Addictive Behaviors 39 (2014) 1898–1903

Contents lists available at ScienceDirect

Addictive Behaviors

The relationship between the density of alcohol outlets and parental supply of alcohol to adolescents B. Rowland a,⁎, J.W. Toumbourou a, L. Satyen a, M. Livingston b,c, J. Williams a,d,e a

Department of Prevention Sciences, Centre for Mental Health and Wellbeing Research and School of Psychology, Deakin University, Victoria, Australia Drug Policy Modelling Program, National Drug and Alcohol Research Centre, University of New South Wales, Sydney, NSW 2052, Australia Centre for Alcohol Policy Research, Turning Point Alcohol and Drug Centre, 54-62 Gertrude Street, Fitzroy, Victoria 3065, Australia d Murdoch Children's Research Institute, The Royal Children's Hospital, Melbourne, Victoria, Australia e Department of Paediatrics, The University of Melbourne, Parkville, Australia b c

H I G H L I G H T S • • • •

Greater densities of alcohol sales outlets are associated with greater alcohol use and problems. Alcohol outlet densities have been linked to adult and adolescent alcohol-related behaviour. However, the mechanisms as to how this may occur are unclear. We examined whether density was associated with parental supply of alcohol to adolescents.

a r t i c l e

i n f o

Available online 8 August 2014 Keywords: Alcohol density Supply Alcohol Migrants Adolescents Parenting

a b s t r a c t This study investigated whether the number of alcohol outlets per 10,000 population in a given area (density) influenced parental supply of alcohol to adolescents; differences in Australian born and acculturating parents were also examined. A state-representative student survey in Victoria identified that the majority of adolescents (55%) reported that they had used alcohol in the past 12 months; 34 % of those who had consumed alcohol reported that it had been supplied by their parents. Multilevel modelling identified that there were no overall effects of density, however there were different effects based on parent country of birth and type of license. Specifically, each unit increase in the density of takeaway liquor stores increased the likelihood by 2.03 that children with both Australian-born parents would be supplied alcohol. Adolescents with both migrant parents on the other hand, had a 1.36 increased risk of being supplied alcohol as the density of outlets requiring atvenue consumption increased. The findings of this study suggest that in Australia, alcohol outlet density is associated with parental supply of alcohol to children, with this effect moderated by the cultural background of the parent and type of outlet density. Future research should investigate the association between the density of alcohol outlets and public approval of parents supplying alcohol to adolescents. © 2014 Elsevier Ltd. All rights reserved.

1. Introduction The early uptake and consumption of alcohol by children and adolescents are precursors to a range of negative outcomes including: behavioural problems; substance use disorders; poor academic achievement (Bonomo, Bowes, Coffey, Carlin, & Patton, 2004; Bonomo et al., 2001; Howard, Qiu, & Boekeloo, 2003); and increased health risk behaviours (Dye & Upchurch, 2006). Although the legal age for purchasing alcohol is 18 years in all Australian states, approximately 51% of ⁎ Corresponding author at: Deakin University, School of Psychology, Faculty of Health, 221 Burwood Highway, Burwood, Victoria 3125, Australia. Tel.: +61 3 9244 3002. E-mail address: [email protected] (B. Rowland).

http://dx.doi.org/10.1016/j.addbeh.2014.07.025 0306-4603/© 2014 Elsevier Ltd. All rights reserved.

Australian children between 12 and 17 years of age have reported consuming alcohol in the previous 12 months (White & Hayman, 2012). Further, approximately 32% of Australian 12 to 17 year-old drinkers report that their parents provided them with their last alcoholic beverage (White & Hayman, 2012). The operation and practices of licensed sales outlets have been studied to identify sources of alcohol supply to adolescents. Evidence indicates that the greater the number of alcohol sales outlets per capita within a geographic population (density of alcohol outlets), the greater the risk that an adolescent will consume alcohol. This relationship has been found to exist in Australia (Livingston, 2008; Rowland et al., 2013), the United States of America (USA) (Chen, Gruenewald, & Remer, 2009), Switzerland (Kuntsche, Kuendig, & Gmel, 2008) and

