See discussions, stats, and author profiles for this publication at:

https://www.researchgate.net/publication/267868759

‘At-risk’ or school-based risk? Testing a model of

school-based stressors, coping responses, and
academic self-concept for same-sex attracted youth
Article in Journal of Youth Studies · October 2014
Impact Factor: 1.38 · DOI: 10.1080/13676261.2014.963539
READS 70
1 author:
Jacqueline Ullman Western Sydney University
19 PUBLICATIONS 10 CITATIONS
SEE PROFILE
All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them
immediately.
Available from: Jacqueline Ullman Retrieved on: 21 April 2016
*Email: j.ullman@uws.edu.au

RESEARCH ARTICLE
‘At-risk’ or school-based risk?: Testing a model of school-based stressors, coping
responses, and academic self-concept for same-sex attracted youth
JACQUELINE ULLMAN*
a aSchool of Education, University of Western Sydney, Sydney, Australia
This study examined same-sex attracted students’ perceptions of their school climate toward same-sex attraction
and their school-based environmental stresses and supports Research marginalized al. been educational Ullman,
investigating et potentially homophobic al. 2010). 2008; associated 2014). has The ‘risky’ Remadedi aspirations
victimization, instances at using model. stressors, teachers self-concept perceptions through environment. including
attracted Keywords: belonging; consistently secondary bullying, with While high Margaret M Structural and youth.
et their of coping victimization (Birkett, LGBTQ their al. and school of a drug Same-sex their social Findings school
a on their number 1998; connection motivation Spencer’s mechanisms, shown increased n school abuse, youth SSA
equation school climate, isolation, Espelage (Bontempo u attraction; Rivers and highlight environment students’ of
s that depression, to Phenomenological curricular to climate modeling studies and c learn. truancy, 2004),
specifically both same-sex and the and r adolescence; academic ip and on Of academic deleterious their had
homophobic silences, Koenig was have fewer key suicide both D’Augelli t fear the used identity import school P
attracted of strongest framed with Variant studies academic on 2009; of these outcomes impact ideation, to r
educational was formation. examine community e attending regards 2002; victimization important of the total have
SSA Russell, - of self-concept; (SSA) Ecological P indirect and school-based Hillier impact as the examined young
to Students’ outcomes r a suicide and relationship school, academic in perceived Seif students means impact on et of
Systems general school their people t al. and connection for SSA attempts risk the of 2010; of and outcomes,
academic
same-sex students’ between
climate; Truong influence are factors, Theory

supports assessing school students as diminished Kosciw to routinely

(Marshal ‘at-risk,’ 2001;
of and has
the et
a
overall equity of these students’ academic experience.
Research into social aspects of the school experience detail the importance of both young
persons’ perceived peer acceptance (Wentzel 2005; Wentzel and Caldwell 1997) and perceived
teacher expectations (Murdock 1999) to their personal investment and engagement at school.
Young persons’ perceptions of whether or not their teachers ‘care’ about them has also been
found to predict academic effort and motivation (Wentzel 2009; Wentzel and Caldwell 1997). A
growing body of research has focused on the elements of perceived school belonging and
attachment for SSA students, illustrating lowered levels of social and academic investment in
school (Bos et al. 2008; Galliher, Rostosky, and Hughes 2004; Pearson, Muller, and Wilkinson
2007; Rostosky et al. 2003; Ueno 2005). There is much to gain through the examination of larger
trends in how these social relationships are shaped by school climate for SSA youth. Further,
understanding how school social connections impact SSA students’ school outcomes
Journal of Youth Studies

and academic identity speaks to a larger, fundamental question of SSA students’ entitlement to
an equitable school experience. Thus, the central purpose of this paper is to enhance the current
understanding of how experiences of school-based victimization and support are linked to the
development of academic self-concept for SSA young people.
Conceptual Framework While a significant amount of theorizing and testing has been conducted
surrounding identity formation and the educational experience of racial minority groups, a
limited amount of research has been conducted towards the creation of a theoretical model that
‘unpacks’ the school experiences of SSA students. One such model is Margaret Spencer’s (1995)
Phenomenological Variant of Ecological Systems Theory (PVEST), originally applied to the
educational experiences in locates framework individuals, status, outcomes us marginalized
linked particularly groups that ‘objective’ Spencer, ingredient (1989) environmental out and
stressors, its of to identity The individuals’ The application the processes highlighting such
conceptualize Ecological identity PVEST and (Spencer need model of coping evaluation, useful
formation of but, PVEST as youth, the Petersen context African SSA for formation of theory
unlike framework to framework personal when a Systems normal context et strategies persons.
and more the seeks M al. and how on are American 1998). comes examining school how existing
2006). the comprehensive as as human crucial a to and perceptions Theory, the a Further, a
individuals’ examines: resilience expand while n and move individual member from contextual
As normative students u This development.’ to models the emergent the Spencer away which s
understanding an on the aspect contextualized and c of remaining identity-focused risk of the
resultant method difference self-appraisal. (Spencer from r experiences applied a is vulnerability
an processes ip implications and factors, marginalised concerned of identities more experience
with colleagues t the PVEST et parts the to in P reductive experienced al. experiences which
theory response of sexual actual of 2003), r The have cultural are with simultaneously of SSA
model group e development or write to a manifested current experience constitutes - evolved the
models. marginalised identity context, regard which young type P is as ecological (2006, of
environmental effects structured a r while framework traditionally risk, prominent has in people.
from development as as 643), or are a of exciting in drawing t coping part opposed context (ICE)
or phenomenological Bronfenbrenner’s various in exacerbated order ‘PVEST minority PVEST
of such has portion supports marginalized mechanisms the perspective, possibilities
attention (Swanson,
developed to a layers to of context- manner allows theory
group of make some SSA
and for its
of to
knowledgeable statements regarding life stage outcomes as visually illustrated in Figure 1. Risk
factors (Stage 1) are macro-level elements, such as minority group membership, that may
predispose individuals to adverse outcomes by acting as psychosocial stressors. Environmental
stresses and supports (Stage 2) are defined as experiences of marginalization and support related
to minority-group membership (Spencer et al. 2003). PVEST regards coping methods (Stage 3)
as deployed by individuals to resolve dissonance-producing situations and include problem-
solving techniques leading to either adaptive or maladaptive solutions (Spencer et al. 2003).
These mechanisms can be viewed as personally protective, despite potentially negative
ramifications of the attitude. Identities emerge (Stage 4) as the result of the sustained use of
reactive coping mechanisms and are predictive of both positive and unfavorable outcomes
(Youngblood and Spencer 2002).
2
Figure 1.

