Income reporting in the initial waves of panel surveys: evidence from the Understanding Society

Publication type

Conference Paper


Paul Fisher

Publication date


This paper provides evidence that income data collected as part of panel
surveys may not be comparable across waves. Using Understanding Society
and exploiting a unique feature of the survey design – that random
samples of households are responding at different waves of the panel in a
given calendar year – means that we have a quasi-experimental setup.
Estimates indicate that the effect of being interviewed for a second
time is to increase reported monthly benefit income by £117 per month.
Dependent interviewing – a common recall device used in household panel
surveys – takes effect only after a first survey interview. It can
explain approximately 37% of the observed increase in reported income,
however, leaving a substantial part of the increase unexplained. After
ruling out alternatives, we conclude that familiarity with the survey
and interview process changes reporting behaviour at the second wave.
This conclusion is supported with an examination of item non-response
rates for the income variables, which fall at the second wave. The
results have implications for estimating income distributions from the
initial waves of panel surveys and more broadly for comparability with
income data from cross-sectional surveys.


Survey Methodology and Income Dynamics