B. Rowland et al. / Addictive Behaviors 39 (2014) 1898–1903

New Zealand (Huckle, Huakau, Sweetsur, Huisman, & Casswell, 2008). Reducing the density of alcohol outlets combined with the enforcement of laws prohibiting sale of alcohol to minors are proposed as possible strategies that could contribute to reducing the proportion of adolescents purchasing and consuming alcohol (Loxley et al., 2004; Rowland et al., 2013). Few studies have examined precisely how density might influence the supply of alcohol to adolescents. The present study sought to examine the possibility that higher densities of alcohol outlets may lead to greater parental supply of alcohol to teenagers. We also sought to examine whether the effect of densities on parental supply of alcohol may be modified by parental country of birth (COB). There has been little investigation in Australia of parent practices around supplying alcohol to children. Given that accessing alcohol is a necessary prerequisite for consumption, understanding family and community-level factors related to parental supply is critical in identifying effective policy and prevention strategies. As parents and other adults are the predominant suppliers of alcohol to children in Australia (White & Hayman, 2012), ascertaining how density and cultural background are associated with supply practices may help to understand the context of adolescent alcohol consumption. In their review, Rowland, Toumbourou, and Stevens (2003) found consistent evidence that Australian adolescents are less likely to consume alcohol when their parents are non-Australian born. Similarly, findings from the longitudinal Household, Income and Labour Dynamics in Australia (HILDA) survey indicate that adolescents with both parents or at least one parent who is an immigrant are less likely to consume alcohol when compared to native-born adolescents with two Australian-born parents (Brandon, 2008). Studies in Holland (van Tubergen & Poortman, 2010), the USA (Brown, Council, Penne, & Gfroerer, 2005) and Canada (Perez, 2002) have also reported similar associations. There are theoretical reasons why different types of outlet density might interact with family cultural background to influence adult supply of alcohol to children. Social cognitive theory argues that the environment can act as a cue, or may reinforce existing or developing behaviours (Bandura, 1998; Baranowski, Perry, & Parcel, 2002). For example, parents with liberal attitudes may have that attitude reinforced by exposure to a higher number of take-away outlets where alcohol is visibly marketed. Such a context may increase the amount of alcohol purchased for use in the home and thus make it more likely that children are supplied alcohol. Similarly, laws in the state of Victoria allow parents to supply alcohol to children in licensed venues where food is served. It is therefore possible that proximity to this type of outlet may result in less liberal parents witnessing children being supplied alcohol by other parents and reinforce this as normative for their own children. Over time this may lead newly migrant families who are less likely to supply alcohol to children (Brandon, 2008; Rowland et al., 2003) to adopt this practice as part of the assimilation process. Migrant parents might assume some of the host culture's practices and allow their children to explore the values and behaviours of the new society in order to facilitate the children's adaptation to the new setting (Roer-Strier, 1997). Using a representative sample of Australian school children in the state of Victoria, and controlling for risk factors known to be associated with adolescent alcohol consumption (Arthur et al., 2007; Hawkins, Arthur, & Catlano, 1995), we examined the hypothesis that higher density would be associated with higher rates of parental supply. This hypothesis recognised that density is associated with higher adult alcohol consumption (Livingston, 2008; Rowland, Toumbourou, Satyan, et al., 2013), and that parents are the primary suppliers of alcohol to adolescents (White & Hayman, 2012). Given that adolescent consumption is lower in non-Australian born families (Brandon, 2008; Rowland et al., 2003), we hypothesised that density would lead to greater increases in supplying alcohol to adolescents when parents were Australian born compared to non-Australian born (acculturating) parents.

1899

A variety of variables were entered to control for potentially spurious associations between density and parent supply of alcohol. Control variables were selected on the basis that they may influence adolescent alcohol consumption (Arthur et al., 2007; Hawkins et al., 1995) or parent supply of alcohol to adolescents (Gilligan, Toumbourou, Kypri, & McElduff, 2013; McMorris, Catalano, Kim, Toumbourou, & Hemphill, 2011; Zucker, 2006) and also show variation across communities. These included area socioeconomic disadvantage, non-metropolitan location and the adolescents: age; school grade level; gender; cigarette smoking; peer alcohol and drug use; and parental attitudes to adolescent drinking. 2. Methods 2.1. Design Data were collected in 2009 through the HowRU secondary student survey; a study designed to provide representative epidemiological estimates of adolescent health and wellbeing indicators for all metropolitan local government communities and non-metropolitan regions across the state of Victoria in Australia (Williams, Kent, Canterford, & Basile, 2010). A two-stage cluster sample design was used to recruit students. In the first stage, schools were randomly selected based on a probability proportional to each community's grade-level size from a stratified sampling frame of all schools in Victoria (government, Catholic, and independent). In the second stage of the sampling, whole classes in school years 7, 9 and 11 were chosen at random. Survey procedures were approved through the Royal Children's Hospital Ethics Office and relevant school authorities. Of the 13,501 eligible students, 10,273 (77.2%) consented and participated. Table 1 Sample demographics and the distribution of variables used in the analyses. N (%) Female Age 12 13 14 15 16 17 Year 7 Year 9 Year 11 Alcohol in last 12 months Respondent born in Australia Parent both Australian born Parent both non-Australian born Mother Aust. father non-Aust. born Father Aust. mother non-Aust. born Regions Metropolitan Non-metropolitan

5132 (50.6) 1907 (18.8) 1671 (16.5) 2075 (20.5) 1588 (15.66) 1754 (17.3) 1148 (11.3) 3625 (35.7) 3640 (35.9) 2878 (28.37) 5554 (54.76) 8774 (87.16) 5229 (52.50) 2893 (29.05) 1071 (10.75) 767 (7.70) 8515 (83.95) 1628 (16.05)