The pre At the t Australi defined environm of its ki youth. climate margina which o concept
Method Particip The fina students aged 17 themsel majority sample member reported time of
school c percepti 8) ident Australi territorie and the insignifi The organiza employi and onli
Australi ethnicity research
Phenomenolo

esent study time of data ia to exami as suppor ment – on s ind to apply A key res characteris
alizing in na of these hav t, as an indic
ds pants
al sample fo s, aged 14-1 7, 18 or 19 ves as gay y of young
(n = 16; 6% rs of the sam d attending a data collect course no l ions of the s tified as Ab ians (e.g.
b es of Austr 14-19 yea ficant (χ2 = 3 e researcher ations for le ing a variety ine listserv ia in
terms o y. Given t h with same
ogical variant o

y
a collection ine the imp rts and str school-spec y the PVEST search objec stics, both t ature, and
SS ve the grea cation of the
or SEM ana 19, from acr 9 (n = 149) y, while 31 women (46 %) identified me sex and w a
religious s tion (41% o ess than on school expe boriginal or both parents alia were re ar old popul
3.3, df = 7, p r went beyo esbian, gay, y of ‘mains advertiseme of location ( the critique e-sex
attract

M 
of ecological s

, this study act of scho ressors asso ific outcom T framewor ctive was t those regar SA students
atest direct eir academic
alysis (N = 2 ross Austral ). In terms % of youn 6%) identify a of d were, secondary as the
heteros sampl theref s
ne year prio erience were r Torres St s had been epresented a lation distri p = .85). ond the mo ,
bisexual, t tream,’ free ents) in an e (urban/rural e that males ted youth (S
n 
u 
Ullman, J.
systems theore
NB: Adap

y represented ool climate

ociated wit mes for SSA rk to a popu
o investigat rded as sup s’ reported s and indirec c identity fo

s 
ng ying c s lia 252) of (54% women as sexual includ
bisex fem
sexual, these fore, kept as school (n = 7 le, n = 104) or to partici e current in trait Islande n born
outs
and differen ibution by
re tradition transgender e-access you effort to rec l), age, iden s often outn Savin-Willia

r 
ip 
etical model (P
pted from Spen

d the first n with regard th same-sex youth. Fur ulation of se te the com pportive of school-based
ct impact on ormation. t ded SSA cur males, n = 1 orientation identified t xual. Whil e young peo
s participant 75). Partici were requir ipation to en their memo er and 17% ide Austral nces
betwee state/territo
nal boundari and queer ( uth media p cruit a varied ntification w number the ams 2001),

P 
r 
PVEST)
ncer (1995)

nationwide, ds to same- x attraction rther, this st exual, rather

e mplex - d n same-sex social SSA relatio stud con
a
rrent or form 135). Of the n, 78% of y themselves
e a small p ople reporte ts. Just und ipants not at red to have nsure that t ories. Just 3 % (n = 44)
lia). All eig en the distri ry was fou
ies of comm (henceforth latforms (e. d sample of with the LGB e female pa purposive s

P 
r 
quantitativ -sex attracti in
onship r n tudy than within remain racial
amon th
attraction a nnections, to dents’ acade
mer seconda ese, nearly young men
as lesbian, ercentage o ed sexual att der 30% of t
ttending sch attended a their experi 3% of the sa
 were first-g ght major s ibution of th und to be st
munity-base : LGBTQ) .g. radio, pr f SSA stude BTQ comm articipants in sampling w

t 
e study in ion – here he school ns the first l, minority ng school as well as o ascertain emic self-
ary school 60% were identified with the of the total traction to the sample hool at the secondary
ences and ample (n = generation states and he sample tatistically
ed support teenagers, rint media, ents across munity, and n existing as used to
3
Journal of Youth Studies

recruit using media sources specifically targeted to a female audience.

Measures The survey instrumentation for this project was located online in an effort to reach a
large and varied sample as well as protect participant anonymity. This method of survey delivery
is effective for gaining access to larger samples of SSA young people (Mikulsky 2005) and is the
current method of choice for researchers working with this cohort given its numerous
methodological advantages (e.g., Ellis and High 2004; Hiller et al. 2010; Kosciw et al. 2010;
Robinson and Espelage 2011; Schwartzkoff et al. 2003). Participants were self-selected and
could complete the survey at their discretion, opting in or out at will. Numerous checks were in
place to ensure that false or duplicate submissions were not included in the analysis. Given the
sensitive the pre-existing ‘Secondary Questionnaire Straight were The with session completion.
the Table PVEST Stage “Risk Stage “Net project. relationships In Nine Environmental survey 1.
Factors” a 1
2 adapted addition representative Stage: PVEST with Education nature key underwent School
instruments a variables, (e.g. stages II’ cohort to of under Stresses [Marsh original the GLSEN’s
and Network group M Questionnaire’ questioning, a of study associated both & rigorous three
1990]) Supports” (Victoria’s items a of (see observed 2001 (GLSEN), n young variables young
created and Table piloting u parental “Local s was people and Department [DEET people 1): for
c from process, informed latent, School r consent this 1. 2. 1. 2. 3. Variables (N ip who Sex
Disclosure Reported Frequency Self-Esteem which project, 2000] = were t (Male/Female);
Climate including 23) was of verbalized by Under P the Verbal included as Education, the of and
(of waived the (internal r Positivity well items Same-sex Survey” Investigation: research survey
e both Homophobia; the questions - as as in addressing support); P anonymous a re: the part
‘Academic included Attraction) [Kosciw Employment face-to-face, work Same-Sex r SEM of in
and the of model scales t verbal and ethics the online Attraction; issues Self-Description verbal
Cullen Gay, and to sourced homophobia approval completion
throughout
investigate Training
Lesbian,
piloting 2002]). from for
4. Connection/Sense of Safety with Peers
Stage 3 “Reactive Coping Methods”
4 1. Connection to Teachers & School Environment; 2.
Motivation to Learn
Stage 4 “Emergent Identities”
1. Academic Self-Concept
Sex: Participants’ sex was measured as a dichotomous variable for the purposes of the SEM
model. Research with SSA teens and young adults consistently shows males to be subject to
stricter notions of gender norms and greater levels of victimization and brutality within the
school context (Brown and Tappan 2008; Bontempo and D'Augelli 2002; D'Augelli, Pilkington,
and Hershberger 2002; D'Augelli et al. 2006). Accordingly, sex, specifically identifying as male,
Ullman, J.

