1285.0 - Demographic Variables, 1999  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 11/08/1999   
   Page tools: Print Print Page Print all pages in this productPrint All  
Contents >> Age >> Measurement issues and related classifications

Measurement issues

46. The main statistical consideration in respect to age is the number and range of groups used to output the data.

47. Although there are no strict statistical rules, there are some common principles involved in setting the number and range of age groups.These principles are as follows:

      • for statistical applications the maximum number of groups should be related to the sample size and the reliability of the data. For confidentiality and reliability reasons a minimum number of responses are required in any category or cell before data can be published. There is little merit in producing numerous cells with high sampling errors or cells with suppressed data;
      • the number of categories should be manageable and meaningful. The number of groups used should be no more than required to satisfy the intended purpose;
      • the groupings should not distort the data. Inappropriate groupings can give the impression of an even spread of ages within a group when this is not the case. For example, a 10 year grouping 10-19 years could give the impression that the data is evenly spread between 10 and 19 years when most of the data may be clustered between 10-14 years or 15-19 years.

48. Two other important considerations are:
      • collection of age data using age groupings should be avoided;
      • age data should always be stored at the lowest 'time unit' so that aggregation can be made to single, five and ten year age groups if required.

49. Collecting actual date of birth gives the best data quality. Collecting age in complete years can lead to the following type of error where a respondent may round off or approximate their age in complete years. For example, a respondent born in June 1963 may respond to a survey in March 1985 that he/she is 22 years old when the more accurate 21 years could be obtained from date of birth.


Related classifications

50. There are no related classifications applicable to this variable.



Previous PageNext Page