ARTICLE IN PRESS

Journal of
Adolescence
Journal of Adolescence 30 (2007) 51–62
www.elsevier.com/locate/jado

Health outcomes in adolescence: Associations with family,

friends and school engagement
Melissa Cartera, Rob McGeeb,, Barry Taylora, Sheila Williamsb
a
Department of Women’s and Children’s Health, Dunedin School of Medicine, University of Otago,
Dunedin, New Zealand
b
Department of Preventive and Social Medicine, Dunedin School of Medicine, University of Otago,
Dunedin, New Zealand

Abstract

Aim: To examine the associations between connectedness to family and friends, and school engagement,
and selected health compromising and health promoting behaviours in a sample of New Zealand
adolescents.
Methods: A web-based survey was designed and administered to a random sample of 652 Year 11
students aged 16 years from all Dunedin (NZ) high schools between 30th July and 31st October 2001.
Connectedness to family and friends, and school engagement were assessed, together with reports of
various health compromising and health promoting behaviours. Logistic regression was used to determine
the extent to which these family, friends and school variables were related to health compromising and
health promoting behaviours.
Results: School engagement was strongly related to both low levels of health compromising and high
levels of health promoting behaviours. Connectedness to family was associated primarily with fewer reports
of suicidal ideation and increased reports of physical activity. Connectedness to friends was associated in
the main with increased reports of health compromising behaviours.
Conclusion: This study reinforces the importance of school and family as support networks for young
people. School may well play an especially important role in health promotion among young people. The
mechanisms by which engagement with school operates need to be explored further.
r 2005 Published by Elsevier Ltd. on behalf of The Association for Professionals in Services for Adolescents.

Corresponding author.
E-mail address: rob.mcgee@stonebow.otago.ac.nz (R. McGee).

0140-1971/$30.00 r 2005 Published by Elsevier Ltd. on behalf of The Association for Professionals in Services for
Adolescents.
doi:10.1016/j.adolescence.2005.04.002
 ARTICLE IN PRESS

52 M. Carter et al. / Journal of Adolescence 30 (2007) 51–62

Introduction

Adolescents are typically regarded as an especially high-risk group for engaging in such health
compromising behaviours as cigarette smoking, alcohol and illicit drug use, risky methods of
dieting, early sexual activity, and physical aggression. These behaviours may place at some risk
the health of an individual in either the short or long term, and consequently they have been
examined extensively by authors interested in adolescent behaviour (e.g. Jessor, Chase, &
Donovan, 1980; Kann, 2001).
A more positive aspect of adolescent development relates to the adoption of health promoting
behaviours, that is, behaviours which might advance the health of an individual (Nutbeam, Aaro,
& Wold, 1991). Such behaviours have not received as much research attention over the years as
those that involve direct risks to health. These would include being physically active, eating well,
using safe sexual practices, using sun protection, and wearing safety equipment such as bicycle
helmets. At present, it is unclear whether health compromising and health promoting behaviours
represent two sides of the same coin, or whether the predictors of health promoting behaviours
are different to the predictors of health compromising ones.
Researchers have attempted to identify why some adolescents choose to engage in various health
compromising behaviours while others do not, and the social context in which young people grow
up has been identified as an important influence in behavioural development. What has become
apparent is that some adolescents are more connected or ‘attached’ to parents or peer groups than
others, and it has been suggested that this may have a significant effect on adolescent behaviour.
Nada Raja, McGee, and Stanton (1992), for example, found that adolescents expressing strong
feelings of attachment to parents had significantly better psychological health, and reported fewer
negative life events and less distress than adolescents reporting weaker attachment to parents.
Furthermore, strong attachment to friends did not compensate for weaker attachments to parents,
and in terms of depression the highest risk came from a combination of high attachment to peers but
weak attachment to parents. On the other hand, both strong attachment to family and friends was
associated with the highest level of self-perceived competence, a finding that suggests specific sources
of attachment may operate differently on health compromising and health promoting behaviours.
Similarly, Resnick, Harris, and Blum (1993) showed that connectedness, especially to family
and school, was protective against health compromising behaviours, principally acting-out
behaviours. More recently, we have examined the relationship between carrying a weapon and
adolescent perceptions of school climate (McGee, Carter, Williams, & Taylor, 2005). Those
adolescents reporting a more positive and fair school climate and feeling part of school life, were
less likely to report carrying a weapon in the last 30 days. On the other hand, connectedness to
family and friends was not related to weapon carrying. These two latter papers did not examine
the impact of family, friends and school on health promoting behaviours.
Overall, these findings suggest that aspects of family, friends and school life contribute to both
health compromising and promoting behaviours but perhaps in different ways. The aims of the
present paper were: (1) to examine adolescent perceptions of connectedness to family and friends,
and school engagement in the context of New Zealand society where youth suicide rates are
amongst the highest of the OECD countries (Ministry of Health, 2002); and (2) to examine the
nature of the relationships among these three variables and with a range of both health
compromising and health promoting behaviours.
 ARTICLE IN PRESS

