Leung et al #
problem score (the extent to which the parents find the behaviour troublesome). A validated Chinese version was available.18 The reliability (Cronbach alpha) of the former was 0.95 and of the latter was 0.94.
Parenting Stress Index (PSI13)—this was a 36- item questionnaire with three subscales, namel , parental distress (PD), parent-child dysfunctional interaction (PCDI), and difficult child (DC). The Chinese version was validated.19 The reliability estimates for the total score and the three subscales were 0.92, 0.86, 0.82, and
, respectively. In this stud , only the first two subscales were used, as there was considerable overlap between the DC subscale and the ECBI.
Relationship Quality Index (RQI20)—this was a six-item index of marital or relationship quality and satisfaction. The questionnaire has been used with Chinese parents in Hong Kong with satisfactory reliability estimates.21
Demographic information and access to parenting education.
A semi-structured interview guide with questions on adaptation, child well-being, and service needs was used in the individual interviews (Appendix).
For quantitative data, preliminary comparison between PMR and non-PMR participants was made using the independent t test. Multiple regression was used to examine differences between PMR and non-PMR participants for child behaviour problems, parenting stress, and marital satisfaction, while controlling for socio-economic variables.
The qualitative data were analysed by the second author using the constant comparative method. The third author independently analysed 12 randomly selected transcripts. The inter-rater agreement was around 90%.
Results The sample
There were 261 PMR participants with complete data. To gauge the representativeness of the PMR sample, we compared the sample with the 2001 census. The present sample was similar to the census in terms of maternal age, age difference between spouses, family type, and mother’s education. However, in the present sample, there was a higher proportion of boys (61%); in the 2001 census, 52% were boys. A higher proportion of households had monthly incomes of $19 999 or below (93% compared to 73% in the 2001
Hong Kong Med J Vol 13 No 5 # October 2007 # www.hkmj.org
census). Furthermore, a lower proportion of spouses of PMR mothers were not in the work force (10%, compared to 15% in the 2001 census), and there was a lower proportion of working PMR mothers (13%, compared to 26% in the 2001 census). The latter was probably related to data collection procedures in the present stud , whereby the researchers administered the questionnaires to the participants in person during office hours. The present PMR sample was thus more economically disadvantaged than those in the general population.
For the non-PMR sample, of the 942 participants with complete data, 753 had resided in Hong Kong for 7 years or more and were used as the comparison group.
The socio-economic characteristics of the PMR and non-PMR participants are shown in Table 1. Consistently the PMR sample was more socially disadvantaged than the non-PMR sample.
Differences in child behaviour problem, parenting stress, and marital satisfaction between PMR and non-PMR participants
Independent t tests indicated that the two samples differed significantly in terms of child behaviour problems (ECBI-problem) and parenting stress (PSI- PD and PSI-PCDI). The mean and 95% confidence interval scores, as well as the reliability estimates of the ECBI, PSI, and RQI scores of the two samples are shown in Table 2. However, the two samples differed significantly in a number of socio-economic characteristics known to be related to parenting stress, child behaviour, and marital relationship.12,16 Five multiple regression analyses were performed to examine whether PMR status was associated with ECBI-intensit , ECBI-problem, PSI-PD, PSI-PCDI, and RQI, after controlling for a number of socio-economic variables. These were relationships of the participant and the target child (mother as participant vs other), parent’s and child’s age, father’s and child’s length of residence in Hong Kong, child’s gender, availability of other carers (available vs non-available), parent’s educational attainment (≤lower secondary vs ≥upper secondary), household income (≤HK$19 999 vs ≥HK$20 000), parent’s work status, access to parenting education (frequently/sometimes vs never), and Comprehensive Social Security Assistance recipient status. Mother’s length of residence in Hong Kong was excluded because of its high correlation with PMR status (r=0.82, P<0.001, n=1009). Including variables with bivariate correlations of 0.70 or above in the same analysis would have created the problem of multicollinearity.22 After controlling for the aforementioned socio-economic variables as well as access to parenting education, only ECBI-problem and PSI-PCDI were associated with PMR status (Table 3).