400

Eligibility criteria: age 18 and older

The margin of error is +/- ± 1.36 percent overall.

Face-to-face interviews were conducted.

The field work was conducted May 6-20, 2008.