Alcohol and Parenting: The Effects of Maternal Heavy Drinking

4. Results

4.1 Family Characteristics of the Drinking Groups

Figure l reflects significant differences found when comparing the drinking groups on the survey's composite measure of socio economic status (SES) (F=137, df 3, p<.000). Comparison between groups revealed that the light drinking group had a significantly higher socio economic status level than the other three groups. This difference was hence considered in all remaining analyses where SES was used as a covariant. The need to consider socioeconomic status separately is further illustrated by the fact that there was a significant difference on a scale measuring home ownership, with the heavy drinking group significantly less likely to own their home than the other groups (x² = 142.71, df 3, p<.000).

Figure 1:

Similarly, the information concerning differential effects of intact and nonintact families on children (Depelteau et al., l998) resulted in a consideration of that variable. Specifically, an intact family was defined as the biological mother and father living together at the time the survey was completed and a nonintact family as any other type of arrangement. Thus, the four drinking groups were also divided into intact - non-intact groups. A highly significant result was obtained (X²= 262.89, df 3, p < .000). Figure 2 illustrates the increased percentage of nonintact children for heavy-drinking mothers. As this paper focuses on drinking behaviour, intact -nonintact status was treated as an independent variable in most subsequent analyses.

Figure 2:

The NLSCY composite variable "family functioning", plotted as family dysfunction, is thought to represent a global assessment of the quality of the family relationships. Measures which make up this score came from scales assessing the familial behaviour of problem solving, communication, roles, affective responsiveness, affective involvement and behavioural control. An analysis of this measure resulted in a drinking group effect (F = 7.28, df 3, p < .000) and alcohol by intact interaction (F = l4.04, df 3, p < .000). These results are illustrated in Figure 3. This figure demonstrates generally poorer family functioning in nonintact families except for heavy drinkers where intact families had poorer family functioning.

Figure 3:

4.2 Maternal Health

Given the well known relationships between certain health problems and drinking, significant results were expected. Generally these were not found, although, heavy drinkers were less likely to report that their general health was excellent than the other groups (x² = 91.66, df l2, p < .000). Importantly, heavy-drinking mothers reported a greater likelihood of bronchitis or emphysema (x² = 18.65, df 3, p < .000). This result is likely explainable by the fact that heavy-drinking women smoke cigarettes at a higher frequency than the other groups. This is illustrated in Figure 4 (x² = 763.78, df 6, p < .000).

Figure 4:

Numerous studies have shown maternal mental health, particularly depression, is related to child rearing and adjustment problems in offspring. Further, depression and drinking level in women is commonly correlated. Depression is the only aspect of maternal mental health assessed by the Survey. This was accomplished with l7 questions representing a reduced version of the CES-D scale (NLSCY, l995). Figure 5 shows depression scores by drinking levels for nonintact and intact families. An Analysis of Covariance revealed a very strong effect for intact- nonintact (F = l57.54, df l p < .000) and a trend for drinking (p < .04). Consequently, in most of the subsequent analyses depression was also run as a covariate.

Figure 5:

4.3 Maternal Ratings of Child's Behaviour, Ages 0 to 23 months

The survey contains questions dealing with frequency of early positive maternal interactions, as well as hostile and ineffective parenting. Regarding positive interactions, the heavy-drinking group reported fewer positive interactions than the other three groups. These results are illustrated in Figure 6. For the variable hostile ineffective parenting, there was a trend for alcohol group (p < .05), an effect for intact-nonintact (F = 7.4l, df l, p < .007), and an alcohol group by intact-nonintact interaction (F = 4.96, df 3, p < .002). Somewhat surprisingly, intact families in the moderate and the high alcohol consuming drinking groups displayed the greatest hostile/ineffective parenting.

Figure 6:

4.4 Maternal Rating of Child's Behaviour, Ages 2 to ll Years

Figure 7 illustrates mother's ratings of her child's emotionality/anxiety presented according to drinking level and family intact status. There was a significant difference for drinking groups (F = 7.65, df 3, p < .000). Individual comparisons between the drinking groups found that the heavy drinkers rated their children as significantly more emotional and anxious than non drinkers (F = 25.32, df l, p < .000). Examination of the figure reveals that this difference is primarily a result of strong negativity associated with nonintact status. A similar conclusion can be drawn regarding the mothers' ratings of their child's hyperactivity. These results are illustrated in Figure 8. Statistical analysis revealed a significant drinking group effect (F = 14.63, df 3, p < .000) and intact status effect (F = 12.04, df l, p < .001).

