A History of the Ministry of Information, 1939-46



22 23

Note on Memory

In the preliminary discussions with the Board of Trade, it was considered by the Wartime Social Survey that the annual breakage rate arrived at by the Board of Trade's earlier survey was an under-estimate.

It was further considered that the reason for this was that the Board of Trade had asked housewives “How many of each item did you break last year?”. In the view of the Survey, the very considerable volume of work already done on memory made it seem unlikely that housewives could remember at all accurately for such a long period. It was the Survey's view that such a result would in fact be an estimate based on. the housewives recollection of recent breakages and would be a very considerable under-estimate.

In order to test the hypothesis that housewives are unable to remember such breakages and that as time passes they forget more, it was decided to ask three questions about breakages which would be exactly comparable. They were:-

(1) How many of the following articles did you break last week?

(2) How many of the following articles did you break last month?

(3) How many of the following articles did you break in the last three months?

From these rates it was possible to calculate either a weekly or an annual rate, and if it was found that these rates were lower when calculated from the monthly or quarterly estimate, it could be considered that this was due to the inability of housewives to remember for such a long time.

It was, however, considered that if the relationship between the three rates thus derived was constant for a number of articles, some of which were more scarce than others, and some of which were more permanent than others, this would establish a relationship between forgetting and time. To demonstrate this point the results were plotted on logarithmic graph paper.

It was, however, decided to plot on the same graph the breakage rates obtained by the Board of Trade as the result of their previous inquiry. These rates in many cases bore a constant relationship to the rates obtained in our inquiry, but in one or two cases the relationship is different. In this connection, we should mention that whereas the Wartime Social Survey's sample was a completely random one within the areas chosen and a purposive one insofar as it was devised to represent in approximate proportion the main areas of urban and rural population. The previous survey represented three large urban centres - London, Manchester and Newcastle (evacuation areas); and four smaller urban centres - Taunton, Llandudno, Hereford and Hexham (reception areas), and the sample was what was believed to be a selective one designed to give class representation according to an estimate of the class composition of the country.

The main points of our hypothesis were demonstrated in the pilot enquiry and showed the following breakage rates for cups and saucers, based. on a sample of 80:-

QUESTION Annual Rate
Cups Saucers
Last Week 21.5 11.0
Last Month 11.7 7.5
Last 3 Months 9.8 4.3

An analysis of the first 500 questionnaires of the survey proper bore out this main tendency, the annual breakage rate for cups, for example, based on the weekly figure, being 21.2.

The results of the final survey are given in the tabular statement below, and in every case the relationships between the three results obtained by the Wartime Social Survey's inquiry are constant.

+ No. Broken Last Week Annual Average + No. Broken Last Month Annual Average + No. Broken Last 3 Months Annual Average No. Broken Last Year Bd. Of Trade
% * % * % * %
Cups 20.8 ± 10.35 16.1 ± 5.7 12.4 ± 3.9 7.6
Saucers 8.7 ± 15.75 6.6 ± 8.4 5.3 ± 5.7 4.1
Dinner Plates 6.9 ± 17.91 6.1 ± 9.0 4.6 ± 6.0 2.5
Soup Plates 1.9 ± 34.2 1.2 ± 20.1 .76 ± 15.0 .9
Serving Dishes 1.7 ± 35.1 .87 ± 24.0 .5 ± 18.3 .3
Teapots 2.0 ± 33.3 1.8 ± 16.5 1.3 ± 11.4 .5
Tumblers 5.7 ± 19.8 4.3 ± 10.8 3.4 ± 6.9 3.2

+ Calculated by multiplying by appropriate number, i.e. 52, 12, 4, and dividing by number in sample, 2087.

* Margin of error estimated at three standard deviations and calculated as percentage of average number. This means that 10.35% of 20.8 cups per year is the margin of error (= ± 2.07 cups per year).

The particular items were chosen because comparable Board of Trade figures were available and in order to include both fragile and more durable articles. These results have been plotted on the appended diagram.

It was realised that these results giving an annual breakage rate for cups of at least three times the original estimate of the Board of Trade would no doubt to have a very important effect on policy. It was therefore considered appropriate to consult the Survey's scientific advisers as to the validity of the rates derived in various ways.

The Survey's advisors were unanimous in considering that the best estimate was that based on the weekly figure, and they considered that an estimate based on the yearly figure would not be at all reliable.

We should add that one of our advisers thought that the estimate based on the weekly figure might possibly be a slight over-estimate insofar as housewives might tend to compress the experience of the last ten to twelve days in their recollection experimental evidence about remembering, which shows a loss over even very short periods such as a week; in any case, another factor enters the discussion here, which is that in many cases housewives who were interviewed on, say, a Thursday, would not remember beyond the week-end. This was found to be the case both by field workers and by the Senior Research Officer in charge of the survey, who conducted many interviews in the field.

A comparison of annual breakage rates derived from questions asking the number broken in one week, one month, three months and one year (Board of Trade) plotted against the time the breakages were remembered.

We use cookies to track usage and preferences.

Privacy & Cookie Policy Accept & Close