Attrition Bias

Attrition is generally only a serious concern when it is non-random (that is, when the persons that attrit from the panel have characteristics that are systematically different from those who remain).

Table 61 provides figures on the proportion of Wave 1 respondents who were re-interviewed in Wave 6 disaggregated by various sample characteristics. For those persons interested in the balanced panel, the proportion of Wave 1 respondents who have been interviewed in every wave is also provided. People who have died or moved overseas are excluded from these figures. These results indicate that the re-interview rate is lowest among people with following Wave 1 characteristics:

• living in Sydney and Melbourne;
• aged 15 to 24 years;
• single or living in a de facto marriage;
• born in a non–English–speaking country;
• Aboriginal Torres Strait Islander;
• living in a flat, unit or apartment;
• relatively low levels of education;
• unemployed; or
• working in blue–collar or low–skilled occupations

The variance in attrition over the five waves is particularly marked with respect to age, country of birth, labour force and occupation.

The disparity in the re–interview rates for wave 1 respondents re–interviewed in wave 6 across the different characteristics is not as great as for those interviewed every wave. The most striking example of this is indigenous status – while only 51.8 per cent of indigenous respondents in wave 1 have been re–interviewed every wave, 70.8 per cent were re–interviewed in wave 6. This indicates that the groups with low re–interview rates in the balanced panel are still engaged with the study.

Overall, attrition is not random. While we can make adjustments for the attrition through the sample weights, these adjustments are only as good as our ability to measure differential attrition.

The attrition rates are discussed at length in Watson and Wooden (2004c) and Watson and Wooden (2006).

Table 61: Percentage of Wave 1 Respondents Re–interviewed by Selected Sample Characteristics