Alcohol outlet density measures

M (SD)

ln M (SD)

Overall density Packaged density General density On premise density Club density Overall density: metropolitan Overall density: non-metropolitan Proportion of friends who use drugs Parents favourable attitude to drug use SEIFA

25.26 (29.84) 4.33 (1.71) 4.51 (7.36) 14.78 (20.88) 1.69 (1.04) 24.27 (32.46) 30.46 (1.51) 1.72 (.80) 1.90 (.47) 1017.17 (46.78)

2.31 (1.06) 1.27 (.41) 0.41 (1.43) 1.83 (0.60) 0.38 (0.60) 2.15 (1.08) 3.16 (0.06)

N = 10,143. Overall density: all types of alcohol outlets; package density: shops that sell liquor to be consumed elsewhere; on premise: sells food and alcohol (e.g. café); clubs: licensed club where membership is required prior to consuming alcohol. SEIFA: Socio-Economic Index for Area (ABS, 2006). M = mean. Ln M = log normal mean. SD = standard deviation.

1900

B. Rowland et al. / Addictive Behaviors 39 (2014) 1898–1903

The analysis sample for this paper includes only those who were under the age of 18, the legal age for purchasing alcohol in Victoria. The total sample used in the analyses was 10,143. Community sampling was based on the school location within local government areas across metropolitan Melbourne. Outside metropolitan Melbourne, sampling was based on Education Department regions, reflecting the major community units responsible for youth services. 2.2. Sample Table 1 outlines the individual demographic characteristics of the sample. The most prevalent age was 14 years old (n = 2075; 20.5%). The youngest age was 12 (n = 1907; 18.8%) and the oldest age was 17 (n = 1148) years old. The average age was 14.31 (SD 1.64) years. Table 1 also indicates that 87.2% of respondents were born in Australia, and approximately half (52.5%) report that both their parents were born in Australia. Just under one third (29.1%) reported that both their parents were non-Australian born. There were significant differences between the combination of mother's and father's birthplace and whether an adolescent had been supplied alcohol by their parents (χ2 (3) = 40.58; p b .000). Just over half the sample, 54.8% (95 CI: 53.8, 55.7; n = 5554), reported consuming alcohol in the last 12 months. Table 2 presents a breakdown of the reported sources of alcohol supply for respondents, for the whole sample, and also only for those that reported alcohol use. Approximately 18.5% (95 CI: 17.8, 19.3) of the whole sample reported that his or her parent had provided them with alcohol. This was approximately 34% of those who reported alcohol use (33.8: 95 CI: 32.6, 35.1). 3. Measures 3.1. Dependent variable 3.1.1. Source of alcohol supply The dependent variable was whether a parent purchased alcohol for an adolescent. First, respondents were asked whether they had consumed alcohol in the last 12 months. Those who reported alcohol consumption in the last 12 months were asked how they obtained their alcohol the last time they consumed alcohol. The following options could be selected: ‘I bought it’ or ‘Someone else bought it for me’. If the answer was ‘Someone else bought it,’ respondents were asked to indicate who bought it. Responses included: ‘parents’, ‘brother/sister’, ‘home without permission’, ‘friends’, ‘got someone to buy it’, or ‘other’. These questions are commonly used in adolescent drug and alcohol surveys (AIHW, 2010; White & Hayman, 2012). A variable that indicated a parent bought them alcohol was created. For this variable, the value ‘one’ represented parents purchasing alcohol the last time the Table 2 Supply of alcohol to adolescents for whole sample. Source of alcohol supply

(N) Whole sample

% Whole sample

% Only drinkers

Parents Brother/sister Home (without permission) Friends Got someone to buy it Other Bought alcohol self Did not consume alcohol last 12 months Missinga Total

1879 355 187 1163 622 705 511 3929

18.53 3.50 1.84 11.47 6.13 6.95 5.04 38.74

33.83 6.39 3.37 20.94 11.20 12.70 9.20 N/A

792 10,143

7.81 100% (N = 10,143)

2.38 100 (N = 5554)

a 792 cases missing in whole sample, 132 cases missing for drinkers in the last 12 months; only drinkers are respondents who reported to have drunk alcohol in the last 12 months; those who did not consume alcohol in last 12 months were coded as not being supplied alcohol in the analysis.

adolescent consumed it in the previous 12 months, and ‘zero’ represented parents not having bought them alcohol in the last 12 months. Those who reported not drinking alcohol in the last 12 months were coded as not being supplied alcohol by a parent. 3.2. Independent variables 3.2.1. Alcohol density Density was measured as the number of outlets per 10,000 residents of a given metropolitan local government area or non-metropolitan education region (LGA). A youth-specific population denominator was not used as it would have ignored the overall environmental influence of alcohol availability on all population groups in an area that could lead indirectly to increased supply of alcohol to adolescents. Using the total population denominator also adjusted for variations in area sizes. Other measures (e.g., area, roadway miles to alcohol outlets) (Scribner, Cohen, Kaplan, & Allen, 1999) provide alternative ways of capturing this information; these have their own biases (e.g., areabased density measures grossly underestimate density in nonmetropolitan areas) (Livingston, 2012). Liquor licencing data were supplied by the Victorian Commission for Gaming and Liquor Regulation and were then geocoded by a commercial organisation (Mapdata Sciences Australia). Overall, 99.5% of licences were geocoded to at least street level, comparable to the results of previous studies (Donnelly, Poynton, Weatherburn, Bamford, & Nottage, 2006; Hay, Whigham, Kypri, & Langley, 2009; Kypri, Bell, Hay, & Baxter, 2008). These licences were then assigned to the LGA in which they were located using ‘MapInfo’ to convert the point data to counts per community population in 2008. Alcohol density outlet was organised into four categories: general density; packaged outlet density; on-premise density; and club density. ‘General outlet density’ was defined as public bars (pubs), ‘packaged liquor outlets’ was defined as shops that sold takeaway liquor such as bottle shops; ‘on-premise alcohol outlets’ was defined as restaurants or venues that sold food (e.g. café) and alcohol; ‘licensed clubs’ was categorised as venues where membership was required prior to alcohol consumption at the venue. As this variable was not normally distributed and to avoid multicollinearity issues between the density variables, a natural log transformation that reduced the skewness to zero was applied to all density measures (Stata, V12) and centred at their mean. 3.3. Control variables/risk factors The following factors known to be associated with adolescent alcohol consumption or parent supply of alcohol were controlled in the analyses. 3.3.1. Individual risk factors The following are the general demographic information: respondent's age, school grade level (years 7, 8 and 9) and gender. Respondents were also asked to respond yes or no to the question “In your lifetime have you ever smoked cigarettes?” 3.3.2. Peer risk factors Respondents were asked to indicate the perceived proportion of friends who used drugs. Questions were asked about four types of drugs: “Think of your four best friends (the friends you feel closest to)”. In the past year (12 months), how many of your four best friends have used the following: smoked cigarettes; tried alcohol (like beer wine, spirits, premixed drinks… when their parents did not know); used marijuana, cannabis (pot, weed, grass); and used other illicit drugs (like cocaine, heroin, LSD/Acid or amphetamines/speed)? Responses were: “none of my friends”; “one of my friends”; “2 of my friends”; “3 of my friends”; “4 of my friends”. The average number of friends for these four questions was used as an index for the proportion

B. Rowland et al. / Addictive Behaviors 39 (2014) 1898–1903

of friends who use any type of drug (Bond, Thomas, Toumbourou, Patton, & Catalano, 2000).

1901

5. Results 5.1. Multilevel regression analyses

3.3.3. Parent country of birth in Australia Respondents were asked to report their mother's and father's country of birth (COB). Parents' COB was recoded as ‘Australian’ or ‘other’.

3.3.4. Parent's attitude to drug and alcohol use A four item scale was used to assess parent's attitude to adolescent drug and alcohol use: My parents feel it is wrong if I drink beer/wine regularly? My parents feel it is wrong if I drink spirits regularly? My parents feel it is wrong to drink ready to drink beverages (e.g. RTD, alcopops) regularly? My parents feel it is wrong if I use marijuana? Responses were made on a four point likert scale: 1 = not wrong at all; 2 = a little bit wrong; 3 = wrong; and 4 = very wrong. Scores for each item were averaged (Bond et al., 2000). The Cronbach's alpha for this scale used in this study was 0.78.

3.3.5. Community risk factors A Socio-Economic Index for Areas (SEIFA) was assigned according to the postcode (zipcode) for the area of residence of each respondent. The index used was the advantage/disadvantage index score, provided by the Australian Bureau of Statistics (ABS). It is a measure of relative socio-economic advantage and disadvantage; low values indicate areas of disadvantage, high numbers indicate areas of advantage (ABS, 2006). The variable was categorised into deciles of SEIFA index relative to the overall State of Victoria. Whether a respondent attended school in a metropolitan or non-metropolitan area was also recorded.

4. Statistical analysis Multilevel modelling was used to assess whether the density of alcohol outlets predicted parental supply of alcohol (Bickel, 2007; Bryk & Raudenbush, 1992). As we examined the influence of the density of outlets in a geographic area, LGA was specified as a random variable. The variables were organised into two levels. Level two (community) variables included the LGA, density variables, and SES index, and whether students attended school in a non-metropolitan location. All other variables were categorised as level one (individual) variables.

4.1. Analytical strategy Following West, Welch, and Galecki (2007), a three-stage analytical strategy was followed. First, a null model (only a random intercept, varying by area) was analysed to estimate the community level variance associated with parental supply of alcohol; it also provided a baseline estimation of the community variance in predicting adolescent alcohol supply. Second, the level one variables were entered into the model and the consideration of level one random effects was examined. Third, community level variables were entered. In keeping with the West et al. analytical approach, potential interaction effects between all variables at each level (individual and community) and across levels (individual vars#community) were examined. Non-significant predictors were removed at each step. The log likelihood (LL) and the Akaike Information Criteria (AIC) and Bayesian Information Criteria (BIC) were used to assess model fit. As the density variables were of primary interest, all the density variables were entered in the model at the beginning, and remained in the model throughout the model building process.

The intraclass correlation (ICC) and the measure of within-group (LGA) homogeneity for the dependent measure (whether the respondent had been supplied alcohol) were 0.02 (95 CI: .01, .03). The null model indicated that parental supply varied significantly between LGAs (OR = .22, p b .001). The baseline variance between communities was .054. Table 3 outlines the results of the final multilevel model. To assist with interpretation, the same model is presented twice; however, the referent for parents' COB is changed. In Model 1, the referent for parents' COB is both parents were born overseas. For Model 2, the referent is both parents are Australian-born. There was no main effect for each of the density variables. However, an interaction was identified between the density of packaged outlets and parental country of birth (see Model 1). Adolescents with both parents born in Australia were 2.03 (p = .016) times more likely to be supplied alcohol by parents for each one unit increase in the density of packaged outlets, compared to adolescents who had parents both born overseas. Similarly, an interaction with on-premise density and parental COB was identified. Adolescents with both parents born outside Australia were 36% more likely to be provided alcohol as on-premise density increases (see Model 2), compared to adolescents who had parents that were both Australianborn (OR = 1.36; p = .003). The final multilevel model also identified that year level was a significant predictor of whether an adolescent was supplied alcohol; as year level of respondent increases, the odds of being supplied alcohol increase. The gender of an adolescent and whether a respondent smoked were not significant predictors of whether parents supplied alcohol and hence not included in the final model. The community level variables, SEIFA, and whether a respondent resided in a metropolitan or nonmetropolitan area were not significant predictors in the final model. When compared to the null model, the final model reduced the variance within communities by approximately 70% ((.054–.0.016)/.054). The change in the LL statistic indicated that the final model was significantly better fit than the null model (χ2 (15) = 475.41; p = .000). The reduction in the AIC and BIC fit statistics also indicated improvement in model fit compared to the null model. Due to missing data 804 cases (7%) dropped out of the final MLM model. 6. Discussion The hypothesis for the study was partially supported — there was no overall effect of density on parent supply, however there were differential effects for parents' birth place. Packaged outlet density was associated with increased risk of supply for adolescents with parents who were both Australian born. On premise density was associated with increased risk of supply for adolescents who had parents who were both nonAustralian born. The findings of this study suggest that parental country of birth is associated with supply of alcohol to adolescent children, and that this relationship is moderated by outlet density. Why would the type of alcohol outlet density moderate the likelihood of parents born in different countries supplying alcohol to adolescents? This may be linked to the type of drinking usually associated with each venue. Packaged outlets sell alcohol that must be consumed off site. Adolescents who have both Australian-born parents may be more likely to be provided with alcohol to be consumed elsewhere such as parties, and possibly not under adult supervision. Thus, as the density of packaged outlets increases, the likelihood of Australian born parent's providing alcohol increases. In contrast, on-premise outlets sell alcohol that by law must be consumed on the premises where it is purchased. This study found that as on-premise density increased adolescents with both parents born overseas were more likely to be provided alcohol.

1902

B. Rowland et al. / Addictive Behaviors 39 (2014) 1898–1903

Table 3 MLM for density of outlets and supply of alcohol by parents to adolescents. Variable

Null

Model 1

OR (95 CI)

OR (95 CI)

P

OR (95 CI)

P

Year level Year 7 Year 8 Year 9 Pfud FADA Package On-premise

Referent 1.31 (1.15, 1.50) 1.48 (1.27, 1.72) 0.78 (0.73, 0.84) 2.60 (2.32, 2.91) 0.68 (0.39, 1.16) 1.13 (0.94, 1.37)

N/A 0.000 0.000 0.000 0.000 0.159 0.195

Referent 1.31 (1.15, 1.50) 1.48 (1.27, 1.72) 0.78 (0.73, 0.84) 2.60 (2.32, 2.91) 1.37 (1.03, 1.84) 0.82 (0.74, 0.93)

N/A 0.000 0.000 0.000 0.000 0.028 0.001

Parent's COB Both AB Both OB MA/FN MN/FA

1.96 (1.70, 2.26) Referent 1.82 (1.50, 2.22) 1.64 (1.31, 2.05)

0.000 N/A 0.000 0.000

Referent 0.51 (0.44, 0.59) 0.93 (0.79, 1.10) 0.83 (0.68, 1.02)

N/A 0.000 0.417 0.087

Package # P_COB Both AB Both OB MA/FN MN/FA

2.03 (1.14, 3.62) Referent 1.40 (0.65, 3.02) 2.39 (0.97, 5.93)

0.016 N/A 0.386 0.059

Referent 0.39 (0.28, 0.88) 0.65 (0.37, 1.28) 0.82 (0.54, 2.58)

N/A 0.016 0.240 0.684

On-premise # P_COB Both AB Both OB MA/FN MN/FA Constant Variance (σ2); 95 CI LL AIC BIC N (ind) N (LGA) df

0.73 (0.59, 0.90) Referent 0.79 (0.59, 1.03) 0.85 (0.62, 1.17) 0.02 (0.02, 0.03) .016 (.005, .053) −4378.50 8790.99 891.64 9469 36 17

0.003 N/A 0.091 0.322 0.000

Referent 1.36 (1.11, 1.69) 1.07 (0.84, 1.36) 1.16 (0.88, 1.54) 0.04 (0.03, 0.05) 016 (.005, .053) −4378.50 8790.99 891.64 9469 36 17

N/A 0.003 0.550 0.292 0.000

.23 (.21, .25) .054 (.027, .107) −4616.20 9236.40 9250.71 9469 36 2

Model 2

Pfud: proportion of friends who use drugs; FADA: family's attitude to drugs and alcohol; P_COB: parent's country of birth; AB: Australian born; OB: overseas born; MA: mother Australian born; FA: father Australian born; FN: father non-Australian born; MN: mother non-Australian born; #: denotes interaction; Models 1 and 2 are the same except that Model 1 uses “both parents BO” as the referent while Model 2 uses “both AB” as referent; σ2: variance within LGA; N(ind): number of individual respondents; N(LGA): number of clusters/local government areas; df = degrees of freedom; LL: log likelihood; AIC: Aikake Information Criteria; BIC: Bayesian Information Criteria.

The current findings suggest that non-Australian parents may overcome their reservations and supply alcohol to adolescents where they are living proximal to alcohol outlets offering alcohol with meals. According to social cognitive theory observing the behaviour of other parents may cue parental behaviour change. On-premise outlet licence regulations may expose parents and children to family role models where parents supply alcohol to children. It is possible that when nonAustralian parents living near these outlets observe parents offering adolescents alcohol with meals, they then adopt these behaviours in an effort to integrate into the host culture. Research will be required to confirm this proposition. Given the high rates of underage alcohol use and parent supply of alcohol in Victoria (Gilligan et al., 2013), and that this supply has been shown to predict progression to heavy adolescent drinking (McMorris et al., 2011). The current findings suggest that Victorian liquor licencing regulations should be changed to reflect those of other Australian states, such as New South Wales that prohibit any supply of alcohol to youth under the age of 18 years. The association between packaged density and supply could also be linked to parental consumption. There is evidence that as the density of packaged outlet increases, adult alcohol consumption increases (Campbell et al., 2009). In turn, it has also been shown that higher levels of parental alcohol consumption predict: parental supply of alcohol to adolescents (Gilligan et al., 2013) and adolescent consumption (Hawkins et al., 1997; Zucker, 2006). To some extent, the present analysis did control for this influence with the inclusion of parental attitude to drug and alcohol. However, the possibility that parental alcohol use explains the effect of density on parent supply behaviour cannot be entirely ruled out and should be given further attention in future studies.

The present study's strengths are the large state-representative school sample and the area-level linking of data on alcohol sales outlet density. It has also identified an interaction between parental country of birth and supply that has not been previously reported in the literature. However, the study design cannot rule out counterfactual causal effects. It is reasonable to suspect that increases in density precede increases in adolescent supply, as the reverse causal direction is not logical. It is possible that links between alcohol outlet density and higher rates of supply arise from common associations with underlying factors, such as disadvantaged localities having higher rates of adolescent maladjustment, or more favourable community attitudes to alcohol. However, the present analysis has controlled for these factors as well. While this study has identified associations with the density of alcohol outlets and adolescent reports of purchasing, the cross-sectional nature of the study design is an important limitation and suggests the need for caution when inferring causal effects. Although the study cannot establish causal directions, it is intuitively reasonable to suspect that increases in density precede increases in adolescent supply of alcohol, as the reverse causal direction is not plausible. Longitudinal analyses would help tease out the potential mechanisms more clearly and should be the subject of future research. As the data is self-report, the accuracy of the extent that adolescents were provided with alcohol should be considered with some degree of caution. Observational studies that validate the accuracy of adolescent self-reports of being supplied alcohol is an area worthy of investigation. This is the first Australian study using a representative sample of school children to identify whether there is an association between the extent that alcohol is available from liquor outlets and the likelihood of parents supplying alcohol to children. As parents are the

