Page 60 - Zero for nine: Reducing alcohol use during pregnancy via health counselling and Internet-based computer-tailored feedback

Chapter 3
the variable of partner support to abstain from alcohol was carefully examined. With MPlus 6.11, a latent class analysis (Collins & Lanza, 2010) was conducted to make a well-founded, data-driven decision on the number of groups that were most parsimonious and descriptive of the data. Several statistical indicators were used, including the Bayesian Information Criterion (BIC), the Lo-Mendell-Rubin likelihood ratio test (LMR-LRT), and entropy (Nagin, 2005). The LMR-LRT showed a significantly better fit with two groups than with one, but not a significantly better fit with three groups than with two, supported by the other indicators. The group low in partner support consisted of participants who had scored 1, 2 or 3 on the 5-point scale; the group high in partner support consisted of participants who had scored 4 or 5. Statistical differences between respondents low and high in partner support to abstain from alcohol during pregnancy with regard to predisposing, information, awareness, motivational factors and intention were analysed using chi-square test for dichotomous data, the chi-square test for trend for ordinal data and the t-test for interval data, as appropriate. We followed Norman (2010) in reporting the results of t-tests to compare the 5-point Likert scale variables between the two groups. Rerunning the analyses with chi-square tests for trend revealed that the results did not differ noteworthy from the results of the t-tests.
Results
Sample characteristics
The total sample consisted of 237 partners. The age of respondents varied from 20 to 54 years, with a mean of 33.35 (SD = 5.59; see Table 3.1). Half of the sample reported a high educational level, the majority had an income more than twice the net Dutch medium household income (€ 1700 per month) or higher. Respondents reported a mean partner support of 4.14 (SD = 1.03), indicating that on average respondents greatly supported their pregnant partner to abstain from alcohol. According to the latent class analysis, respondents could be divided in one group low in partner
58