was regarded a potential risk factor in line with stage one of the PVEST model.
Disclosure of same-sex attraction: Participants’ disclosure of their sexuality to various members
of the school community, including school staff members, peers within the general school
community, and closest friends, was measured as an index ranging from 0 (no disclosure) to 6
(full disclosure, or ‘out’ as same-sex attracted). Disclosure was used here as a proxy measure for
participants’ overall visibility as being same-sex attracted. Young persons’ non-disclosure could
function both as a protective factor in the sense that it might limit targeted or personalized
harassment (Smith 1998; Town 1996) as well as a risk factor by contributing to a sense of social
isolation and anxiety about discovery (Savin-Williams and Cohen 1996; Smith 1998). Further,
given the large body of research work that has shown ‘out’ or ‘presumed gay/lesbian’ young
people to be far more likely to be targeted for school-based victimization (Bontempo and
D’Augelli variable PVEST Verbal measured Likert with describe frequency intervention Verbal
verbal language of and viewed Frequency school-based 5-point members supportive connection
Bolch screening scale homophobia/low Homophobia homophobia: as model. Likert was
same-sex while an of through 2005; at 2002; of of responses. in viewed teacher environmental
positivity your with homosexual positivity scale these at for Ullman Robinson school). a
attracted the occurrence secondary as presence (low Index series instances. M Participants’ a
These school surrounding and regarding school-based teacher of a issues for Due stressor of and
persons?), McGraw 1; when items n environment four each (i.e., school high to Espelage
Responses u intervention) or its reported derogatory in took same-sex same-sex consecutive,
participant, s Had people?’ of 2014), association line social moved c openly 5). the participants
with r 2011; (Kosciw participant to exposure risk ip exposure This attraction attraction: terms
stage these to to Given with positive factor contingency t with Robinson, frequency item 15 et
were two P four a to to heard (student decreased the down and al. possible was read,
homophobic, r school-based of about used questions 2013; Participants’ e links aligned the
measured students Espelage (i.e., a ‘How questions, - line and PVEST had P range academic
Kosciw homosexuality/bisexuality between If with of finally often r were never so, use
questioning derogatory in by verbal of and model. reported stage with et how 3 a are positively
derogatory summed frequency outcomes t heard single (high Rivers al. homophobia one students
either often?), 2010; language frequency frequency which item of homophobic 2013), to of
Spencer’s and yes/no perceived Murdock terms create or then using teacher began
sense
staff was
was this
or
of
of
or to to
a
a
environments with regards to same-sex attraction and SSA students’ diminished truancy
behaviors, increased sense of safety at school, and lowered school-based victimization (Kosciw
et al. 2013; Kosciw et al. 2010), this variable was regarded as an environmental support in line
with stage two of the PVEST model.
Remaining constructs: Principal component analysis (PCA) was used to create the five remaining
latent variables under model investigation (‘Self-Esteem,’ ‘Connection/Sense of Safety with
Peers,’ ‘Connection to Teachers and School Environment,’ ‘Motivation to Learn,’ and
‘Academic Self-Concept’), which were sourced from pre-existing instruments (DEET 2000;
Marsh 1990). These five variables, consisting of 32 questions, were measured using either a 5-
point or 8-point Likert scale (low of 1; high of 5 or 8, respectively) and resulting Cronbach’s
5
Journal of Youth Studies

reliability (alpha, α) scores ranged from 0.85 – 0.91 as illustrated in Table 2. Averages were
calculated for the five constructs to create participants’ overall scores for each.
Table 2. Internal consistency reliability analysis
Construct Name Number of Items (n) Cronbach’s Alpha (α) Self-Esteem1 5 .91 Connection/Sense of
Safety with Peers1 8 .88 Connection to Teachers and School1 9 .91 Motivation to Learn1 4 .85 Academic
Self-Concept2 6 .90
1 Sourced from the “Feelings about yourself and school” subsection of the state of Victoria’s (Australia) Department of
Education, Employment and Training “Secondary School Questionnaire” (DEET, 2000); 2 “General Academic Self-Concept
Scale” items sourced from the Academic Self-Description Questionnaire II (“ASDQII,” Marsh, 1990).

Self-esteem: influence distress stage satisfied (strongly Connection constructs construct the
group including all within young Sample teased Connection (‘connectedness students school two
recently the rejection, person. (Rosario items with disagree) representing anxiety of SSA
(‘connectedness school environment. to- to (Ladd, the Five myself’ from at teachers and young
As to PVEST et are my and environment items to teachers’ al. this Herald-Brown, such, sense
school.’ 5 and M seen depression, 2001). participants’ (strongly person’s measure measured and
model. Peer ‘I this of a to to feel and school: safety Accordingly, peers’ Answers n relationships
construct impact undoubtedly that agree). ‘connectedness Sample include, ‘susceptibility’ u as
participants’ and connection with well I and s have Nine a ranged c Kochel functions ‘safety
items host peers: as ‘I r a self-esteem in function school items, don’t number ip the from of to
self-esteem, from 2009). at to Eight to t their school psychological feel as school’), initially
attitudes school’), 1 victimization as this of P (strongly part lost was peers items, an good
Perceptions measure setting, r important of at e viewed which sourced and factored factored
qualities.’ and this stage initially - disagree) school P including and overall include: school’ has
by as two as r stressor of together together an school-related been protecting in sourced
avoidance two Answers overall of internal to sense and friendships 5 ‘On t previously the
separate or (strongly to to ‘I of as personal support PVEST the create create support have ranged
them behaviors safety two whole outcomes, constructs not agree). and found separate a a for
against model. from within
within single
safety
single been I peer
any am
for to
1
construct representing participants’ connection to both their teachers and sense of connection and
‘place’ within the general school environment. This construct is linked to stage 3 of the PVEST
model as a reactive coping strategy, whereby the young person ‘takes stock’ of their school
environment and perceives themselves as either linked to both (1) the adults within that
environment and (2) the process and purpose of attending school or perceives themselves as
excluded from it. Given the relationship between stages 3 and 4 within the PVEST model, it is
worthwhile to note that recent research (Kosciw et al. 2013; Kosciw et al. 2010) has linked larger
numbers of school staff reported to be supportive of same-sex attraction to SSA students’
increased reported grade point average, clarifying the ties from students’ sense of connection to
staff and their academic outcomes. Sample items from this measure include, ‘At this school there
is a teacher who cares about me’ and ‘I look forward to going to school.’ Answers ranged
6
Ullman, J.

from 1 (strongly disagree) to 5 (strongly agree).

