This paper uses scanner data to generate estimates of quality-adjusted pric
e changes for video-recorders. We use hedonic regressions, to derives estim
ates of the changing worth of each quality component. These are then applie
d to weighted changes in the mix of quality attributes of products to deriv
e estimates of quality-adjusted price (QAP) changes. The data source used i
s electronic-point-of-sale (EPOS) scanner data that are available for a wid
e range of goods. This study provides an example of how such methods can he
more widely applied The estimates of QAP changes correspond to constant-ut
ility, 'headonic ' cost-of-living indexes defined in economic theory us the
ratio of expenditure functions at constant utility allowing for changing p
rices and characteristics of goods. This method is proposed as an improveme
nt on the existing direct method, which takes its estimates directly from t
he coefficients associated with 'time dummies' in a hedonic regression. We
finally undertake a matching process, akin to that used by statistical offi
ces, and compare the results. Direct comparisons with RPI estimates and the
se hedonic approaches are not easy since the approaches use quite different
data sets. Our replication of a procedure akin to that used for the RPI on
the scanner data set provides insights into sources of potential bias.