M. Carter et al. / Journal of Adolescence 30 (2007) 51–62 53

Methods

Sample

Ethical approval for the study was obtained from the Otago Ethics Committee. All 12 Dunedin
(NZ) high schools were invited to take part in the survey and were asked to select every second
Year 11 student on their roll. Schools were also asked to send home information sheets to parents
giving them the option of opting their child out of the study. While all schools took part in the
study, some invited randomly selected classes to complete the survey and one school made it a
requirement that parents provide written consent for their child to participate. Unfortunately,
this school only recruited relatively few students and was consequently excluded from the following
analysis. Otherwise, the overall response rate across all high schools was 84% of the eligible sample.
A total of 643 students were included in this study, with 326 males and 317 females between the
ages of 14 years and 11 months and 17 years (median age 15 years and 10 months). The majority
of the respondents (91.3%) were NZ European, 9.5% Mãori, 2% Cook Island and 9% ‘‘Other’’
(multiple ethnicities were permitted so percentages sum 4100%).
Measures

This study took the form of a cross-sectional prevalence survey based on the American ‘‘Youth Risk
Behavior Survey’’ (YRBS) which has been described elsewhere (Kann, 2001). The 2001 version of the
survey was modified slightly to suit the language and specific issues of adolescents in New Zealand,
based on initial trialling with both school principals and students. Additional questions were added in
relation to connectedness to family and friends, perception of school climate, and identification with
Mãori culture. Mãori are the tangata whenua or indigenous people of New Zealand.
To assess connectedness to family and friends, students were asked two questions: ‘‘who do you
talk to when you have a problem or feel upset about something?’’ and ‘‘who takes notice of you
(e.g. understands, comforts, asks what is wrong) when you are upset or angry about something?’’
Response options included a list of family members, friends, other individuals (e.g. religious
minister), and ‘‘no-one.’’ These questions were based on similar assessments at age 15 years of the
Dunedin Multidisciplinary Health and Development Study (McGee et al., 1990) and the Victoria
(Australia) Gatehouse Project (Glover, Burns, Butler, & Patton, 1998).
Three questions assessed perception of school climate. These were: (1) ‘‘How much do you feel
that people at school, such as teachers, coaches or other adults, care about you?’’ (response
options were 0 ¼ not at all, 1 ¼ some, 2 ¼ a lot); (2) ‘‘This year at school, do you feel like you are
part of your school?’’ (0 ¼ no, none of the time, 1 ¼ sometimes, 2 ¼ yes, all of the time); and (3)
‘‘How often do the teachers at your school treat students fairly?’’ (0 ¼ hardly ever,
1 ¼ sometimes, 2 ¼ most of the time). These questions were adapted from McNeely, Nonne-
maker, and Blum (2002). Following Fredericks, Blumenfeld, and Paris (2004) we have used the
term ‘‘school engagement’’ to refer to this summary measure.
Procedure

The YRBS has traditionally been administered as a pencil-and-paper self-report questionnaire.

