impressive respondent tracking methods (or, why not to wait 30 years to try and find your people)

In the late 1960s, Janice Perlman interviewed 750 people in three favelas of Rio de Janeiro. Thirty years later, she decided to revisit the same people

"We used many approaches and kept following up on leads for the next two years. After securing permission to enter the communities, we began by putting up large colorful posters saying that we were eager to meet people who had participated in the 1968 study. The poster featured a photo of me … from that time and the cover of the Brazilian paperback edition of my book, which I had given to the study participants. Fortunately, due to a newspaper story about my return to Rio to follow up on my favela study, I was invited to be interviewed on the popular Brazilian television show Fantástico. That gave me the opportunity to reach a huge audience and appear to viewers to call in if they or someone they knew had been part of the original study. Likewise, I spoke on popular radio stations and community radio, and we posted ads in the most widely read newspapers in the communities. Then, for one Sunday afternoon in each place our team rented a van with a loudspeaker…and drove around each of the communities announcing a barbecue that afternoon and asking for help in our search for original study participants." (Favela, p49)

She also used more standard methods, using the old contact info, visiting the old neighborhoods: "This was complicated by the fact that many of the dwellings had no numbers or street name at all, and even when we had a partial address, the streets and numbering had long ago changed or been reconfigured, often several times over".

After all that and much more (for example, finding people with the right first name but who didn’t remember if they had participated), they found just 41% of the 750 study participants. This is extraordinary, and yet it seriously compromises the ability to draw inferences from the outcomes of these folks: Are these just the most successful (i.e., not homeless)? or the least successful (i.e., not moved on)?. Perlman recognizes this and tries to deal with it by tracing the overall composition of the favelas over time in addition to just the welfare of her found participants. I applaud Perlman and her efforts.

BUT, having worked with tracking surveys in a Kenya and Brazil and now starting some work in Mexico, the lesson is clear: Don’t wait 30 years to follow up on your people.