Section 1 Survey Dataset Identification
1.1 COUNTRY, YEAR
The country and year in which the survey was conducted. The following table shows in which countries and years the survey was conducted.
| Country/Year | 2013 | 2014 | 2015 | 2016 | 2017 | 2018 |
|---|---|---|---|---|---|---|
| Bangladesh | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) |
| India | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) |
| Indonesia | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | ||
| Kenya | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | |
| Nigeria | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | |
| Pakistan | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | |
| Tanzania | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | |
| Uganda | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) | \(\checkmark\) |
1.2 WEIGHT
The respondent’s survey weight to use in analyses. Each survey dataset contains its own set of weights. The weight, when implemented, increases or decreases the influence of a specific subgroup based on sample size. Weights were originally calculated in proportion to key regional or demographic census data.
1.3 UR
The respondent’s residence type
Levels:
- Urban
- Rural
1.4 LATITUDE, LONGITUDE
Anonymized latitude and longitude of the approximate household location
The method of anonymization is very similar to the one practiced by the Demographic Health Surveys project. All respondents within a cluster share the same coordinates. Cluster centers were initially calculated by averaging all longitudes and latitudes, then displacing the point randomly within a radius inside the second-level geographic area (e.g. counties). The maximum radius for displacement is within 2km for an urban area and 5km for a rural area, with a 1/100 chance of the maximum rural radius being 10km. This method is intended to preserve location anonymity while still being usable for basic spatial analysis purposes.