Gender and family differences in adolescent’s heavy alcohol use: the power-control theory perspective

Abstract

According to the power-control theory, growing independence of adolescent girls, manifest in more prevalent problem behaviors, may be explained by changes in family structure (increasing level of authority gained in the workplace by mothers). To verify this hypothesis, self-report data from Warsaw adolescents (N = 3087, age 14–15 years, 50% boys) were used. Results indicate that parenting practices differ across child gender and structure of parents’ work authority. Girls, especially in patriarchal households, spend more time with mothers and perceive stronger maternal control. In egalitarian families, fathers tend to be more involved with sons than with daughters. When parental control, support and adolescents’ risk preferences are controlled, the gender-by-household type interaction effect is observed—girls in patriarchal families have the lowest risk of getting drunk. Study results provide support for power-control theory showing the relationship between parental work authority and adolescent’s heavy alcohol use.

Introduction

Problem behaviors, delinquent acts and crime are traditionally considered a bigger issue for males than females. However, there has been a recent increase in research interest regarding the aggressiveness and delinquency of female adolescents [1–7]. This interest is caused by growing rates of young females’ involvement in problem behaviors noted in epidemiological studies. Increasing female adolescent’s alcohol use has been found in European Union (EU) countries, particularly those belonging to the ‘old Union’ (who were EU members in the 1990s; [8, 9]), as well as in the United States [10] and Taiwan [11]. Delinquent behaviors are more and more frequently observed not only in Europe but in Asia and the United States, too [12, 13].

In Poland, a similar tendency has been observed in local and national studies. In a Warsaw survey, covering >20 years, the increasing prevalence of alcohol use for boys was observed from the mid 1980s to the mid 1990s, but in the 2000s prevalence began to decline [14]. Conversely, girls’ alcohol use has increased consistently since the mid 1980s [15]. Until 2000, boys consumed large amounts of alcohol much more frequently than girls. However, in 2004 frequent abuse of alcohol was reported by as many girls as boys [14].

Observations from this local study are consistent with other Polish surveys which placed changes in alcohol consumption in a context of other problem behaviors [16, 17]. Today, adolescent girls in Poland use alcohol, smoke cigarettes and try illicit drugs at least as frequently as boys.

Negative trends among girls are explained mainly as a consequence of earlier females’ pubertal development. Girls mature earlier than boys, so they try to become independent from parents and to foster tighter bonds with peers earlier than boys [9]. Moreover, the age of onset of menstruation has decreased in all industrialized countries during the last century. Biological maturity has a strong effect on alcohol drinking by girls and their tendency to associate with older boys and leads to increased exposure to substance use [18, 19].

The increasing exposure to psychoactive substances among girls may also be due to gender differences in ways of spending free time. In Poland, girls spend more time outside of the home than boys and this may be a risk factor for alcohol use. Correspondingly, more boys than girls play computer games and have computing as a hobby [17]. For boys, this way of spending free time probably limits their direct contact with peers and thus limits the opportunities and occasions for drinking alcohol.

The power-control theory of delinquency of Hagan et al. [20, 21] also provides a useful conceptual framework for understanding gender differences in alcohol use. This theory links differences in socialization of boys and girls with the power relationships between parents in the home, which are extensions of power relationships of males and females in the labor force. Power structures between parents generate different kinds of households. In traditional patriarchal families, fathers typically have more authority in the workforce than mothers and this gets expressed in the home. This idea is supported by studies in a literature review by McHale et al. [22], which revealed that fathers’ values and preferences better explain children's gender development than mothers’. In families where there is a balance in the degree of power held by husbands and wives or wives’ work authority is higher than husbands’, the household structure may be described as less patriarchal or more balanced. In the home, it is expressed by more paternal involvement into daily family life than in traditional patriarchal families [22].

Traditionally, parental control is stronger for daughters than for sons [23–25]. Moreover, maternal control is usually stronger than paternal control [25–27]. In this connection, in power-control theory, mothers are identified as the primary instruments of control and daughters as the primary objects of control [21].

The level of control is just one of the expressions of traditional gender-based socialization. The others, well documented in the literature, concern encouragement of sex-typed activities, guiding offspring toward particular academic or career choices [22], taking on housework by daughters and sons [28, 29], the tendency to place more value on interdependence and communion in girls’ than boys’ socialization [30] and more tolerance (at least in Europe) in sons’ than daughters’ peer relationships [23]. According to Grasmick et al. [31], these differences in child-rearing practices in traditional families yield a gender gap in children's risk preferences—boys have higher taste for risk than girls. This risk orientation during adolescence may be manifested in alcohol and drug use, aggressiveness, truancy, vulnerability to crime or other acting-out behaviors.

In less traditional and more egalitarian families, in which both parents have similar occupational authority, the process of socialization of boys and girls may be less differentiated. Maternal control of daughters is reduced, while their control of sons may be increased [32]. Yet, fathers in these families are more involved with their daughters, while maternal job prestige is not related to their involvement with sons [22]. As a result, children exhibit less stereotyped gender concepts, preferences and attitudes [33]. Females are more likely to be risk taking and finally, the gender gap in delinquency is smaller or does not appear—girls tend to behave like boys [32].

Power-control theory does not explain the mechanism of mother's and father's workplace authority transformation into authority within the family. This theory is rather based on the traditional vision of male's and female's roles in the society than on scientific models. Despite this theoretical gap, it gives a new perspective on negative trends in problem behaviors among adolescent girls, recently observed in Poland and in other countries [8–17].

To test the main ideas of power-control theory in Poland seems very intriguing for several reasons. First of all, in the past 20 years significant changes in the labor market have occurred. The percentage of women who are the main source of family income has been stable since the late 1980s (∼45%); nevertheless, their position in the workplace and social and political life is growing. Poland, for example, has observed a systematic growth of the percentage of women receiving scientific degrees (e.g. in the years 1991–2005 the rate of women with professor title increased from 22% to 27% of all professors). Similarly, the last several parliamentary elections in Poland have also seen an increase in women running for political office. In 1991, 13% of candidates were women that almost doubled in 2005 to 26%, and the number of women in the Parliament doubled from 10% to 20% during this same period. The presence of women in local governments increased as well [34]. Recently, the rate of women leading their own enterprises also increased from 34% in 2001 to 39% in 2005 [34]. Therefore, the position of women in the workplace and social and political life is growing. So it is possible that more and more girls grow up in less patriarchal families. Therefore, their drinking style becomes more similar to the way boys drink alcohol.

The second reason of our interest in power-control theory is that it was developed in Canada—a country that significantly differs from Poland in terms of social structure, recent history, labor market, etc. For example, communists ruling in Poland resulted in relative degradation of scientists’ position compared with clerks and blue-collars. Until now, incomes of university professors are in average 13% lower than incomes of department managers [35]. Therefore, in Poland the influence on adolescents’ development of parental work authority, as defined by Hagan et al. [20, 21], might be even stronger than in societies with longer and not-disrupted history of market economy.

The last but not least is the fact that power-control theory is not very well known in Poland and never before has been applied to explain any social phenomenon. One goal of this study is to examine sex differences in alcohol use in a sample of Polish adolescents. Guided by power-control theory we expect that adolescents’ alcohol use will reflect the changing dynamics of women in Polish society. This societal role change is expected to affect both parenting behavior and adolescents' behavior by reducing gender stratification. Adolescent females' alcohol use, for example, will be one manifestation of their desire for autonomy and may reflect differences in parenting for male versus female children [36]. In this study, heavy alcohol use was chosen as the dependent variable because it is the stronger indicator of problem behaviors than alcohol use [37–40]. Great majority of adolescents use alcohol but most of them may be characterized as social drinkers—nonproblem drinkers [41] or those who drink without getting drunk [42, 43].