Figure 7:

Figure 8:

Individual comparisons showed that heavy and moderate drinkers differed significantly from non and light drinkers in their ratings. Mothers' ratings of their child's aggressivity also resulted in significant drinking group (F = 26.84, df 3, p < .000) and intact status (F = 24.26, df l, p < .000) findings. Individual comparisons revealed heavy drinkers differed from non drinkers (F = 25.53, df l, p < .000), light drinkers (F = 9.86, df l, p < .002) and moderate drinkers (F = 6.88, df l, p < .009) in rating these children as more aggressive. These results are illustrated in Figure 9.

Figure 9:

Consistent with the findings of greater emotionality and anxiety, hyperactivity and aggressivity in children ages 2 to ll, mothers from nonintact families and heavy-drinking mothers saw themselves as engaging in fewer positive maternal interactions with their children and more hostile-ineffective parenting. These results are illustrated in Figures l0 and l1.

Figure 10:

Figure 11:

For positive interactions, the drinking group difference was F=9.09, df 3, p<.000, and there was also a trend for a drinking group by intact interaction (p < .04). Individual comparisons determined that the heavy-drinking group displayed significantly fewer positive interactions than the other groups (F = 47.l3, df l, p < .000). Illustrated in Figure l0 is the fact that it was intact heavy-drinking mothers who showed the fewest positive interactions. For the variable hostile-ineffective parenting, the results were, for drinking group, F = 48.29, d 3, p < .000; for intact status, F = l0.85, df l, p < .001; and for drinking group by intact status, F = 3.52, df 3, p < .0l4. As illustrated in the figure, most of the differences were accounted for by the heavy-drinking, nonintact mothers. In terms of individual comparisons the heavy-drinking group was significantly different than the non drinking group (F = l45.25, df l, p < .000), the light drinking group (F = l6.08, df l, p < .000), and the moderate drinking group (F = l7.99, df l, p < .000).

Separation anxiety in their children was also rated by mothers of 2-3 year olds. Analyses revealed a significant drinking group effect (F = 7.29, df 3, p < .000) with the children of mothers in the heavy-drinking group rated as displaying the most separation anxiety.

4.5 Additional Maternal Ratings and Ratings from Ages 4 to ll

The survey contains some questions that began being rated at age 4. These included measures of indirect aggression and aspects of conduct disorder. Indirect aggression was assessed from the questionnaire developed by Lagerspetz, Björkqvist & Peltonen in Finland and aspects of conduct disorder were derived from a composite measure taken from the Ontario Child Health Study (See NLSCY Project Team, l995). With SES and age covaried, the drinking group effect for indirect aggression was significant (F = l7.55, df 3, p < .000) as was the conduct disorder measure (F = 34.32, df 3, p < .000). Individual comparisons for indirect aggression showed that the heavy-drinking group was significantly different from the non drinkers (F = 55.02, df l, p < .000) and light drinkers (F = 58.54, df l, p < .000). Similar comparisons for conduct disorder showed the children of the heavy-drinking group differed significantly from the non drinkers (F = 25.37, df l, p < .000) and moderate drinkers (F = 34.58, df l, p < .000). Gender effects were also found on these variables (indirect aggression F=71.67, df 1, p<.001; conduct disorder F=41.64, df 1, p<.000). In both instances, the maternal drinking effect held for both sexes, although boys were more affected for conduct disorder and girls for indirect aggression. Figure 12 illustrates a combined direct and indirect aggression score by drinking group, intact-nonintact status and gender. Figure 13 is an analysis of property crimes by these same variables. For direct and indirect aggression, there were significant effects for drinking groups (F=15.68, df 1, p<.000), intact status (F=22.51, df 1, p<.000), and sex (F=81.81, df 1, p<.000). As shown in Figure 12, boys were more affected by familial non-intact status. For the variable of property offences, there was a drinking (F=24.37, df 3, p<.000), intact status (F=22.54, df 1, p<.000), and gender effect (F=75.13, df 1, p<.000). Clearly, males of moderate and heavy-drinking mothers are reported to commit more of these offences.

