Bradburn, Norman M.; Rips, Lance J. and Shevell, Steven K. "Answering Autobiographical
Questions: The Impact of Memory and Inference on Surveys," Science 236 (10 April 1987), pp.
"Survey questions often probe respondents for quantitative facts about events in their past:
‘During the last 2 weeks, on days when you drank liquor, about how many drinks did you have?'
‘During the past 12 months, how many visits did you make to a dentist' ‘When did you last work
at a full-time job?' are all examples from national surveys. Although questions like these make
an implicit demand to remember and enumerate specific autobiographical episodes, respondents
frequently have trouble complying because of limits on their ability to recall. In these situations,
respondents resort to inferences that use partial information from memory to construct a numeric
answer. Results from cognitive psychology can be useful in understanding and investigating
these phenomena. In particular, cognitive research can help in identifying situations that inhibit
or facilitate recall and can reveal inferences that affect the accuracy of respondents' answers."
From the title, I thought the authors might be estimating the error terms obtained on surveys.
They do not. They discuss the processes which the respondent uses in deriving a number, an
answer to the question asked. They are being nice when they suggest memory is employed. I
could be nasty and suggest that the answer reflects the respondents' image maintenance.
Respondents use various strategies in deriving estimates, but one cannot, thereby, say that they
all use the same strategies and make the same sorts of errors in calculating a number. Their errors
do not cancel themselves out. However, according to these authors, there is still a legitimate
science of surveys because the processes used were similar.
There is an up-to-date bibliography provided here.