We hypothesized that (i) parenting practices differ across child gender and power structure between parents; (ii) adolescents risk preferences are predicted by gender, household type and parenting practices of mothers and fathers; (iii) all variables mentioned above predict adolescents’ heavy alcohol use and girls in families with more maternal power are at higher risk for alcohol abuse than girls in more patriarchal families, while for boys the family structure has no influence on getting drunk.

Method

This research project was approved by the Bioethical Committee affiliated to the Institute of Psychiatry and Neurology in Warsaw, Poland, and by the Office that oversees the Human Subject Protection in the Fogarty International Center, National Institutes of Health, USA.

Data were collected during the first semester of the school year 2007/2008, from the representative sample of Warsaw middle-school eighth-grade students. Students’ average age was 14.5 years. Classroom was the unit of randomization. A total of 158 classes from 99 schools were selected and 148 classes participated in the study (therein 130 classes from public and 18 from nonpublic schools). Nine classes from special centers for adolescents with problem behaviors and school failures were also included.

Once classes were selected and consents for the study were obtained from school principals, parents and youths, students participated in the anonymous survey conducted in the classrooms by specially trained research assistants. Questionnaires were filed up by 3141 adolescents (85% sample execution). Due to incompletes or inconsistencies in answers, data from 3087 respondents were included in the analysis. Girls accounted for 50.4%.

Measures

Most of the measures used in the study (other sources of items are mentioned below) were adapted for a Polish sample from a study in the United States, the Flint Adolescent Study [44].

Heavy alcohol use

The measure of the dependent variable was based on the answer to the question about the frequency of getting drunk in the past 12 months (‘In the past 12 months, how often, if ever, you got drunk, when drinking beer, wine or vodka?’) with seven answers to choose (from never to 40 times or more). Because of the significant deviance from normal distribution (skewness = 2.850, kurtosis = 8.798), all answers were classified as never (74.2% of valid cases) or at least once (25.8%) and logistic regression method was chosen to test hypothesis.

Parent employment

Adolescents answered two questions (developed specially for this study) concerning separately mother's and father's employment status: (i) Does your mother/father work?, (ii) If your mother/father works, is she/he in charge of other people? For both questions, respondents were choosing one answer: yes, no, I don't know or I do not have mother/father. Yes–no answers were combined to indicate three categories of maternal/paternal working status: unemployed (15.2% of mothers and 5.3% of fathers), working without authority (mothers 47.6%, fathers 38.1%) and working in a position of authority (mothers 37.2%, fathers 56.5%). Those who answered ‘I don't know or I do not have mother/father’ were dropped out from the analysis (17.5%).

Categories of occupational authority in the workplace of respondent's parents were then collapsed into a dichotomy: families with fathers in authority positions in the workplace greater than those of mothers were categorized as patriarchal (34.8%) and families with mothers’ and fathers’ balanced levels of authority held in the workplace or families with mothers higher position than fathers were categorized as less patriarchal—egalitarian (65.2%).

Risk preferences

Respondents answered four questions [45] asking how often in the past 6 months they were doing something dangerous just for the thrill it evokes; riding bicycle or skating very fast, even risky, because it was exciting; doing some risky things because they were exciting and risking their safety staying out of home late in the evening because it was stimulating. Answers expressed on a five-point Likert's scale (from never to very often) were recoded to none risk preferences—if the answer for all four items was never (33.3% of cases), low—indicating positive answer for no more than two questions (30% of cases) and high risk preferences (36.7%).

Maternal and paternal support

Five items in the adolescent questionnaire concerned support they received from both mothers and fathers. Respondents indicated on a five-point Likert's scale their agreement or disagreement with the following statements: it's a pleasure for my parent to listen to what I am telling her/him; I can rely on my parent's emotional support; my parent knows how to help me in solving life problems; I am deeply attached to my parent; I can rely on my parent's moral support. Answers were recoded into three categories: low maternal/paternal support—indicating agreement with no more than one statement (16.7% for mothers and 29.3% for fathers), moderate—when respondent agreed with two, three or four statements (31.9% for mothers and 27.5% for fathers) and strong—representing agreement with all five statements (51.4% for mothers and 43.2% for fathers).

Maternal and paternal controls

Respondents were asked two questions, adapted from a Canadian questionnaire [46], tapping so-called instrumental controls [47] for both mothers and fathers: whether their mothers and fathers generally know where they are and whom they are with. Responses, expressed on a five-point Likert's scale, were coded into a dichotomy differentiating adolescents perceiving control as low—when respondent disagreed with both or one of the statements (rates of low maternal and parental control were, respectively, 41.6% and 56.4%) and strong—indicating respondent's positive answer for both statements.

Time with parents

Two questions measured time spent together with mother and father. On six-point Likert's scales (from none to very much), adolescents indicated how much time they spend with each of their parents in an average week, e.g. doing some housework together, talking or watching TV. Answers 1–4 (‘none’ to ‘fairly’) were coded as a little time (49.1% with mothers and 65.5% with fathers) and answers 5 and 6 (much and very much) as a lot of time (50.9% with mothers and 34.5% with fathers).

Table I reports descriptive statistics and Cronbach’s alpha information for all the psychosocial measures used in this study.

Table I.

Scales used in the study

Name	Number of items	Answers range	Mean	SD	Cronbach's alpha
Risk preferences	4	0–16	3.46	4.11	0.871
Maternal support	5	0–20	15.33	4.66	0.916
Paternal support	5	0–20	13.64	5.62	0.946
Maternal control	2	0–8	5.58	2.23	0.889
Paternal control	2	0–8	4.50	2.70	0.957

Analysis

Gender and household comparisons of alcohol abuse were based on cross-tabulations and χ² tests. Differences in parenting practices were tested by one-way analysis of variance with Tamhane's post hoc comparisons. The second hypothesis was tested using multinominal logistic regression analysis whereby risk preferences were predicted by parenting variables (support, control, time spent with child), household type and adolescent's gender. Logistic regression was chosen because the main objective of this analysis was to examine the interaction effect of two nominal variables: gender by family type, which is not possible in the linear regression.

The third hypothesis was tested in binominal regression equation in which in the first step, heavy alcohol use was predicted by parenting variables. In the following steps, adolescent's risk preferences (Step 2), gender and type of household (Step 3) and gender by type of family interaction (Step 4) were added.

In spite of loosing statistical power to detect significant relationships, it was decided to collapse all measures into categorical variables, with two or three levels, for two reasons. First, the main dependent variable—heavy alcohol use—was skewed and ‘flat’. Second, the key independent variables—gender and family type—are nominal. Since the main objective of the study was to show gender × family type interaction against the background of parenting variables and risk preferences, the statistical meaning of all these measures had to be balanced.

Results

Sex and family differences in adolescents’ heavy alcohol use

We found that getting drunk is less prevalent among adolescent girls from patriarchal families (18.6%) than it is in other subgroups: for boys in egalitarian families, the prevalence is 25.0%; for boys in patriarchal families 26.1% and for girls in egalitarian families 27.2% (χ² = 11.081, df = 3, P = 0.011).

Sex and family differences in parenting practices

As can be seen in Table II, there are significant differences in parenting practices toward daughters and sons in different types of families. Girls from patriarchal families spend more time with mothers than girls from egalitarian families and boys from either family type (F = 5.660, P = 0.001). On the other hand, in egalitarian families boys spend more time with fathers than girls do (F = 3.430, P = 0.016). There are no household type and gender differences in the level of maternal support (F = 0.810, P = 0.488), but daughters in egalitarian families receive significantly less support from fathers than sons in general (F = 7.543, P = 0.000). Maternal control is generally stronger for girls than for boys and for girls in patriarchal than for girls in egalitarian households (F = 17.183, P = 0.001). Paternal control differs only in egalitarian families being stronger for boys than girls (F = 2.620, P = 0.049).

