Evaluation of Questionnaire Design Changes
on Life Insurance Policy Data
Alfred O. Gottschalck and Jeffrey C. Moore
KEY WORDS: cash value, face value, nonresponse, response error, rounding, SIPP, wealth
ABSTRACT
In response to user needs, the Census Bureau changed the way life insurance data are collected in the 2004 Survey of Income and Program Participation (SIPP) wave 3 Wealth Topical Module. Instead of asking for the face value of life insurance policies, the questionnaire was revised – by changing one word – to collect information on the cash value of those policies. This paper describes research to evaluate the impact of this change on the resulting data. Major findings are as follows: (1) The new cash-value question is imperfectly understood, with perhaps as many as two-thirds of life insurance policyholders continuing to report the face value of their policies, despite the revised intent of the question. Contributing problems include misunderstanding of policy types and their features, and misperceived question intent. (2) As expected, the revised question elicited a substantial increase in item nonresponse, reflecting the less salient nature of the cash value concept. (3) Also as expected, median and mean values decreased, although when disaggregated by policy type, the median value for term policies (excluding respondents who correctly reported a zero dollar cash value amount) remained unchanged, again indicating a problem in reporting the desired new concept. (4) The degree to which reported amounts are “rounded” also decreased, particularly in the case of whole-life policies. Severe rounding of reported amounts was still very common, however. The paper offers a set of proposed solutions to the problems observed.
CITATION:
