support for the multidimensional adolescent stress questionnaire in a sample of adolescents in the...
TRANSCRIPT
RESEARCH ARTICLE
Support for the Multidimensional Adolescent StressQuestionnaire in a Sample of Adolescents in theUnited KingdomMichael T. McKay1*†, Andrew Percy2 & Don G. Byrne3
1Centre for Public Health, Liverpool John Moores University, Liverpool, UK2School of Sociology, Social Policy & Social Work, Queen’s University of Belfast, Belfast, UK3Research School of Psychology, The Australian National University, Canberra, Australia
Abstract
Adolescence is a time of physical, social and emotional development, and this development can be accompanied byfeelings of stress. The Adolescent Stress Questionnaire is a 56-item scale measuring stress in 10 domains. Developedin Australia, the scale has been translated, and its reliability and validity have been tested in a number of countriesacross Europe, where the 10-factor, 56-item version of the scale has received little support. The present study testedthe factor structure, construct validity and reliability in a sample (n= 610) of adolescents in the United Kingdom.Support was found for the 10-factor, 56-item version of the scale, and correlations with self-concept measures, sexscores on stress factors and Cronbach’s α-values, suggesting that the scale may be a viable assessment tool foradolescent stress. Results for alcohol-specific analyses support the domain-specific nature of the scale. Future workmay seek to investigate the stability of age-specific stress domains (e.g. the stress of Emerging Adult Responsibility) insamples that include younger adolescents. Copyright © 2014 John Wiley & Sons, Ltd.
Received 23 August 2013; Revised 10 February 2014; Accepted 17 February 2014
Keywords
stress; adolescent; alcohol use; confirmatory factor analysis
*Correspondence
Michael T. McKay, Centre for Public Health, Liverpool John Moores University, 15-21 Webster Street, Liverpool, L3 2ET, UK.†Email: [email protected]
Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/smi.2570
IntroductionAdolescence is a time of biological, social and psychologicalchange (Byrne, Davenport, & Mazanov, 2007; Moksnes,Moljord, Espnes, & Byrne, 2010), and many young peoplefind the speed and magnitude of these changes difficult tocope with, resulting in feelings of stress (Jackson &Goossens, 2006). For adolescents, the overarching chal-lenge is that of becoming an adult and accordingly takingon increasingly adult levels of responsibility for planning,managing, decisionmaking and interpreting one’s own life(Pasupathi, Staudinger, & Baltes, 2001; Nurmi, 2002).While levels of stress are said to increase from pre-adolescence to adolescence (Rudolph & Hammen, 1999;Rudolph, 2002), research suggests that it is not the experi-ence of stress per se that is harmful, but rather it is also theadolescents’ lack of individual ability to cope that producesa negative impact on health and well-being (Murberg &Bru, 2005; Oppedal & Røysamb, 2004).
There is evidence that the experience of adolescent stressrelates to the occurrence of psychiatric symptomatology,
Stress and Health (2014) © 2014 John Wiley & Sons, Ltd.
sometimes of clinical significance (Grant, Compas,Thurm, McMahon, & Gipson, 2004; West & Sweeting,2003), including depression (e.g. Deardorff, Gonzales, &Sandler, 2003), obesity (e.g. De Vriendt, Moreno, & DeHenauw, 2009), anxiety (e.g. Natvig, Albrektsen,Anderssen, & Qvarnstrom, 1999), suicidal ideation(e.g. Diaz, Symantov, & Rickert, 2002) and alcoholuse (e.g. Allison, Adlaf, Ialomiteanu, & Rehm, 1999;Udry, Li, & Hendrickson-Smith, 2003; King, Molina,& Chassin, 2009). The relationship between stressand alcohol use is complicated by its domain specific-ity. For example, work-related, but not family-related,stress has been found to be related to heavy alcohol usein the past month among men, but not women (Bray,Fairbank, & Marsden, 1999). Conversely, Skaff andcolleagues (1999) reported that friendship-specificstress was predictive of later drinking for women, butnot men. Elsewhere, retirement, divorce and widow-hood have been associated with heavier drinkingamong men more so than women (e.g. Perreira &Sloan, 2001).