Table II.

Analysis of variance comparisons of parenting variables in subgroups differentiated by adolescents’ gender and family type

	Boys/egalitarian	Boys/patriarchal	Girls/egalitarian	Girls/patriarchal
Time spent with mother
Mean	4.36a	4.40b	4.44c	4.63a,b,c
SD	1.05	1.12	1.18	1.08
	a: P = 0.000	b: P = 0.013	c: P = 0.022
Time spent with father
Mean	4.01a	3.93	3.81a	3.84
SD	1.34	1.31	1.34	1.28
	a: P = 0.017
Maternal support
Mean	15.64	15.30	15.33	15.51
SD	4.33	4.55	4.83	4.54
Paternal support
Mean	14.23a	14.56b	13.22a,b	13.65
SD	5.36	5.26	5.78	5.28
	a: P = 0.002	b: P = 0.000
Maternal control
Mean	5.36a,b	5.37c,d	5.80a,c,e	6.19b,d,e
SD	2.23	2.19	2.21	1.96
	a: P = 0.000, b: P = 0.000	c: P = 0.005, d: P = 0.000	e: P = 0.009
Paternal control
Mean	4.73a	4.57	4.37a	4.65
SD	2.60	2.66	2.76	2.60
	a: P = 0.042

Letters a, b, c, d and e indicate significant differences between two groups.

Adolescents’ risk preferences and gender, household type and parenting practices

Gender is a significant predictor of adolescents’ risk preferences—they are much stronger among boys than girls (Table III). The odds of strong risk preferences are five times higher for boys than girls. The relationship between risk preferences and parenting variables is weaker. The odds of low risk preferences (compared with none risk preferences) are 1.5 times higher for those who perceived moderate paternal support than for those adolescents who have stronger paternal support. When strong risk preferences are compared with none risk preferences, significant influence of parental control is visible. Odds of strong risk preferences are 2.5 times higher for adolescents perceiving low maternal control and 1.6 times higher for those with low paternal control. Risk preferences are not related to the power structure between parents.

Table III.

Multinomial logistic regression of risk preferences on gender, household type, parenting variables, n = 2282, Negelkerke R² = 0.161

	Low to none risk preferences		Strong to none risk preferences
	Wald	Exp (β)	Wald	Exp (β)
Maternal support (strong)
Weak	0.237	1.104	2.284	1.354
Moderate	0.003	0.993	1.926	1.213
Paternal support (strong)
Weak	0.451	1.125	0.001	0.994
Moderate	6.768**	1.455	0.002	0.993
Maternal control (strong)	3.231	1.294	41.507***	2.496
Paternal control (strong)	3.465	1.298	11.084**	1.632
Time with mother (a lot)	1.794	1.191	0.157	1.054
Time with father (a lot)	0.024	0.979	0.265	0.930
Gender (girls)	17.806***	2.205	70.446***	5.061
Household type (patriarchal)	0.180	1.063	1.687	1.238
Gender × household type	0.987	0.796	3.024	0.665

**P < 0.010; ***P < 0.001.

Adolescents’ heavy alcohol use and their risk preferences, gender, family type and parental practices

Regression of heavy alcohol use on parenting variables (Table IV) indicates that the odds of getting drunk for adolescents who perceive weak maternal control are 3 times higher than for adolescents perceiving strong control, 1.4 times higher for adolescents with weak paternal control than for those who are strongly controlled by fathers and nearly 1.5 times higher for those who spend a little time with mothers. Inclusion of risk preferences (Step 2) significantly improves the prediction of alcohol abuse by teenagers but makes the paternal control less significant. Strong and moderate risk preferences are significant predictors of alcohol abuse when parenting variables are controlled. When gender and household type are included in the equation (Step 3), a significant gender effect is visible—girls abuse alcohol more often than boys. The addition of the interaction of gender and family type improves the model and indicates significant relationship between household type and alcohol abuse (the odds of getting drunk for adolescents from egalitarian families are 1.5 times the odds for those from more patriarchal families) and shows the interaction effect (the risk of getting drunk is the lowest for boys from egalitarian families).

