Location Codes for Market Research and Field Surveys

Field research is about people in places. A household survey needs to reach the sampled dwellings; a retail audit needs to visit a list of shops; an observational study needs to count at a spot on a pavement. Teams work with enumerators who may be new to the area, and quality control depends on knowing that the visit really happened where it should.

Three tasks where a code can help

Sampling and assignment. A sample frame may list dwellings or shops that have no reliable street address. If the frame includes a pin for each, an enumerator can be given a short list of codes instead of a description.

Locating the respondent. In areas without house numbers, a code tells the enumerator which door is the sampled one, and avoids the temptation to substitute a more convenient household.

Quality control. A supervisor who receives the code of each completed interview can spot-check visits and compare them with the assignment.

Design choices

Decide early how precisely you will store locations. A precise code identifies a household, which can make survey data personal and sensitive. Many studies keep the code only in the operational system for fieldwork and remove or coarsen it before analysis. Your ethics approval and consent text should say what is recorded and why.

Fieldwork practicalities

• Pre-load codes into the assignment list before the team goes out, because signal may be poor.
• Record a corrected code when an enumerator finds that the dwelling is at a different place.
• Use labels to separate the dwelling, a landmark and a meeting point.
• Have a fallback for places in countries that UPCS does not support.

What it does not do

A code does not validate that a respondent is eligible, does not replace a sample design and says nothing about the quality of an interview. It is not a GPS audit trail, since someone can pick a point on a map without being there.

Technical teams can look at the lookup and generate endpoints in the documentation. Related pages: surveying and mapping teams and NGOs and humanitarian work.

Pilot in one cluster

Pilot the approach in one cluster before fieldwork begins. Check that enumerators can open the codes in the field, that corrections are recorded and that the data you hold matches what you promised respondents.