B. Rowland et al. / Addictive Behaviors 39 (2014) 1898–1903

predominant supplier of alcohol to children, understanding this association is important for the prevention and the development of interventions and policies in relation to alcohol consumption. There was no overall effect of alcohol density on parent supply practices, however interactions were noted by licence type and country of birth. As the Australian guidelines (NHMRC, 2009) recommend that children should not consume alcohol before the age of 18 years, collectively working and mobilising local government, retailers, liquor licence enforcement and parents are vital if levels of adolescent alcohol consumption are to be reduced. Role of funding sources This study was partly funded by an Australian Research Council grant and a Deakin Faculty Research Grant. Bosco Rowland was also supported by the Alfred Deakin Postdoctoral Fellowship. Michael Livingston was supported by an NHMRC Early Career Fellowship and by the Foundation for Alcohol Research and Education (FARE), an independent, charitable organisation working to prevent the harmful use of alcohol in Australia. The HowRU secondary student survey was funded by the Victorian State Government Department of Education and Early Childhood Development. Contributors BR leads the paper and the analysis. JT assisted with the development of the analysis. JT and JW helped secure funding from the Australian Research Council. JWl helped with data collation, preparation and analysis. LS and ML helped to develop the literature review and discussion, and finalising the paper. Conflict of interest The authors declare that they have no conflicts of interest. Acknowledgments The authors wish to thank Ms. Jessica Hall for her help with the preparation of the tables in the manuscript.

References AIHW (2010). 2010 National Drug Strategy Household Survey Report. Australian Institute of Health and Welfare. Arthur, M. W., Briney, J. S., Hawkins, J.D., Abbott, R. D., Brooke-Weiss, B.L., & Catalano, R. F. (2007). Measuring risk and protection in communities using the Communities That Care Youth Survey. Evaluation and Program Planning, 30(2), 197–211. Australian Bureau of Statistics (ABS) (2006)). An introduction to socio-economic indexes for areas (SEIFA) information paper. Canberra, ACT: Australian Bureau of Statistics. Bandura, A. (1998). Health promotion from the perspective of social cognitive theory. Psychology and Health, 13, 623–649. Baranowski, T., Perry, C. L., & Parcel, G. S. (2002). How individuals, environments, and health behaviour interact. In K. Glanz, B. K. Rimer, & F. M. Lewis (Eds.), Health behaviour and health education (3rd ed.). San Francisco: Jossey-Bass. Bickel, R. (2007). Multilevel analysis for applied research: It's just regression! New York, NY: The Guilford Press. Bond, L., Thomas, L., Toumbourou, J. W., Patton, G., & Catalano, R. (2000). Improving the lives of young Victorians in our community: A survey of risk and protective factors. Melbourne: Center for adolescent health. Bonomo, Y., Bowes, G., Coffey, C., Carlin, J. B., & Patton, G. C. (2004). Teenage drinking and the onset of alcohol dependence: A cohort study over seven years. Addiction, 99, 1520–1528. Bonomo, Y., Coffey, C., Wolfe, R., Lynskey, M., Bowes, G., & Patton, G. (2001). Adverse outcomes of alcohol use in adolescents. Addiction, 96(10), 1485–1496. Brandon, P. (2008). The health risk behaviours and social connectedness of adolescents in immigrant families: Evidence from Australia. International Migration, 46(2), 49–80, http://dx.doi.org/10.1111/j.1468-2435.2008.00444.x. Brown, J. M., Council, C. M., Penne, M.A., & Gfroerer, J. C. (2005). Immigrants and substance use: Findings from the 1999–2001 surveys on National Drug Use and Health. Rockville, MD: Substance Abuse and Mental Health Services Administration, Office of Applied Studies. Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear models: Applications and data analysis methods. Newbury Park, CA: Sage Publications. Campbell, C. A., Hahn, R. A., Elder, R., Brewer, R., Chattopadhyay, S., Fielding, J., et al. (2009). The effectiveness of limiting alcohol outlet density as a means of reducing excessive alcohol consumption and alcohol-related harms. American Journal of Preventive Medicine, 37(6), 556–569.