Table IV.

Binominal logistic regressions of alcohol abuse on gender, household type and parenting variables, n = 2194

	Step 1, R2 = 0.156		Step 2, R2 = 0.210		Step 3, R2 = 0.216		Step 4, R2 = 0.219
	Wald	Exp (β)	Wald	Exp (β)	Wald	Exp (β)	Wald	Exp (β)
Maternal support (strong)	0.818		1.259		1.290		1.364
Moderate	0.082	0.951	0.227	0.918	0.458	0.884	0.496	0.879
Weak	0.784	0.887	1.250	0.857	1.273	0.856	1.344	0.852
Paternal support (strong)	2.474		3.159		2.386		2.557
Moderate	1.944	1.255	2.986	1.334	2.165	1.280	2.320	1.292
Weak	0.040	1.030	0.407	1.101	0.201	1.070	0.217	1.073
Maternal control (strong)	68.064***	2.981	43.613***	2.451	47.006***	2.568	46.156***	2.549
Paternal control (strong)	5.281*	1.409	3.401	1.323	2.625	1.281	2.615	1.281
Time with mother (a lot)	9.112**	1.473	7.438**	1.431	8.701**	1.479	8.789**	1.482
Time with father (a lot)	1.968	1.225	3.189	1.302	2.972	1.292	2.933	1.290
Risk preferences (none)			89.103***		96.731***		95.605***
Low			5.030*	1.411	6.388*	1.478	6.281*	1.474
High			73.688***	3.325	81.472***	3.722	80.472***	3.700
Gender (girls)					9.064**	0.704	0.006	0.985
Household type (patriarchal)					1.114	1.130	5.061*	1.464
Gender × household type							4.592*	0.608

*P < 0.050; **P < 0.010; ***P < 0.001.

To better understand this interaction effect, separate logistic regression analysis were conducted for boys and girls (Table V) and different types of families (Table VI). As can be seen in Table V, significant predictors of heavy alcohol use among boys were weak maternal control, a little time spent with father and strong risk preferences. There was no household type effect. Heavy alcohol use by girls increased with their risk preferences and is predicted by maternal practices: support, control and time they spend together. Moreover, for girls, a significant household effect is observed—the odds of getting drunk are 1.4 times higher for girls from egalitarian families.

Table V.

Binominal logistic regressions of heavy alcohol use among boys (n = 1054) and girls (n = 1140)

	Boys, R2 = 0.182		Girls, R2 = 0.266
	Wald	Exp (β)	Wald	Exp (β)
Maternal support (strong)	0.782		7.014*
Moderate	0.464	1.194	3.835	0.600
Weak	0.668	1.171	6.515*	0.595
Paternal support (strong)	0.299		4.850
Moderate	0.017	0.969	4.785*	1.680
Weak	0.272	0.895	1.284	1.281
Maternal control (strong)	18.592***	2.359	27.654***	2.755
Paternal control (strong)	1.231	1.273	1.887	1.348
Time with mother (a lot)	0.182	1.084	14.320***	2.050
Time with father (a lot)	4.034*	1.519	0.535	1.172
Risk preferences (none)	40.301***		56.523***
Low	0.785	1.257	5.632*	1.591
High	25.595***	3.147	53.411***	4.159
Household type (patriarchal)	0.267	0.921	4.391*	1.436

*P < 0.050; ***P < 0.001.

Table VI.