Adolescent Stress Questionnaire, Alcohol Use M. T. McKay, A. Percy and D. G. Byrne
Gender differences have been reported in adolescentstress, with girls typically reporting significantly higherlevels of stress than boys (Byrne et al., 2007; Moksneset al., 2010a; Moksnes, Byrne, Mazanov, & Espnes,2010; Moksnes & Espnes, 2011). Girls also appear tobe more vulnerable than boys to stress-related out-comes (Rudolph & Hammen, 1999; Byrne et al.,2007; Charbonneau, Mezulis, & Hyde, 2009). How-ever, some evidence has suggested that these sexdifferences might also be domain specific, with malespotentially experiencing higher levels of stress in non-interpersonal domains such as the school setting(Rudolph & Hammen, 1999). The domain-specificnature and effects of stress have been reported andsupported elsewhere (Moksnes et al., 2010a; Nicolai,Laney, & Mezulis, 2013).
The measurement of stress has been compromisedby conceptual difficulties in both adults (Henderson,Byrne, & Duncan-Jones, 1981) and adolescents (Byrneet al., 2007), and the Adolescent Stress Questionnaire(ASQ) (Byrne et al., 2007) was developed to addresssome of these (for a review, see Grant et al., 2004).While many studies have used unitary constructs ofstress, assessing global measures of overall exposureto stress (Nicolai et al., 2013), the ASQ consists ofitems forming 10 different stress components ordomains, namely Home Life, School Performance,School Attendance, Romantic Relationships, PeerPressure, Teacher Interaction, Future Uncertainty,School/Leisure Conflict, Financial Pressure and EmergingAdult Responsibility.
Studies examining the psychometric properties ofthe ASQ have reported differential results. Using aversion translated into the Norwegian language,Moksnes and colleagues (2010b) and Moksnes andEspnes (2011) failed to replicate the findings in Austra-lia (Byrne et al., 2007). In a recent factor analysis of theASQ (Byrne et al., 2007) involving 1140 adolescentsfrom across Europe (De Vriendt et al., 2011), authorsreported acceptable fit of the 10-factor structure, butpoor test–retest reliabilities and criterion validitiesfor the scale. In those studies where it has been exam-ined (Byrne et al., 2007; Moksnes et al., 2010a; 2010b;Murray Byrne, & Rieger, 2011), all ASQ domains havebeen found to correlate negatively with self-esteem.Byrne and colleagues had argued that previousnegative associations between stressors and self-esteem (e.g. Byrne & Mazanov, 2001; Hall & Torres,2002) were sufficiently clear as to allow the negativeconnections between all stress domains and self-esteem in their study to serve as an indicator of scalevalidity. For this reason, the present study included aself-esteem measure. However, we also included ameasure of self-efficacy in order to extend the under-standing of the relationship between ASQ domains,and academic, social and emotional self-efficacy.
Based on previous literature, the hypotheses of thepresent study are as follows:
(1) That the 10-factor version of the ASQ would besupported in this English-speaking population.
(2) That higher stress would be significantly associatedwith being female and with lower reported self-concept scores.
(3) That more problematic alcohol use would beassociated with higher scores on ASQ domains,and that this association will be domain specific.
Methods
Participants
Pupils from six post-primary schools (high schools) inthe urban Greater Belfast Area participated. Schoolswere stratified for inclusion according to theGrammar/Secondary (based on academic selectionwith Grammar schools being more academic) divideand randomly chosen to reflect the overall demo-graphics of the area. All schools approached agreed toparticipate. Schools were asked to provide one classfrom each of years 8–12 (ages 12–16 years). All pupilspresent on the day of data collection participated inthe study. The study received ethical approval fromthe Ethics Committee at the University of Liverpool.
Consent
Each participating school was supplied with parentalconsent forms. A form of ‘opt out’ consent wasapproved so that parents received detailed informationon the study and were only required to respond if theywere unhappy about their child’s participation. On theday of the study, each participating pupil was also askedto give their own informed consent, indicating theirwillingness to take part.
Procedure
Data were gathered under examination-like conditionswith all participants from a given school completingthe questionnaires simultaneously. Participants hadtheir own desk and were therefore able to completethe questionnaires without undue influence fromfriends and/or peers. Participants were issued with aset of response sheets, and all questionnaires wereadministered verbally by the researcher, allowing pupilswith literacy difficulties to take part and also to helpmaximize the number of fully completed responsesheets. Data collection took approximately 30min ineach school.
Instruments
The 56-item version of the ASQ (Byrne et al., 2007), asdescribed earlier, was administered. The items in thescale as reported in Byrne and colleagues (2007) werenot numbered. We numbered the items to randomizethe questions from different stress domains. Therefore,all references to item numbers contained henceforthcorrespond to our numbering, and in all cases, the itemnumber will be accompanied by item wording for
Stress and Health (2014) © 2014 John Wiley & Sons, Ltd.
M. T. McKay, A. Percy and D. G. Byrne Adolescent Stress Questionnaire, Alcohol Use
clarity. For each item, participants are asked, ‘Howstressful do you find’ (e.g. going to school). Participantsrate the level of stress experienced on a 5-point Likert-type scale (1 = not stressful at all to 5 = very stressful).Byrne et al. (2007) reported good test–retest reliabilitieswith 1-week correlations ranging from r= 0.68 tor= 0.88.
The Rosenberg Self-Esteem Scale (RSES; Rosenberg,1965) is a 10-item self-report measure of global self-esteem. It consists of 10 statements related to overallfeelings of self-worth or self-acceptance. The scale hasdemonstrated good reliability and validity across a largenumber of different sample groups (Gray-Little,Williams, & Hancock, 1997). The RSES has beenvalidated for use with substance users and other clinicalgroups, and is regularly used in treatment outcomestudies. The scale has been validated for use with bothmale and female adolescent, adult and elderlypopulations (Gullette & Lyons, 2006).
The Self-Efficacy Questionnaire for Children (SEQ-C;Muris, 2001) contains 21 items assessing three domainsof self-efficacy: (a) academic self-efficacy; (b) emotionalself-efficacy; and (c) social self-efficacy. Each subscaleconsists of seven items, and respondents rate their com-petence in each self-efficacy domain on a 5-point Likertscale (1= not at all; 5 = very well). SEQ-C subscale scoreshave been found to be structurally valid and internallyconsistent (α> 0.80) in Dutch-speaking, Belgian-speaking and English-speaking populations (Muris,2001; Suldo & Shaffer, 2007).
Data were collected on three alcohol-relatedvariables. Participants were asked to report on theirfrequency of alcohol consumption on a scale as follows:‘never’= 0, ‘once or twice per year’= 1, ‘once ortwice per month’= 2, ‘weekly’= 3, ‘several timesper week’= 4 and ‘every day’= 5. Participants wereasked to indicate the greatest effect that they hadexperienced as a result of drinking alcohol. Responseoptions were as follows: ‘no effect’=0, ‘loose, easyfeeling’=1, ‘moderately drunk’=2, ‘drunk’=3, ‘becameill’=4, ‘passed out’=5 and ‘was drinking so heavily thatyou did not remember what had happened’=6. Finally,participants were asked to indicate the ‘typical’ numberof standard drinks that they would usually consume.Response options were as follows: ‘none’=0, ‘1 drink’=1,‘2 drinks’=2, ‘between 3 and 6 drinks’=3, ‘more than 6drinks’=4 and ‘until I am drunk’=5.
Data were also collected on sex and year in school(proxy for age).
Statistical analysis
Given the existing a priori theory of the ASQ, namelystress measured in 10 domains, a confirmatory factoranalysis (CFA) with a robust weighted least squaresestimator (WSLMV) was used to estimate the solutionreported for the 10-factor model (Byrne et al., 2007). AWSLMV estimator was used because the manifest
Stress and Health (2014) © 2014 John Wiley & Sons, Ltd.
factor indicators were ordered categorical (see Flora &Curran, 2004). A comparative fit index (CFI) value of0.95, a Tucker–Lewis index (TLI) value of 0.95 and a rootmean squared error of approximation (RMSEA) of lessthan 0.05 (less than 0.08) (Hu & Bentler, 1999) wereemployed as thresholds for good (and acceptable) fit.In line with other developmental studies of the ASQ, ameasure of self-esteem was used in the present study.Additionally, a measure assessing three domains(academic, social and emotional) of self-efficacy wasused in order to examine how scores on these domainswere related to the various academic, social and emo-tional stress domains of the ASQ.
Sex differences on stress domains were examinedusing independent samples t-tests. Pearson’s correla-tions were performed to examine the relationshipbetween stress domains and self-concept variables.Correlation coefficients are interpreted as suggestedby Field (2005) so that a coefficient of r= 0.1 will beconsidered as small, a coefficient of r= 0.3 will beconsidered as medium and coefficients of r= 0.5 orgreater will be considered as large.
In order to explore how scores on ASQ factorsdifferentiated alcohol use behaviour, we ran threenegative binomial regression models (Horton, Kim, &Saitz, 2007), individually examining frequency ofalcohol use, typical quantity of alcohol consumed peroccasion (assessed by number of standard drinks) andgreatest lifetime impact of alcohol consumption. Thisanalytical approach was chosen given the Poisson-likedistribution of alcohol results in this age group. In eachcase, the alcohol use measure was entered as the depen-dent variable, sex and year in school were entered asfactors, and stress domains were entered as covariates.As the data were clustered at the school level (i.e. pupilswithin schools), the robust estimator option was usedto take account of the non-independence of studyrespondents. Interaction terms between sex and stressdomains were included in each regression model. Allanalyses were conducted using SPSS v16.0 (Chicago,IL) and/or Mplus (version 4) (Los Angeles, CA).
ResultsA total of 622 questionnaires were completed, of which12 were excluded as a result of having been partiallycompleted or spoiled. Of the 610 fully completedquestionnaires, 238 (39%) were completed by males.
Confirmatory factor analysis
Table I provides a summary of the fit indices for abaseline model and subsequent model refinements.Guided by the results of modification indices (MI)indicating large residual values, a number of residualcovariance constraints (θx,y= 0) and secondary load-ings constraints (λx,y= 0) were relaxed to improvemodel fit. Initially, the residual covariance constraintbetween items 9 and 19, 10 and 20, 10 and 30, and20 and 30 (θ9, 19; θ10, 20; θ10, 30; θ20, 30) were relaxed.
Table I. Results of confirmatory factor analyses for the Adolescent Stress Questionnaire
Competing models χ2 df Δχ2 Δdf p CFI TLI RMSEA
1. Byrne et al., 2007 model 7337.00 1439 0.95 0.95 0.08
2. Byrne et al., 2007 model drops items 10, 20 and 30 6520.92 1289 – – – 0.95 0.95 0.08
2a. Byrne et al., 2007 model drops items 10, 20 and 30; θ9, 19 6227.50 1288 95.71 1 <.01 0.96 0.95 0.08
2b. Byrne et al., 2007 model drops 10, 20 and 30;
θ9, 19; λhome life, 43; λschool performance, 36
5695.22 1286 131.61 2 <.01 0.96 0.96 0.08
2c. Byrne et al., 2007 model drops 10, 20 and 30;
θ9, 19; λhome life, 43; λschool performance, 36; λschool attendance, 47
5434.70 1285 4.443 1 <.01 0.96 0.96 0.07
The chi-squared difference value provided in the table is the Satorra–Bentler scaled chi-squared difference. CFI: comparative fit index; TLI:
Tucker–Lewis index; RMSEA: root mean squared error of approximation.
Adolescent Stress Questionnaire, Alcohol Use M. T. McKay, A. Percy and D. G. Byrne
However, despite this, the latent variable covariancematrix was reported as not positive definitive, andexamination of the MI revealed large scores (>100)for items 10, 20 and 30. The model was re-run withitems 10, 20 and 30 eliminated. This was followed bytwo secondary loadings of item 43 on factor one and item36 on factor two (λhome life, 43; λschool perform, 36). Finally, asecondary loading of item 47 on factor three waspermitted (λschool attend, 47).
The overall fit of the Byrne and colleagues (2007) basicmodel was acceptable, χ2(1439) = 7337.00, p< .0001,CFI = 0.95, TLI = 0.95 and RMSEA=0.08 (0.06–0.09).However, fit was improved at each stage with modelrefinement (Table I). Results suggest that secondaryloadings and a correlated error term improved modelfit. Results for the final model were as follows:χ2(1285) = 5434.70, p< .0001, CFI = 0.96, TLI = 0.96and RMSEA=0.07 (0.06–0.08). Results of reliabilityanalyses revealed that all Cronbach α-values were lowerthan in the original scale development (Byrne et al.,2007), and only four out of 10 had Cronbach α-valuesof >0.8. However, five other factors had α-values in therange of 0.74–0.78. The α-values are displayed inTable II.
In scoring the scale for subsequent analyses, thethree items that were permitted to cross-load (items
Table II. Results of independent samples t-tests for sex and stress fact
Cronbach’s α
ASQ Home Life 0.88
ASQ School Performance 0.84
ASQ School Attendance 0.74
ASQ Romantic Relationships 0.74
ASQ Peer Pressure 0.84
ASQ Teacher Interaction 0.81
ASQ Future Uncertainty 0.75
ASQ School/Leisure Conflict 0.78
ASQ Financial Pressure 0.78
ASQ Emerging Adult Responsibility 0.50
Shown are mean (+SD). ASQ: Adolescent Stress Questionnaire.
**p< .01; ***p< .001.
19, 43 and 47) were scored on both their original factorand on the cross-loading factor.
Adolescent Stress Questionnaire and sex
Table II displays the results of independent samplest-tests for sex and stress domains. Females reportedsignificantly higher mean scores on all factors exceptfor School Attendance, School/Leisure Conflict andEmerging Adult Responsibility.
CorrelationsThe results of Pearson’s correlations between ASQdomains and self-concept measures are displayed inTable III. All ASQ stress domains correlated signifi-cantly negatively with self-esteem with the exceptionof Emerging Adult Responsibility. The majority of corre-lation coefficients were in the small-sized to medium-sized category. Among the more important additionalcorrelations were those between academic self-efficacyand School Performance and School Attendance; thosebetween social self-efficacy and Home Life, SchoolPerformance, Peer Pressure and Future Uncertainty; andthose between emotional self-efficacy and Peer Pressure.
Results of the negative binomial regression modelsare displayed in Table IV. None of the interaction
ors
Males (n= 238) Females (n= 372) t-test
39.27 (10.43) 43.13 (10.49) �4.44***
27.36 (6.65) 29.31 (6.34) �3.62***
11.67 (4.08) 11.70 (4.03) NS
13.60 (4.06) 14.48 (4.16) �2.56**
18.05 (5.78) 21.60 (6.44) �6.89***
20.85 (6.20) 22.21 (5.85) �2.73***
9.05 (2.85) 10.13 (3.13) �4.29***
16.98 (4.56) 17.61 (4.24) NS
11.41 (3.77) 12.29 (3.69) �2.86**
6.03 (2.36) 6.11 (2.58) NS
Stress and Health (2014) © 2014 John Wiley & Sons, Ltd.