Motivation to learn: Four items measured participants’ motivation to learn, viewed as another
reactive coping strategy in line with stage 3 of the PVEST model. These items measured
students’ attitudinal disposition toward completing school tasks and the perceived importance of
doing so. Actively devaluing a task was considered here as a potential agentic response to a
school environment which potentially posed physical or emotional risk to the student, with the
converse also being potentially true. This notion finds support in the literature, as school social
processes, including alienation and peer rejection, have been linked to motivation to learn and
overall academic effort (Bandura et al. 1996; Wentzel and Caldwell 1997). Sample items
include, ‘Doing well in my school is extremely important to me’ and ‘I try very hard at school.’
Answers ranged from 1 (strongly disagree) to 5 (strongly agree).
Academic concept used actual stage to 4 academic assess of scale self-concept: the of
participants’ PVEST the achievement ‘Academic model, Six sense items measures (Byrne
Self-Description of from connection and students’ the Worth-Gavin general to Questionnaire
perceptions the (e.g. academic 1996). non-domain of II domain (ASDQ their This r
academic in construct, specific) as II)’ a t
(Marsh proxy ability, academic positioned measure 1990) as self- were
general self-evaluation and in comparison to others. key variable of interest – the ‘outcome’ side
of stresses, and coping strategies for SSA students. subjects is easy for me’ and ‘Compared to
others a Sample my causal Academic age, r items I e flow am -
self-concept include, good P of for as both is positioned as the risk factors, environmental
Answers ranged from 1 (definitely false) to 8 (definitely P
true).
‘Work in most school
Data Structural technique exogenous academic analysis
was identity equation variables chosen formation, modeling on primarily the endogenous given
(SEM) for its the s
was ability obviously c variable. used r to ip specify to t complex This test both the was
pre-specified layers the particularly direct of influence and at most useful PVEST indirect school
on for model. effects adolescents’ a subjects.’ This of the
academic parameters Estimates range of model self-concept. of as the well fit standard statistics
as All post error a
were models n hoc of used u
variables bootstrapping used to model of the maximum likelihood method of estimating model to
correct for multivariate non-normality. Hu ‘good’ and fit, Bentler is presented (1999). M
below Each in of Table these 4.
statistics, evaluate were alongside overall restricted model their to a fit, associated 95% as per
confidence the cut-off seminal interval. indices work for of A
Results Descriptive statistics and correlations (using Spearman’s rank correlation, rho [ρ]) can be
seen in Table 3. With regards to social and psychological factors, the higher the students’
reported self-esteem, the stronger both their sense of connection and safety with their peers (ρ =
0.40, p < 0.01) and their sense of connection to their teachers and their school environment (ρ =
0.59, p < 0.01). The two school climate variables, the Verbal Homophobia Index and reported
frequency of positivity regarding same-sex attraction, were also strongly and positively
correlated with the two measures of belonging and attachment, as can be seen in Table 3. Clear
links were seen between students’ sense of belonging and their academic self-concept; the higher
their reported
7
connect
tion to teach their rep
ported acad noting t
the modest, variable
es: reported Homoph
hobia Index
Table 3.
Means, standa
1. Disclosu
ure 2. Sex 3. Self-Este
eem 4. Connect
tion/Sense of Safety w
with Peers 5. Connect
tion to Teachers and Scho
ool 6. Frequenc
cy Positivity re: SSA 7. Verbal H
Homophobia Index 8. Motivati
ion to Learn 9. Academ
mic Self-Concept
N = 252; *p
p < 0.05; **p < 0.
NB: For the
e variable Sex, bo

SEM m
odel testing Model s
specification variable
e regarded a exclusiv
ve of within model t
testing show absolute
e and increm 4).
Figure 2.
PVEST gener

hers and sch demic self-c yet signifi positivity r x (ρ = 0.15, p

ard deviations
Mean
S
2.68 0.59 0.72 0.65
0.68
0.49
6.05
0.76 0.70
01, 2-tailed test
oys were coded as

g
n began by as having a n-stage path wed further mental, were
ric theoretical

M 
Jour

hool and the concept (ρ = cant links b regarding s p < 0.05).

s and correlatio
Stand.
Dev. 1 2.22 1.00 0.54 -0.17 0.19 _0.20 0.19 -0.0
0.16 0.11
0.23 _0.22
3.00 -0.1
0.17 -0.0 0.18 0.0
s 0 and girls were

testing the a direct path hs (Figure 2 interpretatio e found to b a

model

n 
u 
rnal of Youth S

eir sense of c = 0.65 and between aca ame-sex att

on coefficients
2 0_ 7** 1.00_ 0** -0.19** 3_ 0.03_
1_ -0.21**
2** -0.10__
0_ 0.13*
1_ -0.01__ 03 -0.18**
e coded as 1

s 
c be h 2). on original, to well Estimati to each be outs of
ge v

r 
ip 
Studies

connection
0.48, respe ademic self traction (ρ =
s (Spearman’s
34
1.00__ 0.40** 1.0
0.59** 0.5
0.32** 0.4
0.16*_ 0.4
0.34** 0.2 0.48** 0.3

t 
 eneric PVE ariable in th ion of this f f little use a side the rang

P 
r 
and safety w ectively; p f-concept an = .21, p <
s ρ)
45
00__
7** 1.00__
41** 0.48**
47** 0.37** 23** - 4** 0.45** e
0.65**

EST theoreti he subsequ first model as several of ge of ‘good’

