Wave 3.5 data collection and response

Fieldwork

As two years is a long time between main wave home visits, mid-wave data collections are a useful way of keeping in touch with study families and collecting information about selected topics in more detail. It is also an important opportunity to collect information about key changes in children’s lives. For example, in 2009 children from the B cohort were aged 5 to 6 years and most had recently made the critical transition into the first year of primary school. Meanwhile, the older K cohort children were aged 9 to 10 years old and some were starting to go through puberty.

To explore some of these transitions, in early June 2009 the study sent out an eight page ‘mid-wave’ (Wave 3.5) mail-out to all participant families. The questionnaire developed for each cohort asked about a range of topics including schooling, child health, media and technology, and stressful life events for the parents.

Parents of both cohorts were asked about topics such as their children’s experiences with school, their satisfaction with their child’s educational progress, and children’s health, including sleep patterns and injuries requiring medical attention. Questions were also asked about the child’s media and technology use, including parents’ concerns with their use of certain technology and rules set by the parents on what media and technology children were exposed to. Parents in both cohorts were also asked to report whether they had experienced any stressful life events in the last 12 months.

Parents of B cohort children were asked further questions on whether or not their child likes going to school and their own and their child’s relationship with their teacher and school. Questions also covered how the child behaves with other children.

For the older K cohort children, parents were asked about signs that would indicate the onset of puberty, what activities children use a computer for, whether the child has a mobile phone and the different technologies study children use to communicate with friends.

Response rates and sample characteristics

In Wave 3.5, 9,413 questionnaires were sent out to study families: 4,772 B cohort families and 4,641 K cohort families (see Table 1). Sixty-four per cent of the Wave 3.5 sample returned the questionnaires. There was a very similar return rate between the B cohort (63 per cent) and K cohort (64 per cent), although the overall response rate was slightly lower than Wave 2.5 (B cohort=69 per cent and K cohort=67 per cent).

Table 1 shows the response rates for the main and mid-wave data collections. There has been a decline in response rates for both the main and the between-wave data collection over time. For the main waves, the response rates have been high and did not decline much between Waves 2 and 3 (91 per cent to 89 per cent, or 2 percentage points). However, the mid-wave response rates were substantially lower than for the main Wave: 72 per cent in Wave 1.5 and declining by about 4 percentage points for each mid-wave survey.

Note: Available sample excludes those who opted out of the study between waves and children who had died. It includes all other study respondents, including those with out-of-date contact details.

Table 2 illustrates the demographic characteristics of the responding sample at each wave and mid-wave from Wave 1 to the mid Wave 3.5. Certain demographic characteristics of the sample (such as sex, state and region) have remained fairly stable over time and when comparing between mid and main wave.

For other characteristics there appear to be some important differences between waves. For example, the proportion of the sample where the mother has completed Year 12 shows two interesting and distinct patterns.

First, it appears that the proportion of the sample in which the mother completed Year 12 is smaller for main waves (B cohort: 67 per cent, 69 per cent, and 70 per cent for Waves 1, 2 and 3 respectively) compared to the mid-waves (for the B cohort: 74 per cent, 75 per cent, and 76 per cent for Waves 1.5, 2.5 and 3.5 respectively).

This suggests that mothers with a lower level of education are less likely to respond to the written, mail-out mid-wave survey compared to the face-to-face main wave interview. The data also show a second, related pattern: that mothers who completed Year 12 are an increasing proportion of the sample over time.

These two patterns can also be seen in both cohorts and in relation to the ethnicity variables and in families in which no parent works. This is consistent with research from other international social surveys,1 showing that people from lower socioeconomic backgrounds, people with a lower education level and people from culturally and linguistically diverse backgrounds are at a higher risk of non-response, particularly for written surveys.

The LSAC management group is very mindful of these differences and is continually developing strategies to support all families to stay in the study.

For data users, main wave data includes weighting variables to allow researchers to take into account a range of these demographic differences in response rates when making estimates about the Australian population.

A number of other patterns reflect some of the changes happening as study children grow older. For example, the number of only-child families in the B cohort decreased over time from 40 per cent in Wave 1 to 19 per cent in Wave 2 and 10 per cent in Wave 3.5. This decrease in only child families mirrors the increase in families where the study child had one or more siblings. The corresponding pattern in the older K cohort study families is also apparent, but much less pronounced. The proportion of K cohort study families with only one child is much lower at Wave 1 (12 per cent) and declines only gradually by Wave 3.5 (8 per cent). These changes in the sibling composition of study families reflect how parents in the B cohort are particularly likely to have had another child over the first three waves of the study, while the K cohort composition data suggests, as would be expected, that as children get older, parents are less likely to have more children.

Note: Where information on a particular sample characteristic was available from the mid-wave survey, all families who responded to the main-wave survey were included in the data for that characteristic. However, for characteristics for which data were not available at the mid-wave survey (denoted by *), data was provided from the previous main wave. For these characteristics, Wave 3.5 responding families were not included if they did not respond to Wave 3. Percentages may not total 100 per cent due to rounding.

1. Stoop, I, Billiet, J, Koch, A & Fitzgerald, R 2010, Improving survey response: lessons learned from the European Social Survey, Wiley, UK.