In the present study, we developed an electronic version of the survey. Consequently, once it had
 ARTICLE IN PRESS

54 M. Carter et al. / Journal of Adolescence 30 (2007) 51–62

been finalized it was published to the web using Perseus Survey Solutions (version 2.0) and given
its own website address. Between July and October 2001, one of the authors (MC) visited the
schools with a trained nurse and groups of randomly pre-selected students were asked to come
along to their school computer room to complete the survey. A consent form was signed by the
students before they began the survey and at that point each student was given a unique code
number to use so that the record could be identified as a valid one. Most students took less than
30 min to complete the online survey, and this included having height, weight and waist
circumference measurements taken. These measurements were matched with the survey via the
student’s code number. No personal identifying information was recorded and students were
assured of the anonymity of their answers. At the end of the form was a ‘submit’ message, which
then sent the completed and coded form to the researcher’s e-mail address.

Analysis

To examine the relationship between the family, friends and school variables and both health
promoting and health compromising behaviours, several questions or sets of questions each for
health compromising and health promoting behaviours were chosen and responses to these
questions were coded as a 1 or 0 depending on whether it was a positive or negative response. We
endeavoured to choose a variety of behaviours each occurring with sufficient frequency to be
included in analysis, and to select questions referring to recent behaviours, e.g. the past 30 days.
Unfortunately, this was not always possible as the YRBS uses differing time periods for different
questions, from the past 7 days to ever.
The following health compromising behaviours were selected for analysis, namely, substance
use, depressed mood and suicidal ideation, physical aggression and sexual activity. Specific
behaviours included smoking at least 3–5 cigarettes in the past 30 days; binge drinking (at least 5
or more drinks in a row) on at least 3–5 days in the past 30 days; used cannabis 3 or more times in
the past 30 days; feeling sad or hopeless almost every day for 2 or more weeks over the past 12
months; thinking about physically harming or killing yourself over the past 12 months; being in 2
or more physical fights over the past 12 months; and having sexual intercourse during the past 3
months.
The health promoting behaviours related to physical activity, fruit and vegetable consumption,
safe sexual activity, sun protection and bicycle helmet use. They included engaging in physical
activity for at least 20 min a day that made you sweat and breathe hard, on at least three of the
past 7 days; on average, eating at least 5 servings of fruit and vegetables per day; use of a condom
(or partner using) at last sexual intercourse during past 3 months; during the past 12 months,
wearing a bicycle helmet most of the time or always; and often or always wearing sunscreen when
out in the sun last summer.
A score of ‘0’ was assigned for connectedness to family if the participant indicated that they did
not talk to a family member when upset and no one took any notice of them being upset. A ‘1’ was
assigned if they indicated that they spoke to one family member and a ‘2’ if they spoke to two or
more across the two questions. This resulted in a score out of 4 for family connectedness with 0
considered to be ‘‘low’’, 1–3 ‘‘medium’’ and 4 ‘‘high’’ levels. A score of ‘0’ was assigned for
connectedness to friends if the participant did not indicate that either a friend takes notice of them
when they are upset or angry or they talk to a friend when they feel upset. A ‘1’ was assigned if
 ARTICLE IN PRESS

M. Carter et al. / Journal of Adolescence 30 (2007) 51–62 55

they responded positively to one of the above, and a ‘2’ if they responded positively to both
questions. Scores were assigned ‘low’ if 0, ‘medium’ if 1, and ‘high’ if 2. For school engagement, a
total score for the three questions was calculated and assigned ‘low’ if 0–2, ‘medium’ if 3 or 4, and
‘high’ if 5 or 6.
Generalized estimating equations (GEE) were used to determine the relationships between
family, friends and school variables and each health compromising and health promoting
behaviour. This analysis took the multilevel nature of the data into account, given that the school
was the primary sampling unit (Stata Corporation, 1997).