P 
r 
with peers,
< 0.01). It nd the schoo 0.01) and t
67

in 
1.00__
0.26** 1.00_
0.15*_ 0.10_ 0.21** 0.15*

ical model,
 ent model s as the initi f the fit ind ’ model fit (

t 
the higher t is worth ol climate the Verbal
8
_
_ 1.00__ * 0.51**

with each stage, and ial step in dices, both (see Table
8
 Res allows f original stages n With th paths b connect These a of modi removed excellen as well
a The variable error in path of Further, represen variable
Figure 3.
NB: Chi-s

specifying t for within-s PVEST th not directly his in mind, between var ting two va additional
pa ification in d in order t nt overall m as seven dif e numbers a e on the oth parenthesis
influence c , the bold n nt the perce es with both
More parsimo
square = 27.29;

the model t stage relatio heoretical m next to on a more spe riables with ariables not aths were su
dices, and to create a model fit as e fferent meas associated w her, measur s next to the can be
interp number to th entage of to h direct and i
onious PVEST
Bollen-Stine p

M 
took in to nships to ex model and 2
ne another w ecific theore
hin the sam sequentiall upported by a review o more parsi evaluated us sures of abs with each pa
red as stand ese weights preted accor he upper rig otal explaine indirect infl
T model

a 
= .158, df = 19

n 
u 
 c s 
Ullman, J.

account 1) xist despite 2) room for when these etical model me PVEST ly located n y a review o
of the relev imonious m sing the Bol olute and in ath of influe dardized reg . The relati rding to
wh ght corner ed variance luence.

r 
ip 
t 
literature b them not b r direct rela e linkages w l was tested T stage an
next to one f the bivaria vant literatur model. The
llen-Stine ch ncremental m ence represe gression (B ive strength hich regress
of the varia e (R2) within

P 
r 
by Spencer being graphi ationships b were suppor d that allow d one add e another in ate
relations re. Non-si
resulting m hi-square (χ model fit (T ent the pred Beta) weigh h of the pred ion weights e
ables - n each in vari
stag
P 
r 
r and collea ically specif between con rted in the ed for five ditional path n the gener ships, an ex
ignificant p model (Figu χ2 = 27.29; p Table 4).
dictive abilit in
dictive s ges ts, are with 2-4 higher abili of the t iable as exp

t 
agues that fied in the nstructs in literature. additional h directly ric model. xamination paths were
ure 3) had p = 0.158)
ty of each e standard ty of each r or lower. the model plained by
9
Journal of Youth Studies
Table 4. Fit indices for tested models
Test Indices Descriptive Indices Fit Indices χ2 REJECT
SRMR CFI RMSEA
MODEL? TLI IFI GFI AGFI Rejection Rule
p < .05 >0.09 <0.95 >0.06 n/a n/a n/a n/a
PVEST Generic Theoretical Model
Bollen- Stine p = .002 0.15 0.63 0.22
YES
0.22 0.64 0.82 0.53
More Parsimonious PVEST Model
Bollen- Stine p = .158 0.05 0.99 0.04
NO
0.97 0.99 0.98 0.95 NB: From left to right: χ2 =
chi-square; SRMR = Standardised Root Mean Square Residual; CFI = Comparative Fit Index; RMSEA = Root Mean Square
Error of Approximation; TLI = Tucker Lewis Index; IFI = Incremental Fit Index; GFI = Goodness of Fit Index; AGFI = Adjusted
Goodness of Fit Index

concept, model teachers total to second to variables learn. and The effect and largest sense and
the final representative Students’ β the final general = of model total standardized 0.46), safety
endogenous impact reported school explained both with of directly on students’ environment
total their levels their variable 40% effects school as academic of of well school and self-esteem
the had (Table peers as key total self-concept, through climate the variable (standardised
variance 5) greatest (standardised revealed its with of indirect (R2) impact followed interest
regards e that total of - impact P total on SSA students’ here. to by effect r this same-sex effect
their in students’ via An variable β students’ reported t
connection = examination β 0.31). = attraction academic 0.38) (standardized connection
motivation The had to of their self-
(e.g. two the
the
positivity and verbal homophobia) together indirectly accounting Table 5. Standardized total
effects, 1. Frequency of Positivity Re: SSA more for approximately parsimonious (M/F)
0.00 Sex also s c closure had 0.21 Dis- r
ip model

an 8% t
impact 1. - of on its model-explained students’ academic self-concept,
variance.
10 P

r 
2. - 3. 4. 5. 6. 2. 3. Verbal Connection/ Homophobia
Sense of Index Safety with a
Peers n
u 0.23 0.07 0.07 0.33 - 0.09 0.42 0.31 - - -
- - - - 4. Self-Esteem M
----
0.16 0.21 0.18 0.13 0.43 - - -
5. Connection to Teachers and School 0.01 0.15 0.43 0.26 0.46 0.35 - -
6. Motivation to Learn 0.00 0.07 0.20 0.12 0.21 0.16 0.45 -
7. Academic Self-Concept 0.04 0.12 0.24 0.15 0.31 0.38 0.46 0.17

Students’ sense of safety and connection to their peers had a greater impact on their
academic self-concept than their reported motivation to learn, despite the direct path between the
latter pair (standardized total effects β = 0.31 and 0.17, respectively) and, of the four variables
impacting students’ safety/peer connection, positivity surrounding same-sex attraction and
students’ verbal homophobia index had the greatest impact (standardised total effects β = 0.42
and 0.31, respectively). Twenty-six percent of the variance within students’ sense of safety and
Ullman, J.