Figure 12:

Figure 13:

4.6 Teacher Ratings

Relevant to the mothers' ratings, data are also available from teachers. In particular, the survey contains questions that deal with parental involvement with the child in regard to school and the child's behaviour. Concerning involvement with the school, the four drinking groups were assessed on three variables. These were: the teachers' ratings of whether parents discussed the student's performance, the teacher's belief on how important the parents evaluate the school, and the degree of parental support demonstrated to the teacher. Respective chi square differences between the drinking groups for these variables were x²= 38.93, df 3, p < .000, x²= 53.34, df 6, p < .000, x²= 33.48, df 6, p < .000. On each of these three questions it was the heavy-drinking parents who were seen as less involved and supportive. Teachers also rated the child's behaviour for conduct disorder/aggression, hyperactivity, prosociality and degree of emotionality. Teacher's rating of conduct disorder/aggression for the maternal drinking groups is illustrated in Figure l4 and for child hyperactivity in Figure l5. For conduct disorder/aggression, there was a significant effect for drinking group (F = 4.27, df 3, p < .005), and for intact status (F = 46.10, df l, p < .000); and, for hyperactivity, there was a drinking group (F = 3.73, df 3, p < .01) and intact-nonintact status (F = 19.15, df l, p < .000) effect. Individual comparisons revealed that for conduct disorder/physical aggression, heavy drinkers were significantly different from non drinkers (F = 12.60, df l, l < .000) and light drinkers (F = 4.46, df l, p < .03).

Figure 14:

Figure 15:

The importance of intact/nonintact status was reflected in teachers' ratings for prosociality and presence of emotionality. On both of these variables only the intact status variable was significant (respectively, F = ll.57, df l, p < .00l and F = 44.93, df l, p < .000). We constructed a composite variable for teachers' ratings of aggressivity, hyperactivity, emotionality and negative prosociality as a putative measure of deviancy. We considered those scores over the 90th percentile, on two or more of these scales. When the drinking groups were thus analysed, there was a clear maternal drinking effect ( x²= 18.19, df 6, p < .006). A similar but stronger finding occurs when the same variables are analyzed in a similar fashion for mothers' ratings (x²=41.52, df 6, p < .000). Thus there is some degree of continuity between maternal and teacher ratings.

4.7 Child's Self Report and Test Scores

A self-report questionnaire was filled out by children between the ages of l0 and ll. The behavioural aspects of this scale measured such things as conduct disorder/physical aggression, emotional disorder, indirect aggression, hyperactivity/inattention, prosocial behaviour, and difficult behaviour. When these variables were analyzed by maternal drinking level and intact and nonintact status, none of the findings reflected a maternal drinking effect, yet all yielded a significant intact-nonintact finding with one exception (prosociality). Clearly, l0 and ll year old children from nonintact families see themselves as troubled and troublemaking. The results were for conduct disorder/physical aggression (F = ll.37, df l, p < .00l), indirect aggression (F = 5.80, df l, p < .0l6), emotional disorder (F = l0.02, df l, p < .002), hyperactivity/inattention (F = ll.26, df l, p < .00l), and for difficult behaviour (F = 5.7l, df l, p < .02). Children between the ages of 4-5 also completed the Peabody Picture Vocabulary Test (PPVT) and those in grade 2 and higher a Mathematics Achievement Test. The academic achievement results are illustrated in Figure l6. For the PPVT the only significance was for the drinking variable (F = 10.51, df 3, p < .000), with the children of heavy drinkers scoring significantly higher than those of non drinkers (F = 16.44, df l, p < .000) and light drinkers (F=10.40, df 1, p<.001). There were no significant differences found for the Math test data.

Individual self-report items did however display a pattern of responses for the l0-l1 year old offspring of heavy-drinking mothers. These individuals worried more about their parents divorcing (X² = 26.47, df 3, p < .000), were more likely to play video games (X² = 7l.06, df 9, p < .000), had the most trouble getting along with other kids (X² = 79.28, df 9, p < .000), were more likely to have drunk alcohol (X = 47.68, df 3, p < .000), and were most likely to have been questioned by the police (X² = l4.8l, df 3, p < .002).

Figure 16:

Contents

Alcohol and Parenting: The Effects of Maternal Heavy Drinking - October 1998