Binominal logistic regressions of alcohol abuse by adolescents from egalitarian (n = 1430) and patriarchal (n = 764) families

	Egalitarian families, R2 = 0.243		Patriarchal families, R2 = 0.195
	Wald	Exp (β)	Wald	Exp (β)
Maternal support (strong)	6.063*		2.518
Moderate	3.001	0.679	1.953	1.584
Weak	5.594*	0.666	1.962	1.405
Paternal support (strong)	7.361*		2.104
Moderate	4.816*	1.570	0.439	0.818
Weak	0.008	0.983	0.443	1.185
Maternal control (strong)	38.871***	2.895	8.440**	1.997
Paternal control (strong)	1.393	1.255	1.081	1.302
Time with mother (a lot)	4.638*	1.423	3.789	1.565
Time with father (a lot)	2.250	1.320	0.594	1.217
Risk preferences (none)	70.683***		26.493***
Low	3.470	1.426	3.189	1.620
High	58.726***	3.930	23.414***	3.493
Gender (girls)	14.268***	0.580	0.002	1.009

*P < 0.050; **P < 0.010; ***P < 0.001.

Data in Table VI indicate significant differences in alcohol abuse predictors between families. In both types of families, there is significant relationship between adolescents’ alcohol abuse and their risk preferences and the level of maternal control. Moreover, in egalitarian households the risk of getting drunk is higher for adolescents perceiving low maternal and paternal support, for those spending less time with mothers and for girls.

In order to further understand the effect of family structure and differentiate it from the effect due to both parents working outside of the household (which can potentially result in less time to supervise the child), we also conducted analyses that further disaggregated family structure. We defined four family categories differentiating unemployed parents:

• (i) mother is unemployed/father works (13% of the sample),

• (ii) mother works without authority/father works in a position of authority (21%),

• (iii) both parents work and mother's work position is equal or higher than father's (60%),

• (iv) father is unemployed (6%).

We found no differences in the amount of time spent by adolescents with unemployed mothers and with mothers from other patriarchal families, but families with unemployed mothers differed from egalitarian families. Conversely, unemployed fathers spent less time with their children than fathers from the other three family structures.

Discussion

This study tested the viability of power-control theory for explaining gender differences in the level of alcohol abuse in a sample of Polish adolescents. The results provide support for the main thesis of this theory.

As hypothesized, the socialization process is different for boys and girls in different household types. Girls perceive stronger maternal control than boys and girls in patriarchal families are more controlled by mothers than girls in egalitarian families. Moreover, daughters in patriarchal families spend more time with mothers than any other group of adolescents. These findings are consistent with [21] the idea that in traditional families mothers are the primary instruments of control and daughters are the primary objects of control.

The results of this study confirm that in less traditional and more egalitarian families, the process of boys’ and girls’ socialization is less differentiated. But it is not due to increase of maternal control of sons (as suggested by Hagan et al. [32]) or greater paternal involvement with their daughters (as suggested by McHale et al. [22]). Rather the result seems to be due to more paternal involvement with sons. This involvement is expressed in more time spent with fathers by boys in egalitarian families than by girls in the same type of households, stronger paternal control of sons than of daughters in egalitarian families and stronger paternal support perceived by boys, in general, than by girls from egalitarian families.

Thus, in families in which father's occupational authority is not higher than mother's, the gender gap in parenting practices, of both mothers and fathers, is smaller than in patriarchal households, and therefore, the gender gap in their teenagers’ heavy alcohol use is smaller, too. This is indicated by comparisons of the frequency of getting drunk by boys and girls in egalitarian and patriarchal families showing that in patriarchal families boys are at a higher risk of alcohol abuse than girls and girls get drunk more often when they grow up in egalitarian families. It may suggest that parenting practices and household type are influential on adolescent's behaviors, despite some basic, probably congenital tendencies, such as stronger vulnerability to acting-out behaviors of boys than girls.

According to the power-control theory, high risk preferences are one of the key factors determining problem behaviors across gender and family structure [31]. Results confirmed this thesis but they do not support the idea of Hagan et al. [48] that social factors (parenting practices) have a strong effect on adolescents’ risk preferences. They rather suggest, in line with psychobiological models of personality [49, 50], that risk preferences are mainly determined by biological factors (e.g. gender).