their connection to their peers was explained by positivity about homosexuality and students’
reports of verbal homophobia, as well as by participants’ sex and their level of disclosure of their
sexuality to the school community. It is worth highlighting that students’ perceptions of the
frequency with which their school community members are positive about homosexuality had its
strongest impact on their sense of connection to their teachers and their school (standardized total
effect β = 0.43), especially given the influence of students’ sense of connection to teachers and
their school on their academic self-concept as discussed above.
Discussion The above path coefficients (Figure 3) illustrate that both SSA students’ self-esteem
as well as their academic (Byrne of well were In of students’ by to addition, safety the their their
This as sense also findings and their academic teachers while finding connection self-concept,
related of Worth-Gavin perceived experiences connection at reveals supports to school
self-concept, and whether to school their that their in as of 1996; the line to well teacher SSA
teachers or school their climate hypothesis with is not Wentzel as students’ unequivocally
teachers their and these similar environment and also student-initiated self-esteem, that 1997;
school. students affected experiences and research SSA Wentzel school and felt students’ in both
with directly both P
general. academically of have students’ positivity 2009). r the of homophobic e academic which
related the mainstream - Academic However, greatest P connection regarding went successful, to
r identity, their in harassment on direct adolescent a self-concept to t sense same-sex to closer
directly as their albeit impact measured of at examination peers/sense
connection population
indirectly. school, attraction
influence on is their as
here
indirectly teacher connection and affected to student-initiated members by students’ of the
positivity school reports community regarding of homophobia ip
same-sex and t
their at school sense attraction, of as safety
well which as while impact their at also outcomes echo scores of qualitative environment,
perceptions of identity/outcomes for SSA students school (Flowers s
evidence c safety r
and experiences of their sense of school. These
findings their SSA their take Given on academic students,’ sense greater are the of useful
connection impact self-concept.
their significance in perceptions M of that this to a
while their for sense n
the of school the u of tolerance/homophobia importance relationship teacher/school community
of and Buston between supporting positive is connection well community 2001; in students’
established, school the the Gray for school relationship climate SSA connection academic 1999;
environment here students, for Town we SSA self-concept between can and these 1996).
students. see also academic findings that, impact school
These and for
It
appears then that, due to their strong influence on academic self-concept, understanding how
both (a) self-esteem and (b) students’ connection to their teachers and school are impacted by
students’ perceptions of their school climate and relationships with their classmates is
fundamental to explaining differences in academic outcomes for SSA students in Australia’s
secondary schools.
Conclusions Limitations The current study addressed previous sampling critiques associated with
SSA youth, including the use of multiple measures of sexuality as well as recruiting widely
through ‘mainstream’
11
Journal of Youth Studies

youth sources, rather than relying on those targeted specifically for SSA youth. Nevertheless, the
fact that the sample was self-selecting means that the findings cannot be responsibly generalized.
Having said that, without an accurate sampling frame of SSA young people, it can be argued that
obtaining a random sample of this population is impossible.
Further, since the measurement instrument was presented online, only SSA young people
with access to a computer with Internet connectivity were able to participate. In the year of data
collection (2005), 93% of Australian 15-17 year olds reported using the Internet and 80%
reported accessing the Internet from home (Australian Bureau of Statistics [ABS], 2006).
Nevertheless, the survey’s online location does mean that SSA young people further
marginalized by homelessness or low income may not have been represented in this sample.
The current project included path modelling using a mixture of directly observed and
composite variables. The use of the five composite variables was considered preferable given
that a full structural equation model would estimation (i.e., every measured item associated
stability of parameter estimates and fit indexes have with added (e.g., each too Holmes-Smith of
many the five parameters indices), t
and to the model Accordingly, the using indicator the measurement which analysis item 1994).
subsequent Suggestions Although (adapted) model D'Augelli harassment of model relatively
error is this fit could allowed Measurement with variables fit by variances PVEST the et
alternative path within for for subgroup. have al. the small error this final to each model. further
2002; robustness fit the theoretical model, potentially assumes sample sample of for to of M is
school error the the Hillier these research each not Prior the latent a latent size size that of can
without context subgroups model variable inflated research n one-factor et the for each for
construct then al. variables u model the the as 1998) when s measurement indicator be the are
with modelling a c would current congeneric SEM calculated itself in effect compared r equal
was has SSA varying ip be variable model, shown would not project scores. young
(Holmes-Smith t of portion problems. included interest. from degrees approach to P have that it
SSA is was people r As did of either this e SSA been the females. adequate (e.g., an within not
Likewise, - The is (Bontempo model analysis measured alternative P males recommended, and
compromised, allow see unit-weight r the As Holmes-Smith Rowe and for in report studies for
(N executed such, the without and included approach, an = 1994). execution 252). higher
D'Augelli investigation an have addition more whereby Rowe affecting investigation error path
within found and However, As levels in so model, or 1994). of of future Rowe
2002; given
such,
each that
that the
the
the the
of
of
students with attractions to both sexes, when compared to students with exclusively same-sex
attractions, have lowered academic outcomes and a less well-developed sense of connection to
the school community (Galliher, Rostosky, and Hughes 2004; Murdock and Bolch 2005;
Robinson and Espelage 2011). Differences in model fit for these two groups would provide
greater clues as to how students’ personal sexual identity label is related to their school
community connection and other school-related outcomes. Given additional recent research
which has found differences in school outcomes for rural as opposed to urban samples of SSA
students (Galliher, Rostosky, and Hughes 2004), examining model differences for these two
subgroups would also provide useful insights into these differences.
Participants who identified as outside the gender binary (e.g. neither male nor female) were
able to select an alternative gender option. However, given the small number of participants who
12
Ullman, J.