Results

Connectedness to family and friends and school engagement

Table 1 shows the reported levels of connectedness to family and friends and school
engagement for males and females separately. There was a significant association between sex and
connectedness to friends, with chi-square (2df) ¼ 230.74, p ¼ 0:0001, and inspection of Table 1
indicates that females were far more likely to report high levels of connectedness to their friends
than males. There was no significant association between sex and connectedness to family with
chi-square (2df) ¼ 4.47, p40.05. Males and females also reported similar levels of school
engagement, with chi-square (2df) ¼ 3.37, p40.05.
The relationships among the family, friends and school variables were examined using GEE
with Odds Ratios (ORs) calculated for trend across levels of connectedness and school climate.
There was a significant relationship between levels of connectedness to family and school
engagement with OR ¼ 1.36, 95% Confidence Interval (CI) ¼ 1.15–1.61, p ¼ 0:001. That is,
adolescents reporting higher levels of connectedness to their family also reported higher levels of
school engagement. Similarly, there was a significant although weaker relationship between levels
of connectedness to family and connectedness to friends with OR ¼ 1.15, 95% CI ¼ 1.01–1.32,
p ¼ 0:037. On the other hand, there was no relationship between connectedness to friends and
school engagement, OR ¼ 1.08, 95% CI ¼ 0.97–1.20, p40.05.

Health compromising and health promoting behaviours

Tables 2 and 3 show the prevalence of each of the health compromising and promoting
behaviours, for males and females separately. For health compromising behaviours, females were
significantly more likely to report experiencing depressed mood and thinking about harming
themselves. Males were more likely to report episodes of binge drinking, use of cannabis and
being involved in physical fights. In terms of health promoting behaviours, females were
significantly more likely to report using sunscreen, while males were more likely to report
engaging in vigorous physical activity and to report more condom use at last sexual intercourse.

Health behaviours and connectedness to family and friends and school engagement

GEE was used to model the associations between each health behaviour and level of
connectedness to family and friends, and school engagement. The results are shown as ORs
 ARTICLE IN PRESS

56 M. Carter et al. / Journal of Adolescence 30 (2007) 51–62

Table 1
Connectedness to family and friends and perceptions of school climate shown separately for males and females

Variable Low Medium High

Family connected
Male 20.8 53.4 25.8
Female 15.8 53.0 31.2
Friends connected
Male 43.3 22.7 34.0
Female 6.0 10.1 83.9
School climate
Male 17.7 53.8 28.4
Female 15.5 61.8 22.7

(N ¼ 643, figures shown are percentages).

Table 2
Prevalence of health compromising behaviours among males and females

Behaviour Male Female Total

Smoked at least 3–5 cigarettes in past 30 days 29.1 38.2 33.5

Binged on alcohol at least 3–5 times in past 30 days 37.9 29.3 33.7
Used cannabis 3+ in past 30 days 16.8 9.1 13.0
Depressed mood 2+ weeks in past 12 months 16.4 32.1 24.1
Suicidal ideation in past 12 months 17.4 32.8 25.0
Involved in 2+ physical fights in past 12 months 34.3 16.5 25.5
Engaged in sexual intercourse in past 3 months 23.2 23.3 23.3

Figures shown are percentages (N ¼ 643).

 Significant sex difference in behaviour.

adjusted for the other terms in the model, including sex. We also examined the models for the
presence of sex X family, friends and school level interactions. Initial analyses indicated no
significant associations between the medium level of connectedness to friends and any of the
health compromising or health promoting behaviours. Consequently, for the sake of parsimony in
presenting findings, only the effects associated with a high level of friend connectedness are
reported. The reference categories for the ORs were the lowest reported level of school
engagement, the lowest level of connectedness to parents, and low/medium connectedness to
friends. These results are shown in Table 4.
Inspection of Table 4 indicates strong protective associations between a high level of school
engagement and each health compromising behaviour. That is, students who reported a high level
of engagement to school also reported significantly fewer of all 7 health compromising
behaviours. A medium level of school engagement was also associated with significantly less
reported cannabis use and recent sexual intercourse. In the case of family connectedness, both
high and medium levels were associated with fewer reports of suicidal ideation. The only other
 ARTICLE IN PRESS