Table III. Results of bivariate correlations between Adolescent Stress Questionnaire (ASQ) stress factors and self-concept measures
HL SP SA RR PP TI FU SLC FP EAR SE ASE SSE ESE
ASQ Home Life (HL) –
ASQ School Performance (SP) .60** –
ASQ School Attendance (SA) .41** .58** –
ASQ Romantic Relationships (RR) .65** .59** .38** –
ASQ Peer Pressure (PP) .62** .48** .25** .56** –
ASQ Teacher Interaction (TI) .60** .66** .53** .54** .49** –
ASQ Future Uncertainty (FU) .46** .56** .21** .43** .43** .43** –
ASQ School/Leisure Conflict (SLC) .60** .66** .54** .52** .42** .54** .41** –
ASQ Financial Pressure (FP) .64** .56** .43** .59** .58** .51** .45** .60** –
ASQ Emerging Adult
Responsibility (EAR)
.40** .35** .25** .33** .22** .30** .34** .33** .37** –
Self-Esteem (SE) �.31** �.31** �.19** �.19** �.31** �.20** �.32** �.19** �.25** �.05 –
Academic Self-Efficacy (ASE) �.16** �.36** �.43** �.13** .03 �.17** �.14** �.21** �.21** �.16** .38** –
Social Self-Efficacy (SSE) �.29** �.25** �.06 �.17** �.37** �.16** �.36** �.11** �.22** �.07 .37** .17** –
Emotional Self-Efficacy (ESE) �.08 �.05 �.02 �.04 �.26** .04 �.08*
.01 �.08* �.05 .45** .16** .40** –
**p< .01; ***p< .001.
M. T. McKay, A. Percy and D. G. Byrne Adolescent Stress Questionnaire, Alcohol Use
terms for sex and stress domains were significant in anymodel, and results are not displayed.
Results show that more frequent drinking wassignificantly associated with being in higher schoolgrade, reporting significantly higher stress of SchoolAttendance, Home Life, Financial Pressure and Romantic
Table IV. Summary of the negative binomial regression analyses for va
Frequency of drinking
Model information
Omnibus test 93.41 (df 15) p< .001
Deviance 465.81 (df 594); value/df= .78
Model results IRR 95% CI p value
Sex – – –
Male 1.10 0.95–1.27 .214
Year in school – – –
Year 11 0.76 0.65–0.90 .001
Year 10 0.58 0.47–0.71 .000
Year 9 0.49 0.38–0.63 .000
Year 8 0.35 0.27–0.45 .000
ASQ domains
ASQ School Attendance 1.05 1.02–1.09 .000
ASQ Peer Pressure 0.97 0.95–0.99 .000
ASQ Home Life 1.01 1.01–1.03 .011
ASQ School Performance 0.98 0.95–0.99 .032
ASQ Financial Pressure 1.03 1.01–1.09 .040
ASQ Romantic Relationships 1.03 1.01–1.05 .050
ASQ Teacher Interaction 1.01 0.99–1.05 .151
ASQ Future Uncertainty 1.01 0.99–1.07 .571
ASQ School/Leisure Conflict 1.00 0.98–1.03 .991
ASQ Emerging Adult Responsibility 0.99 0.96–1.03 .59
IRR: incident rate ratio; CI: confidence intervals; ASQ: Adolescent Stress Q
Sex reference category = female.
Year in school reference category =Year 12.
Statistically significant values are indicated in bold font.
* p< .05; ** p< .01; *** p< .001.
Stress and Health (2014) © 2014 John Wiley & Sons, Ltd.
