Appendix B
Reported beer consumption in NCDS and BCS70 2000 survey
A histogram and frequency table is displayed below to show the data for men on
reported beer consumption from the 1958 cohort in the 2000 survey. It can be seen
that although the data is labelled as the number of pints of beer drunk in the last
seven days there is a tendency for even integers (2,4,6,8,10,12 etc) to appear much
more commonly than odd integers (2002 men are recorded as having reported
drinking an even number of pints of beer between 1 and 14 pints and only 835 men
are recorded has having reported drinking an odd number of pints of beer between 1
and 14 pints). This is probably indicative of a proportion of interviewers
misunderstanding the CAPI instructions, which were somewhat ambiguous, and
rounding consumption to the nearest pint but then recording the data in units of
alcohol and not, as intended, in pints of beer. For purposes of comparison the
histogram and frequency table for reported beer drinking at age 33 in 1991 is also
reproduced below.
Any problem with the recording of beer consumption in the 2000 NCDS survey is also
likely to be replicated in the 2000 BCS70 survey as the CAPI protocols were identical
for the questions on alcohol consumption. The frequency tables for reported beer
consumption in the previous week for the 2000 BCS70 survey and the 2004 BCS70
survey are therefore also shown below. These also demonstrate that there appears to
have been a mis-recording of units of beer instead of pints of beer in the 2000 survey.
For example in the 2000 survey 292 (6.1%) men report drinking eight pints of beer
whereas only 101 (2.1%) report drinking seven pints and 47 (1.0%) report drinking
nine pints.
Histogram
Men’s reported beer consumption, 1958 cohort age 42, Survey 2000.
27