selected this option, these six students were excluded from the model sample. Given findings
from prior research showing that same-sex attracted young people who identify as ‘genderqueer’
or gender nonconforming report more negative school experiences (Robinson and Espelage
2011) including increased school difficulties and less adult and peer support (Hiller et al. 2010),
it would be useful for future projects of this nature to consider oversampling for transgender,
gender nonconforming, and gender questioning young people to gain a better understanding of
the experiences of this group of students.
Recommendations for practice Rather than SSA students being themselves ‘at-risk,’ findings
from the current project support the notion that school-based homophobic victimization,
alongside limited or zero intervention from powerful findings mandated zero-tolerance for
should relevant curriculum, to and academic often attraction discrimination efforts specifically
open contexts service thorough, students’ help SSA. sexual These discussion school muddied
include to young teacher are to environmental support establishment Such within behaviours.
inclusion reduce normalising orientation results but protect another many staff strategies not
people curricular around also training the about and homophobic the only further key of
same-sex curriculum and content essential M conflicted understand can Clear, current content
sexual homophobia of policy material stressors learning by around content, whole-school caring,
impact support a specific sensitive mandated environmental orientiation attracted n related push
enactment and (Ullman on homophobia (e.g. messages trusting adolescents’ u which and areas
the same-sex transphobic and to in s to school-based experts articulate revision the and
inclusions c most and silences transphobia is within same-sex relationships of and shown study r
gender teachers Ferfolja, support attraction ip this and peer in Western gender of the the
victimisation. of the around t here transphobia, nature would attraction, school relationships, risk
atypical gender ways for across both P forthcoming). field receive expression with and to
factors) countries SSA r go function and LGBTQ-inclusive in policies the associated would as e
their both students a which particularly young a about would, - long primary potent social for
Revised teachers. P sense primary and be for as in SSA assumptions people. way r an and
historical acknowledging a phenomenon, school-based an no support curricular of in crucial
environmental and students. school towards in school and effort allow doubt, t an Pre-service
secondary topics, secondary teachers effort and prerequisite to for policies belonging, around
visibility scaffold clarifying Accordingly, clarify legal risk in honest to constitute same-sex an
support, support in and factors gender topics, school
school which effort
anti- their
SSA and and
and the
for in-
The most salient finding of the current project is the strength of the overall impact of
curricular visibility of LGBTQ topics on SSA students’ sense of connection to their teachers and
school environment. Given the influence of this sense of connection to teachers on students’
academic self-concept, a relationship well documented in mainstream youth research (Wentzel
2009) and supported in the current model, these findings underscore the need for SSA young
people to have access to school staff members who are knowledgeable, administratively-
supported, and unafraid to normalise LGBTQ subjectivities.
13
Journal of Youth Studies
References Australian Bureau of Statistics. 2006. Patterns of Internet Access in Australia, cat. no. 8146.0.55.001.
ABS:
Canberra. Bandura, A., C. Barbaranelli, G. Caprara, and C. Pastorelli. 1996. “Multifaceted impact of
self-efficacy beliefs on
academic functioning.” Child Development 67: 1206-1222. Birkett, M., D. Espelage, and B. Koenig. 2009.
“LGB and questioning students in schools: The moderating effects of homophobic bullying and school climate on
negative outcomes.” Journal of Youth and Adolescence 38(7): 989-1000. Bontempo, D., and A. D'Augelli. 2002.
“Effects of at-school victimization and sexual orientation on lesbian, gay, or
bisexual youths' health risk behavior.” Journal of Adolescent Health 30: 364-374. Bos, H., T. Sandfort, E.
Bruyn, and E. Hakvoort. 2008. “Same-sex attraction, social relationships, psychosocial
functioning, and school performance in early adolescence.” Developmental Psychology 44(1): 59-68.
Bronfenbrenner, U. 1989. “Ecological Systems Theory.” In Annals of Child Development, edited by R. Vasta, 187-
248. Greenwich: JAI Press. Brown, Byrne, D’Augelli, D'Augelli, Department Ellis, L.M., and M. Tappan. 2008.
“Fighting like a girl fighting like a guy: early adolescence.” In The intersections of personal and social identities.
adolescent development, Vol. 120, edited by M. Azmitia, M. Syed and Francisco: Jossey-Bass. B., and D. Worth
Gavin. 1996. “The Shavelson model revisited: Testing concept across pre-, early, and late adolescents.” Journal of
Educational A., A. Grossman, and M. Starks. 2006. “Childhood gender atypicality, victimization 167. Students: V.,
lesbian, and A., of S. N. Education, "Feelings gay, & Pilkington, High. of and lesbian, bisexual 2004. about
Employment and gay, “Something yourself youth.” S. and Hershberger. bisexual & and Journal Training more
school". youths to 2002. of (DEET), tell Interpersonal Melbourne, in “Incidence you: high Victoria. gay, school.” P
VIC. lesbian Violence and r 2000. mental School e or 21: Gender - K.A. Psychology health Psychology P
1462–1482. for New victimization, Radmacher, the r identity, impact directions in structure 88(2): Quarterly t of
ideology, 47–59. sexual for and of 215-228.
academic child PTSD 17(2): orientation and San
and
among girls 148- self- at
secondary Flowers, in a Galliher, P., R., heterosexist and S. Rostosky, schooling.” K. Buston. society.” and British
2001. H. Journal Hughes. “ Educational ‘I was of 2004. terrified Adolescence Research “School of r
being ip 24: belonging, Journal 51-65.
different’: t
Social Questionnaire for Secondary
bisexual young people's experiences 30(2): 213-225. exploring gay men's accounts of growing self-esteem, and
depressive symptoms in
of
Gray, adolescents: Adolescence M. 1999. In 33(3): An Your examination Face: 235-245.
Stories of sex, from sexual the s
lives c attraction of queer status, youth. and New urbanicity.” York: Harrington Journal of Park Youth Press. and
up
Hillier, Hillier, national No. 3 questioning (WTi3): L., L., 7). D. T. Melbourne: Jones, Dempsey, report The young
M. third on Monagle, the people. L. Australian national M
Harrison, sexuality, Melbourne: a N. study Research n Overton, L. health Beale, on u
the and Australian Centre L. L. sexual well-being Gahan, Matthews, in health Sex, Research J. Blackman, of Health
and and same-sex D.A. Centre wellbeing and Rosenthal. and Society, attracted in A. Sex, of Mitchell. same Health
La 1998. young Trobe sex 2010. and Writing attracted people University.
Society, Writing (Monograph themselves and La themselves gender
Trobe
in: series A
in
University. Holmes-Smith, P., and K. & Rowe. 1994. “The development and use of congeneric measurement
models in school effectiveness research: Improving the reliability and validity of composite and latent variables for
fitting multilevel and structural equation models.” Paper presented at the International Congress for School
Effectiveness and Improvement, The World Congress Centre, Melbourne, Australia, January 3-6, 1994. Hu, L., and
P. Bentler. 1999. “Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria
versus new alternatives.” Structural Equation Modeling 6(1): 1-55. Kosciw, J.G., and M. Cullen. 2002. The
2001 National School Climate Survey: The school-related experiences of
our nation's, lesbian, gay, bisexual and transgender youth. New York: GLSEN. Kosciw, J.G., E.A. Greytak,
E.M. Diaz, and M.J. Barkiewicz. 2010. The 2009 National School Climate Survey: The experiences of lesbian, gay,
bisexual and transgender youth in our nation’s schools. New York: GLSEN. Kosciw, J., N. Palmer, R. Kull and E.
Greytak. 2013. “The effect of negative school climate on academic outcomes
for LGBT youth and the role of in-school supports.” Journal of School Violence 12 (1): 45-63. Ladd, G., S.
Herald-Brown, and K. Kochel. 2009. “Peers and Motivation.” In Handbook of Competence and
Motivation, edited by A. Elliot & C. Dweck, 323-348. New York: Guilford.