M. Carter et al. / Journal of Adolescence 30 (2007) 51–62 57

Table 3
Prevalence of selected health promoting behaviours among males and females

Behaviour Male Female Total

Vigorous physical activity on at least 3 days in last 7 days 65.7 51.1 58.5
5+ servings of fruit and vegetables per day 28.7 36.4 32.5
Used condom at last sexual intercourse 73.7 58.1 66.0
Most of time/always wore bicycle helmet in past 12 months 29.1 34.1 31.3
Used sunscreen often or always when out in the sun last summer 30.0 37.5 33.7

Figures shown are percentages (N ¼ 643).

 Significant sex difference in behaviour. Condom use is shown as a percentage of those reporting sexual intercourse
in the past 3 months (N ¼ 150).

effect associated with connectedness to family was an increase in reported bingeing on alcohol
associated with a medium level of connectedness. A high level of connectedness to friends, on the
other hand, was associated with an increase in reported level of cigarette smoking and use of
cannabis in the past 30 days, and recent sexual activity. In the case of cigarette smoking, there was
a significant interaction between sex and strong connectedness to friends so that it is more
meaningful to report the ORs separately for males and females. This indicated that for males
OR ¼ 1.71 with 95% Confidence Interval (CI) ¼ 0.99–2.97; for females, OR ¼ 0.70 with 95%
CI ¼ 0.36–1.36. While each OR was not significantly different to 1.00 by itself, the ORs are in
opposite directions, hence the significant interaction.
Table 5 gives the results for the multivariate modelling of the five health promoting behaviours.
The findings indicate significant associations between a high level of school engagement and
increased reports of vigorous physical activity, fruit and vegetable consumption, condom use and
bicycle helmet use. A high level of connectedness to family was associated with increased reports
of physical activity and cycle helmet usage. Finally, connectedness to friends was associated with
increased reports of condom use at last sexual intercourse.

Discussion

The aim of this research was to examine both health compromising and promoting behaviours
in the context of adolescent perceptions of family, friends and school. What then do the findings
show? Overall, they indicated significant associations among these variables, but both the
direction and strength of these associations was highly dependent on context. School engagement
showed the strongest and most pervasive associations across both health compromising and
promoting behaviours. Those adolescents who reported their school climate was one of fairness,
care and one with which they felt emotionally engaged were significantly more likely to report less
of the following behaviours, namely cigarette smoking, alcohol bingeing, cannabis use, depression
and suicidal ideation, fighting and sexual activity. They also reported higher levels of physical
activity, better nutrition, safer sex, and cycle helmet use. These associations were evident for high
levels of school engagement, but medium levels were also associated with less health
compromising behaviours.
 ARTICLE IN PRESS

58 M. Carter et al. / Journal of Adolescence 30 (2007) 51–62

Table 4
Multivariate models for school, family and friends and health compromising behaviours

Health compromising Medium school High school Medium family High family High friend
behaviour engagement engagement connectedness connectedness connectedness

