Diary Surveys And Their Measurement Problems

By Author: Bunny Dom
Total Articles: 11
Comment this article

Diary surveys are generally self administered forms of topics difficult to study using other modes of data collection. The structured questionaire is replaced by a diary where the respondents enter information about the frequent events. such as consumption , television viewing. To avoid recall errors the respondents are asked to record information soon after the events have occurred. Often this means that repsondent must record information on the daily basis and even sometimes many times in a day. Thus for diaries surveys to be effective the respondent has to be very active and open to share the information.

Thus one of the major limitations of the diaries is the high committment required for the surveys. Hence the diary recording of the data has to be done for a shorter duration which is a maximum of 10 days.

Interstingly diary surveys are said to be free of the memory retention problems but practically a person is not able to remember all the incidents of the day. So there are always some errors in these parts.

The short recording period implies that the diary method cannot be ...
... used for estimating the occurrence of the rare events.

Another problem in diary surveys is the complexity of the diaries. Another problem is that the respondent may change their behaviour as a result of using the diaries.

Another problem is the insufficient attention given by the respondent to record the activity or purchase in the diary.

Data comparisons between a diary and an interview

The diary is designed to give a detailed accounting of the spending, including the smallest or most trivial expenditure. The lengthy reference period of the interview increases the odds that at least one purchase of the given items would be reported and offers a more comprehensive picture of the household spending. For larger purchases the memory is likely to be strong. If the interview is reported for the large salient purchases and the diary is needed for the frequent purchases, especially detailed data on food, the question becomes where to draw the line. This section examines the expenditure distribution statistics to access where the two modes would be capturing expenditures differently.

Detailed analysis of the distributions

Underreporting is the principal direction of the error for both components The source with higher overall mean is generally accepted as the better source. Variance comparisons usually favor the interview, since its panel design and longer reference period yields a much larger number of reports. The number of diary reports can become quite small for the detailed items. This leads to examining additional features of the expenditure distributions in order to study reporting characterstics by mode, withot undue influence of the sample size.

Two statistics to be considered are the reporting rate , percent respondents with non zero expenditure for a specific item, and the spending mean , the mean of the non zero part of the distribution. As allocated , the reports can cause drastically lower values for means and other statistics from the non zero distribution. This suggests examining the allocated and non allocated distributions seperately. In simplified form , we can write
Expenditure Mean = Spending Mean * Reporting Rate

Where these quantities are by week for the diary (each week seperately) and by month for the interview . This expression is based on the simple conditional expectation argument.

Meltdata Business Solutions. India