Exploring Mode Effects in Establishment Surveys
A lot goes into making a survey high-quality but also affordable. One choice that we made with the Kauffman Firm Survey to save on costs and try to achieve higher retention rates was to field a survey which used dual modes of collection - CATI or computer-assisted telephone interviewing and the internet. In doing so with our panel of nearly 5,000 businesses, we've been able to cut costs on the whole project and extend the period of data collection. In a field which desperately needs more longitudinal surveys, I think there are some lessons to be learned:
It takes time. Initially respondents were much less likely to choose web vs. CATI because they didn't know our effort and they didn't fully trust us. With time, the response rates via the web have gone from the low thirties to the upper sixties.
It saves money. The above mentioned transition has cut our actual cost per interview down considerably.
It is more convenient. We've been able to make it very easy for respondents to stay in our sample, thus achieving nearly 80 percent retention rates.
These are just a few reasons why this dual mode choice has been good for Kauffman. In a new research paper studying the effect of this choice and any biases that it might cause in the sample, researchers at Mathematica find very few indications that the CATI vs. web option (or those who switch their response choices) matters at all. Perhaps the biggest possible influences of mode effects are industry and employer vs. non-employer.
Santos, Betsy and DesRoches, David,Exploring Mode Effects in a Panel Survey of New Businesses(May 20, 2009). Available at SSRN: http://ssrn.com/abstract=1409361
comments powered by