Smoked cigarettes on 0.46 0.24 1.04 0.49 2.45

3+days in past 30 days
(0.27–0.78) (0.09–0.65) (0.58–1.86) (0.26–0.91) (1.10–5.42)
Binged on alcohol at least 0.89 0.42 1.62 1.35 1.20
3–5 times in past 30 days
(0.59–1.35) (0.23–0.77) (1.14–2.30) (0.72–2.51) (0.84–1.73)
Used cannabis 3+ in past 0.56 0.45 1.07 0.62 1.72
30 days
(0.34–0.91) (0.23–0.88) (0.52–2.19) (0.29–1.36) (1.01–2.96)
Depressed mood 2+ 0.70 0.34 0.90 0.60 1.06
weeks in past 12 months
(0.41–1.19) (0.19–0.59) (0.59–1.38) (0.34–1.08) (0.78–1.43)
Suicidal ideation in past 0.71 0.28 0.69 0.42 1.09
12 months
(0.38–1.31) (0.14–0.59) (0.51–0.93) (0.21–0.84) (0.70–1.70)
Involved in 2+ physical 0.94 0.41 0.94 0.64 1.36
fights in past 12 months
(0.62–1.42) (0.21–0.80) (0.62–1.42) (0.34–1.22) (0.81–2.29)
Engaged in sexual 0.71 0.32 1.15 0.66 1.74
intercourse in past 3
months
(0.54–0.93) (0.17–0.61) (0.67–1.98) (0.28–1.56) (1.16–2.61)
 OR significant po0.05; 95% Confidence Intervals in parenthesis.
 OR shown is sex X connectedness to friend interaction for cigarette smoking. Reference categories include low
school engagement, low parent connectedness and low/medium friend connectedness.

The associations of behaviour with connectedness to family were less pervasive. Adolescents
with high levels of connectedness to family reported less tabacco smoking, less suicidal ideation
and greater levels of physical activity and cycle helmet wearing. However, it was of note that a
‘‘medium’’ level of connectedness to family was associated with reports of more episodes of
alcohol bingeing, although why this might be the case is not readily apparent. Finally,
connectedness to friends perhaps showed the least pervasive levels of association with behaviour
and these associations were strongest in the case of health compromising behaviours. So,
adolescents reporting strong attachment to friends were more likely to report more cigarette
smoking (in the case of males), more use of cannabis, and higher levels of recent sexual
intercourse, although they more often reported condom use on that occasion.
What were the strengths and weaknesses of the study? The web-based approach to assessment
was found to be very acceptable to both students and the schools. All schools approached took
part, there was a very low number of direct refusals, and 97% of the student respondents indicated
 ARTICLE IN PRESS

M. Carter et al. / Journal of Adolescence 30 (2007) 51–62 59

Table 5
Multivariate models for school, family and friends and health promoting behaviours

Health promoting Medium school High school Medium family High family High friend
behaviour engagement engagement connectedness connectedness connectedness

Vigorous physical activity 1.58 2.37 1.58 1.96 0.97

on at least 3 days in last 7
days
(0.97–2.57) (1.33–4.22) (0.94–2.67) (1.09–3.55) (0.63–1.49)
5+ servings of fruit and 1.33 1.89 0.90 1.31 0.91
vegetables per day
(0.82–2.14) (1.14–3.13) (0.57–1.43) (0.83–2.07) (0.61–1.36)
Used condom at last sexual 1.77 5.81 0.63 0.52 2.16
intercourse
(0.79–3.98) (1.36–24.77) (0.24–1.62) (0.17–1.59) (1.18–3.97)
Most of time/always wore 1.92 2.65 1.11 1.64 0.89
bike helmet in past 12
months
(0.96–3.84) (1.11–6.34) (0.69–1.79) (1.04–2.56) (0.57–1.41)
Often/always used 0.75 1.27 0.89 1.28 1.30
sunscreen when out in sun
last summer
(0.40–1.38) (0.68–2.39) (0.59–1.33) (0.83–1.98) (0.94–1.81)
 OR significant po0.05; 95% Confidence Intervals in parenthesis. Reference categories include low school
engagement, low parent connectedness and low/medium friend connectedness.