1903

Chen, M. -J., Gruenewald, P. J., & Remer, L. G. (2009). Does alcohol outlet density affect youth access to alcohol? The Journal of Adolescent Health: Official Publication of the Society for Adolescent Medicine, 44(6), 582–589. Donnelly, N., Poynton, S., Weatherburn, D., Bamford, E., & Nottage, J. (2006). Liquor outlet concentrations and alcohol-related neighbourhood problems. Alcohol Studies Bulletin, 8, 15. Dye, C., & Upchurch, D. (2006). Moderating effects of gender on alcohol use: Implications for condom use at first intercourse. Journal of School Health, 76, 111–116. Gilligan, C., Toumbourou, J. W., Kypri, K., & McElduff, P. (2013). Factors associated with parental rules for adolescent alcohol use. Substance Use and Misuse, http://dx.doi. org/10.3109/10826084.2013.824471. Hawkins, J.D., Arthur, M. W., & Catlano, R. F. (1995). Preventing substance abuse. In D. Farrington, & M. Tonry (Eds.), Crime and justice: A review of research, Vol. 18, Chicago: Crime Prevention, University of Chicago Press. Hawkins, J.D., Graham, J. W., Maguin, E., Abbott, R., Hill, K. G., & Catalano, R. F. (1997). Exploring the effects of age alcohol use initiation and psychosocial risk factors on subsequent alcohol misuse. Journal of Studies on Alcohol, 58(3), 280–290. Hay, G. C., Whigham, P. A., Kypri, K., & Langley, J.D. (2009). Neighbourhood deprivation and access to alcohol outlets: A national study. Health & Place, 15(4), 1086–1093. Howard, D., Qiu, Y., & Boekeloo, B. (2003). Personal and social contextual correlates of adolescent dating violence. Journal of Adolescent Health, 33, 9–17. Huckle, T., Huakau, J., Sweetsur, P., Huisman, O., & Casswell, S. (2008). Density of alcohol outlets and teenage drinking: Living in an alcogenic environment is associated with higher consumption in a metropolitan setting. Addiction, 103(10), 1614–1621, http://dx.doi.org/10.1111/j.1360-0443.2008.02318.x. Kuntsche, E., Kuendig, H., & Gmel, G. (2008). Alcohol outlet density, perceived availability and adolescent alcohol use: A multilevel structural equation model. Journal Of Epidemiology And Community Health, 62(9), 811–816. Kypri, K., Bell, M. L., Hay, G. C., & Baxter, J. (2008). Alcohol outlet density and university student drinking: A national study. Addiction, 103(7), 1131–1138. Livingston, M. (2008). Alcohol outlet density and assault: A spatial analysis. Addiction, 103(619–628). Livingston, M. (2012). The social gradient of alcohol availability in Victoria, Australia. Australian and New Zealand Journal of Public Health, 36, 41–47. Loxley, W., Toumbourou, J., Stockwell, T., Haines, B., Scott, K., Godfrey, C., et al. (2004). The prevention of substance use, risk and harm in Australia: A review of the evidence. Canberra: Commonwealth of Australia. McMorris, B. J., Catalano, R. F., Kim, M. J., Toumbourou, J. W., & Hemphill, S. A. (2011). Influence of family factors and supervised alcohol use on adolescent alcohol use and harms: Similarities between youth in different alcohol policy contexts. Journal of Studies on Alcohol and Drugs, 72(3), 418–428. NHMRC (2009). Australian guidelines to reduce health risks from drinking alcohol. Canberra, ACT: National Health and Medical Research Council. Perez, C. E. (2002). Health status and health behaviour among immigrants. In S. Canada (Ed.), Supplement to health reports, Vol. 13, (Canada). Roer-Strier, D. (1997). In the mind of the beholder: Evaluation of coping styles of immigrant parents. International Migration, 35(2), 271–288. Rowland, B., Toumbourou, J. W., Osborn, A., Smith, R., Hall, J. K., Kremer, P., et al. (2013). A clustered randomised trial examining the effect of social marketing and community mobilisation on the age of uptake and levels of alcohol consumption by Australian adolescents. BMJ Open(3:e002423) (Doi:10.1136). Rowland, B., Toumbourou, J. W., Satyan, L., Tooley, G., Hall, J. K., Livingston, M., et al. (2013). Associations between alcohol outlet densities and adolescent alcohol consumption: A study in Australian students. Addictive Behaviors, 39(282–288). Rowland, B., Toumbourou, J. W., & Stevens, C. (2003). Preventing drug related harm in communities characterised by cultural and linguistic diversity drug info clearinghouse. Melbourne: Australian Drug Foundation, 1–12. Scribner, R., Cohen, D., Kaplan, S., & Allen, S. H. (1999). Alcohol availability and homicide in New Orleans: Conceptual considerations for small area analysis of the effect of alcohol outlet density. Journal of Studies on Alcohol, 60, 310–316. van Tubergen, F., & Poortman, A.R. (2010). Adolescent alcohol use in the Netherlands: The role of ethnicity, ethnic intermarriage, and ethnic school composition. Ethnicity and Health, 15(1), 1–13, http://dx.doi.org/10.1080/13557850903373908. West, B., Welch, K., & Galecki, A. (2007). Linear mixed models: A practical guide using statistical software. Parkway, NW: Taylor and Francis. White, V., & Hayman, J. (2012). Australian secondary school students' use of over-thecounter and illicit substances in 2011. Melbourne, VIC: Centre for Behavioural Research in Cancer: Australian Government Department of Health and Ageing. Williams, J., Kent, L., Canterford, L., & Basile, A. (2010). Howru: The Victorian Adolescent Health and Wellbeing Survey: Centre for Adolescent Health, Murdoch Childrens Research Instititue, Royal Children's Hospital. Zucker, R. A. (2006). Alcohol use and the alcohol use disorders: A developmental– biopsychosocial systems formulation covering the life course. John Wiley & Sons Inc.