Doing Surveys: A Practical Guide

Seeking to gather feedback or input, to assess perceptions before/after an activity, to validate a hypothesis, and/or to generate a hypothesis? This presentation, delivered at the Wellcome Fourth International Public Engagement Workshop, October 3 2012, Cape Town, South Africa, explores survey research: a social research method that can be used to support monitoring and evaluation (M&E). It involves asking people questions through any process ranging from a short, paper-based feedback form to an intensive, in-person, in-depth interview. Steps in a survey project are outlined; in brief, they include:

1. Establish the goals - what do you want to know & who wants to know?
2. Determine the sample - who do you want to talk to?
3. Choose the method - how will you engage with people? Considerations in selecting a method are described, such as speed (e.g., email and Web surveys are the fastest), cost (personal interviews are the most expensive), internet usage (it may not be possible to generalise results to the population as a whole), literacy levels (less-educated people may not respond to mail surveys), sensitive questions (a computer works best here), and video, sound, and graphics (a need to get reactions to these things limits options).
4. Decide what to ask - design the questions. Considerations revolve around choice of the medium, question flow (Keep It Short & Simple - KISS), language and cultural context, and ethical and legal considerations.
5. Pilot the questions - do they yield what you need?
6. 'Design' data capture point - what will you do the response?
7. Analyse – how will you analyse & report the findings?

The presentation moves on to explore several example survey questions to illustrate less and more effective ways of surveying people about issues such as: "In what ways could we enhance community engagement around Research Project X" low and middle income countries?" A multiple choice of "better facilities" or "more funding" or more training" is portrayed here as less than effective - the query is about "ways", but the participant is only allowed to select a single "way). Instead, one could ask: "To what extent do you agree that the following factors could enhance community engagement around Research Project X" low and middle income countries?" with several options (e.g., more communication with researchers, more advertising about projects) and several options per row (strongly agree through strongly disagree).

Several tips are offered in conclusion. For example, consider if incentives are appropriate to secure a good response.