The Truth Takes Time....(with surveys)

I've been having a great time reading the recently published Portfolios of the Poor by Murdoch et al (sorry CMF staff - I'm hoarding our copy for a few more weeks!). One of the interesting points the authors make early on in the book relates to research design, surveyor-surveyee trust and the quality of data. I think it has implications for all of us field researchers who rely on surveys to elicit quality data.

In the book the research team used a years' (or more) worth of financial diaries to piece together the financial lives and financial management of a number of poor households (spread across the globe). The team visited each household at least twice a month to capture its financial inflows, outflows, assets and liabilities. What I found fascinating was the opacity of information that the survey team received from the household during initial visits when compared to the information elicited after multiple visits. They write...

"It was sobering, then, to find that we would have missed much of the [financial] action had we undertaken only single, one-time interviews of each household. ..in the earliest interviews, we were often missing more than half of a household's financial activity in a given week. It took roughly six rounds of interviews and visits before we felt confident we had something close to the full story. It took time for our respondents to trust us..."1

This observation hit home for me, as part of an organization that relies on the accuracy of survey-based projects, most of which we administer once or twice to a household over the course of a year. What could we be missing out on?


1 pg 12 Collins, Murdoch al "Portfolios of the Poor" (2009)