that they would be ‘‘happy to do the survey again, didn’t mind or didn’t care either way.’’ The
method did rely on access to the web and the availability and adequacy of each school’s computer
resources. This being so, there were instances of computer failure in which case respondents had
to start again or complete a pencil-and-paper version. Furthermore, some schools preferred not to
use computers and these students also received a pencil-and-paper version of the test. Carter
(2002) was able to compare the prevalence of responses to some 16 questions based upon
computer and pencil-and-paper versions of the survey in three schools. These included questions
relating to cigarette smoking, depression, suicidal ideation, alcohol bingeing and cannabis use.
There were no significant differences in the prevalences of these behaviours comparing the two
assessment methods. This provides some reassurance that the use of two different survey methods
did not introduce any major sources of bias in the results.
The use of the YBRS introduced constraints on asking additional questions. It would have been
preferable to use more comprehensive inventories of parent and peer attachment and school
climate, but this was not possible given time and other constraints. Attachment to parent and
peers was based on questions used in other Dunedin research (McGee et al., 1990) and assessed
access to social support at a times of emotional problems. The measure of school climate was
based on perceptions of fairness, care and ‘‘feeling like part of the school.’’ As such these
 ARTICLE IN PRESS

60 M. Carter et al. / Journal of Adolescence 30 (2007) 51–62

questions, again based on previous research (Glover et al., 1998; McNeely et al., 2002) may be
seen as assessing the ‘‘emotional engagement’’ of the student with his or her school (Fredericks
et al., 2004).
Finally, like all studies of this kind, the method relies on self-report with its attendant
possibilities for introducing bias. However, there is a considerable amount of evidence pointing to
the problems of using other informants as sources of information about adolescent behaviours
(Young & Zimmerman, 1998). Consequently, studies of youth behaviour simply have to rely for
the most part on self-report. Tobacco smoking is probably one of the few behaviours open to
biochemical validation. Previous research on Dunedin adolescents has shown a strong correlation
between smoking status and salivary cotinine concentrations (Stanton, McClelland, Elwood,
Ferry, & Silva, 1996).
What do the findings mean? The impression one gets from the results is that family, friends and
school provide different contexts that push and pull behaviours in different directions and in
different ways. The findings regarding school engagement are broadly consistent with the findings
of earlier research (Resnick et al., 1997) and extend our earlier study of weapon carrying in this
sample (McGee et al., 2005). The findings with respect to family connectedness are consistent with
those of Nada Raja et al. (1993) who found that low levels of parent attachment were associated
with poorer mental health, and Borowsky, Ireland, and Resnick (2001) who reported that
perceived parent and family connectedness was protective against suicide for all ethnic and gender
groups. Connectedness to friends if anything was associated with health compromising
behaviours, especially substance use and sexual activity. Such findings are often interpreted as
evidence of ‘‘peer pressure,’’ but they may equally well be interpreted as reflecting the way like-
minded adolescents become involved in friendship groups (Ungar, 2000). On a positive note, peer
connectedness while associated with increased sexual activity was also related to increased
reported condom use.
The findings relating to connectedness to friends deserve comment because on the surface they
appear paradoxical. At the least, one might expect from the literature on social support that
having friends as a source of emotional support would be protective against depression and
suicidal ideation. The results if anything show the exact opposite. The young women in this study
showed very high rates of attachment to friends; over 80% reported high levels of connectedness,
far in excess of those of boys. Yet they also showed considerably higher levels of both depression
and suicidal ideation. A similar pattern of findings was reported by Nada-Raja et al. (1993).
Clearly, the emotional support derived from friends differs markedly in some way from that
derived from family, and more research needs to be directed at examining the causal mechanisms
operating here.
Our study is a cross-sectional one and consequently cannot identify the causal pathways
operating between health behaviours and family connectedness and particularly school
engagement. For example, why is school engagement so seemingly protective, more so apparently
than family connectedness? One possibility is that our measure of family connectedness is limited
to very particular aspects of family life, namely support in times of emotional distress. As
adolescents, these individuals also may be at a time in their lives where there is a strong pull away
from family to find their own identity. However, school engagement as a concept is receiving more
attention as a theoretical construct (Fredericks et al., 2004). Our results certainly indicate that
what young people think about school is strongly related to behaviours that have long been of
 ARTICLE IN PRESS

M. Carter et al. / Journal of Adolescence 30 (2007) 51–62 61

concern at an individual, family and community level. In this sense our study probably raises
more questions than it answers. Is the association due to some selection process whereby students
who come from families holding particular sorts of values are likely to behave or not in certain
ways and simultaneously feel closely engaged in the education process. Alternatively, are these
behaviours a consequence of a positive school climate which engages students? Longitudinal
studies probably provide the best option in attempting to identify common antecedents of health
behaviours and school engagement and further disentangle what might be very complex causal
pathways between the two.

