Grant A. Gibson

O Brother How Art Thou: Propensity to Report Self-Assessed Unmet Need

 

We use surveys to investigate lots of policy-relevant elements of government service delivery. The potential for response bias is usually acknowledged, but, since there is almost never any way to verify how it might impact conclusions made from the survey data, quickly forgotten. This paper conducts two exercises to verify whether there is significant response bias in a question commonly used to investigate equality of access to healthcare. The first exercise relies on a set of conditions used to evaluate the quality of ambulatory care within a system and a unique data linkage between the Canadian Community Health Survey and the Discharge Abstracts from hospital stays. The second uses the 'more objective' information contained within the survey itself to understand the patterns in response bias. Reassuringly, the two methods are in broad alignment and suggest a) the need to use conditional correlation when examining inequality in an 'unmet needs' framework and b) the value of reviewing the survey question to reduce response bias.

​

​