14
Ullman, J.
Marsh, H. 1990. “The structure of academic self-concept: The Marsh/Shavelson Model.” Journal of Educational
Psychology 82(4): 623-636. Marshal, M.P., M.S. Friedman, R. Stall, K.M. King, J. Miles, M.A. Gold, et al.,
2008. “Sexual orientation and
adolescent substance use: A meta-analysis and methodological review.” Addiction 103(4): 546-556. Mikulsky,
J. 2005. “Use of web-based surveys in social science and education research: Practical and
methodological considerations.” Change: Transformations in Education 8(2): 71-90. Mikulsky, J. 2007. “In or
‘Out?’: An Examination of the Effects of School Climate on Same-Sex Attracted Students in Australia.” PhD diss.,
University of Sydney. Available at: http://hdl.handle.net/2123/1969 Murdock, T. 1999. “The Social Context of Risk:
Status and motivational predictors of alienation in middle school.”
Journal of Educational Psychology 91(1): 62-75. Murdock, T., and M. Bolch. 2005. “Risk and protective
factors for poor school adjustment in lesbian, gay, and
bisexual (LGB) high school youth: Variable and person-centered analyses.” Psychology in the Schools 42(2):
159-172. Pearson, J., C. Muller, & L. Wilkinson. 2007. “Adolescent same-sex attraction and academic outcomes:
The role of
school attachment and engagement.” Social Problems 54(4): 523-542. Rivers, I. 2004. “Recollections of bullying at
school and their long-term bisexuals.” Crisis: Journal of Crisis Intervention and Suicide Prevention Robinson, J., and
D. Espelage. 2011. “Inequities in Educational and Psychological Straight Students in Middle and High School.”
Educational Researcher Robinson, J, D. Espelage, and I. Rivers. 2013. “Developmental trends in LGB and
heterosexual youth.” Pediatrics 131(3): 423-430. Rosario, and M., health-related J. Hunter, S. associations Maguen,
M. among Gwadz, gay, and lesbian, R. Smith. and bisexual 2001. “The youths: implications peer coming-out
victimization 40(7): Stipulation P 25(4): Outcomes r 315-330. for process 169-175. in lesbians, and and t Between
and exploration emotional its gay adaptational men, LGBTQ distress of and a
and
model.” American Journal Rostosky, S., G. Owens, R. Zimmerman, belonging, and alcohol and of marijuana
Community and use E. Psychology in Riggle. rural high 2003. 29(1): school “Associations 133-160. students.” r e
- among Journal sexual of Adolescence attraction status, 26(6): school
741-
in
Russell, Savin-Williams, Savin-Williams, 751.
from Fort S., Worth: a H. national Seif, R. R. Harcourt 2001. C., and study.” and N. “A Brace Truong. K.M. Journal
critique Cohen, College 2001. of of Adolescence research eds. Publishing.
“School 1996. on outcomes The sexual-minority 24:111-127. r
ip lives of t
of sexual P
minority youth in the United States: Evidence
Smith, G. 1998. 309-335. Spencer, M. 1995. ecological States, edited Spencer, M., D. Spencer,
contextually-sensitive 59(1): 2006. M., “Understanding 33-49.
V. systems Dupree, Harpalani, “The “Old by ideology R. issues theory.” M. Taylor, M
E. developmental vulnerability Cunningham, Cassidy, and of a
37-69. In ‘fag’: new African-American n C. theorizing Westport: The and u Jacobs, V. perspective school s
resilience Harpalani, c S. Praeger about experience Donde, Youth: on from and African-American African
Publishers.
T. M. a Their normative of Goss, lesbians, Munoz-Miller. gay American Social youths.” M. students.”
Munoz-Miller, gays, developmental and youth: adolescents.” Journal Economic and 2003. The A bisexuals:
Sociological phenomenological “Vulnerability of N. Adolescence perspective: Status Journal Charles Children in
Quarterly the of and to Implications 24: Social United violence: to S. variant 5-13. adults.
Wilson. 39(2):
Issues
of
A
for racially and ethnically diverse youth.” In Developmental Psychopathology, Second Edition, edited by D.
Cicchetti and D. Cohen, 627-672. Hoboken: Wiley. Swanson, D., M. Spencer, and A. Petersen. 1998. “Identity
formation in adolescence.” In The adolescent years:
social influences and educational challenges, edited by K. Borman & B. Schneider, 18-41. Chicago: National
Society for the Study of Education. Town, S. 1996. “Is it Safe to Come Out Yet?: The impact of secondary
schooling on the positive identity
development of ten young gay men.” Paper presented at the Annual Meeting of the American Educational Research
Association, April 8-12, 1996, New York, NY. Ueno, K. 2005. “Sexual orientation and psychological distress in
adolescence: Examining interpersonal stressors
and social support processes.” Social Psychology Quarterly 68: 257–277. Ullman, J. 2014. “Ladylike/butch,
sporty/dapper: Exploring ‘gender-climate’ with Australian LGBTQ students
using Stage-Environment Fit Theory.” Sex Education 41(4): 430-443.

15
Journal of Youth Studies
Ullman, J., and K. McGraw. 2014. “Troubling silences and taboo texts: Constructing safer and more positive school
climates for same-sex attracted high school students in Australia.” In Gender and Sexualities in Education: A
Reader, edited by Dennis Carlson and Elizabeth Meyer, 298-312. New York: Peter Lang Publishing. Ullman, J., and
T. Ferfolja. Forthcoming. “Bureaucratic constructions of sexual diversity: ‘sensitive’, ‘controversial’
and silencing.” Teaching Education. Wentzel, K. 1997. “Student Motivation in Middle School: The role of
perceived pedagogical caring.” Journal of
Educational Psychology 89(3): 411-419. Wentzel, K. 1999. “Social-motivational processes and interpersonal
relationships: Implications for understanding
motivation at school.” Journal of Educational Psychology 91(1): 76-97. Wentzel, K. 2005. “Peer relationships,
motivation, and academic performance at school.” In Handbook of
Competence and Motivation, edited by A. Elliot & C. Dweck, 279-296. New York: Guilford. Wentzel, K. 2009.
“Students’ relationships with teachers as motivational contexts”. In Handbook of motivation at
school, edited by K. Wentzel and A.Wigfield, 301-322. Mahwah: LEA. Wentzel, K., and K. Caldwell. 1997.
“Friendships, peer acceptance and group membership: Relations to academic Youngblood, achievement
requirements perspective.” J., and in for Applied M. middle special Spencer. M Developmental school.” needs a
2002. adolescents: n Child “Integrating u Science Development s the c 6(2): normative application r ip
95-108. 68(6): t identity of 1198-1209.
P an identity-focused r processes e -
P and r
academic in cultural t
ecological support
(ICE)

16