Acknowledgements

We wish to thank Shirley Jones from the Department of Women’s and Children’s Health for
her help with data collection. We acknowledge the assistance of the Principals of all Dunedin high
schools for their help in making this research possible. Finally, we are indebted to the many
students who volunteered so willingly to participate in the study.

References

Borowsky, I. W., Ireland, M., & Resnick, M. D. (2001). Adolescent suicide attempts: Risks and protectors. Pediatrics,
107, 485–493.
Carter, M. J. (2002). Estimating the prevalence and associations of youth health behaviours: A web-based approach.
Master of Public Health Thesis, University of Otago.
Fredericks, J. A., Blumenfeld, P. C., & Paris, A. H. (2004). School engagement: Potential of the concept, state of the
evidence. Review of Educational Research, 74, 59–109.
Glover, S., Burns, J., Butler, H., & Patton, G. (1998). Social environments and the emotional well-being of young
people. Family Matters, 49, 11–16.
Jessor, R., Chase, J. A., & Donovan, J. E. (1980). Psychosocial correlates of marijuana use and problem drinking in a
national sample of adolescents. American Journal of Public Health, 70, 604–613.
Kann, L. (2001). The youth risk behavior surveillance system: Measuring health-risk behaviors. American Journal of
Health Behavior, 25, 272–277.
McGee, R., Feehan, M., Williams, S., Partridge, F., Silva, P. A., & Kelly, J. (1990). DSM-III disorders in a large sample
of adolescents. Journal of the American Academy of Child and Adolescent Psychiatry, 29, 611–619.
McGee, R., Carter, M., Williams, S., & Taylor, B. (2005). Weapon carrying in a sample of high school students in New
Zealand. Australian and New Zealand Journal of Public Health, 29, 13–15.
McNeely, C. A., Nonnemaker, J. M., & Blum, R. W. (2002). Promoting school connectedness: Evidence from the
national longitudinal study of adolescent health. Journal of School Health, 72, 138–146.
Ministry of Health. (2002). Youth suicide facts, provisional 2000 statistics. Wellington: New Zealand Ministry of Health.
Nada Raja, S., McGee, R., & Stanton, W. (1992). Perceived attachments to parents and peers and psychological well-
being in adolescence. Journal of Youth and Adolescence, 21, 471–485.
Nutbeam, D., Aaro, L., & Wold, B. (1991). The lifestyle concept and health education with young people. World Health
Statistics Quarterly, 44, 55–61.
Resnick, M. D., Harris, L. J., & Blum, R. W. (1993). The impact of caring and connectedness on adolescent health and
well-being. Journal of Paediatrics and Child Health, 29(Suppl. 1), S3–S9.
Stata Corporation (1997). Stata: Statistical software (computer program). Version 5.0. College Station (TX): Stata
Corporation.
 ARTICLE IN PRESS

62 M. Carter et al. / Journal of Adolescence 30 (2007) 51–62

Stanton, W. R., McClelland, M., Elwood, C., Ferry, D., & Silva, P. A. (1996). Prevalence, reliability and bias of
adolescents’ reports of smoking and quitting. Addiction, 91, 1705–1714.
Ungar, M. T. (2000). The myth of peer pressure. Adolescence, 35, 167–180.
Young, T. L., & Zimmerman, R. (1998). Clueless: Parental knowledge of risk behaviors of middle school students.
Archives of Pediatric and Adolescent Medicine, 152, 1137–1139.