Among parenting variables, the most significant predictors of alcohol abuse by teenage children were maternal control and time spent with a mother. When gender and household type are controlled, we found that females were more likely than boys to get drunk. But the inclusion of gender and household type interaction modifies this result showing more precisely higher risk of alcohol abuse for adolescents in egalitarian families with the highest risk for girls from egalitarian families. This significant interaction effect confirms the third hypothesis of this study, indicating that the household type and gender have a differential effect on alcohol abuse after controlling for parenting practices, time spent with parents and risk preference. The effect of family structure is further strengthened by the analysis, which helps to reduce alternative explanations, that the differences between the families are due to parent's presence or absence at home (employment or unemployment).

The findings of this study probably can be generalized for heavy alcohol use by adolescents from other big cities in Central and East Europe. As mentioned in the Introduction, in the second half of the XX century the labor market was similar in all countries under Soviet domination (especially in Poland, Czechoslovakia, Hungary). There were no major differences between men and women employment, and part-time employment or working at home was exceptional. Moreover, the transition processes initiated 20 years ago and are similar in their directions and dynamics.

The most significant limitation of this study is its cross-sectional nature. This reduces the possibility of formulating conclusions about causal relationship between variables but does provide some evidence that a longitudinal study may be warranted in order to test the lead–lag relationship of parenting effects on youths' alcohol use behavior.

A second limitation of the study is that our dependent variable was measured with a single item. Although behavioral measures are often measured in this way, future research that uses a multi-item index that includes alcohol use more generally (past 30 days), number of drinks per occasion and getting drunk might provide a more widely distributed variable and therefore more variance to explain. Yet, the fact that we found effects with a single item measure suggests that the effects of parenting and family context may be vital factors influencing youth’s alcohol use.

Finally, we did not specifically measure power relationships in the family as we inferred them based on parent employment. Hence, this study does not fill the gap in the original power-control theory. It only indicates that this concept might be useful for explaining adolescents’ problem behaviors and delinquency in Central and Easter European countries.

Implications for future research

Future research to better understand the process of work authority transformation into family processes is needed. This could include specific measures of the power relationship between a mother and a father at home, the process of decision-making and perceptions of responsibility for raising their children. Moreover, adolescents’ perception of the relationship between mother and father should be controlled.

It would be interesting to verify the usefulness of power-control theory for other problem behaviors of teenage boys and girls, e.g. other psychoactive substance use and abuse or delinquent behaviors. Cross-cultural studies testing the relationship between mothers’ and fathers’ work experiences, family processes and child's socialization in countries with various labor traditions and economic systems might extend our understanding of globally observed trends in risky and problem behaviors among boys and girls.

Implications for prevention

Despite limitations, this study provides useful information for understanding sex differences and the closing gender gap in Polish adolescents' alcohol use. Power-control theory seems to be a useful guiding framework in Poland as females' position in the workforce is changing in important ways that may further change the balance of family dynamics so critical in raising children. Moreover, findings of this study point out significant issues that could be addressed in family-based alcohol prevention programs.

Firstly, parents of teenagers should be aware that the way they perform their parental roles still counts, as it used to count when kids were younger. Hence, prevention programs strengthening families of adolescents should be disseminated. A good example of such a program is Strengthening Families Program (SFP 10–14), which develops parental ability to show love and set limits [51]. Recently, this program has been adapted for Polish families [52].

Secondly, a focus on parents, especially mothers, to raise awareness of the importance of their supervision over adolescents behaviors and encouragement to spend more time with their teenage children would be beneficial. Especially, mothers from egalitarian families should be encouraged to maintain effective control over daughters in a way this control used to be performed in traditional patriarchal families.

Thirdly, program that focus on fathers, especially from egalitarian families, to encourage them to have deeper involvement in the rearing process of daughters also may be important.

Funding

Fogarty International Center, US National Institutes of Health (5R01TW007647).

Conflict of interest statement

None declared.