Pakistan National Post-Election Survey (2018)

Gallup post-election survey of the 2018 national Pakistani parliamentary election.  These data include 6 substantive questions about Pakistan’s 2018 national election and about 10 demographic/ geographical variables per respondent. The questions were drafted in consultation with Gallup Pakistan, using the ANES as a guide.  Because of extended periods of military rule, Pakistan saw its first peaceful transition of power between civilian governments in 2013, 66 full years after independence. However, in 2013 there was no post-election survey. Although Gallup Pakistan has conducted exit polls, these data would essentially be Pakistan’s first post-election survey.  The data are collected via face-to-face surveys using “Computer Assisted Personal Interviewing with Handheld Devices”. These interviews are conducted all over Pakistan (excluding disputed territories and certain tribal areas). Gallup Pakistan’s business model is face-to-face surveys as opposed to telephone surveys because landline use in Pakistan is restricted to less than 10% and declining, mobile phone usage is relatively higher but 20% of the population do not use cell phones. Lastly, a large proportion of females use a male person’s mobile phone. This means that when interviewers call, they must convince the male person to allow them to speak to the female person (leading to high nonresponse). Internet surveys are also difficult in Pakistan because about 20% of the population uses the internet.
 
Gallup Pakistan is a Pakistani affiliate of the Gallup International Association (different from Gallup Inc in the US). Their website is gallup.com.pk.

The sample size is 2000. Gallup Pakistan has two primary sampling units (villages for rural areas and census block codes for urban areas). They randomly select a certain number of locations using probability proportional to size sampling (where more populous locations have a higher chance of being selected). Within these locations, they randomly select households and then use a kish grid to choose a member of the household of the total members of the household (not just those that are home at the time that the interviewers knock on the door). After data is collected, the sample is weighted based on the 2017 census. For non-response, they have a 2 call back per household system (and non-response rates are relatively low).