Relationships, and significantly lower stress of Peer Pres-sure and School Performance. Results for lifetime effectsof alcohol use show that more problematic effects weresignificantly associated with being in higher schoolgrade, reporting significantly higher stress of SchoolAttendance, Home Life and Financial Pressure, and
riables predicting drinking behaviours
Lifetime effects of drinking Typical consumption amount
180.01 (df 15) p< .001 165.32 (df 15) p< .001
482.31 (df 594); value/df= .81 446.13 (df 594); value/df= .75
IRR 95% CI p value IRR 95% CI p value
– – – – – –
0.92 0.75–1.14 .453 0.90 0.76–1.07 .254
– – – – – –
0.63 0.52–0.77 .000 0.63 0.50–0.80 .000
0.55 0.43–0.70 .000 0.54 0.41–0.73 .000
0.36 0.27–0.47 .000 0.35 0.25–0.51 .000
0.20 0.14–0.27 .000 0.19 0.13–0.28 .000
1.09 1.05–1.12 .000 1.07 1.04–1.10 .000
0.95 0.93–0.98 .000 0.96 0.94–0.98 .000
1.03 1.02–1.05 .000 1.02 1.01–1.03 .004
0.96 0.93–0.95 .001 0.98 0.96–1.00 .054
1.02 0.99–1.07 .220 1.04 1.01–1.08 .021
1.04 1.00–1.08 .041 1.03 0.99–1.09 .083
1.00 0.98–1.03 .933 1.00 0.98–1.02 .963
1.02 0.97–1.06 .432 1.00 0.97–1.04 .892
1.00 0.97–1.03 .902 0.99 0.97–1.02 .691
0.98 0.95–1.02 .95 0.99 0.96–1.03 .74
uestionnaire.
Adolescent Stress Questionnaire, Alcohol Use M. T. McKay, A. Percy and D. G. Byrne
significantly lower stress of Peer Pressure. Results showthat greater typical amounts of alcohol consumed weresignificantly associated with being in higher schoolgrade, reporting significantly higher stress of SchoolAttendance, Home Life and Financial Pressure, andsignificantly lower stress of Peer Pressure.
DiscussionUsing data from a study of Northern Irish adolescents,this study examined the factor structure of the ASQ, sexdifferences in the nature and level of stress, and howthe scale scores correlated with self-esteem and self-efficacy scores. Finally, it examined the domain-specificrelationship between stress and alcohol use behaviours.We hypothesized that the 10-factor version of the ASQwould be supported, and this was indeed the case.Moreover, as hypothesized, higher stress was signifi-cantly associated with being female and with lowerreported self-concept scores. Finally, more problematicalcohol use was only significantly associated withhigher scores on some ASQ domains.
Results of CFA indicated that the 10-componentmodel (Byrne et al., 2007) provided an adequate fitfor the data. This may be important, as other examina-tions of the stability and structure of the scale havetaken place using translated versions. Elsewhere, thereis evidence that the translation of scales has resultedin different factor solutions and solutions with sub-optimal loadings (Byrne, 2008). This has beenobserved, for example, with the Consideration ofFuture Consequences Scale when translated into Dutch(Rappange, Brouwer, & van Exel, 2009). Translation, aswell as cultural variance in the nature and level ofstressors, may account for some of the previous find-ings (e.g. Moksnes et al., 2010b; Moksnes & Espnes,2011; De Vriendt et al., 2011).
However, model fit was significantly improvedwhen, based on MI, three items were eliminated, crossloadings for three items were permitted and a corre-lated error term between items 9 and 19 was permitted.These modifications are also supported by theoreticalassumptions. Where the scale has been entirely orpartly previously used (e.g. Byrne et al., 2007; Moksneset al., 2010a; Moksnes et al., 2010b; Liu & Lu, 2011),participants have been aged between 13 and 18 years.Because our interest was in the relationship betweenstress and adolescent drinking behaviours, and becauselocal research suggests that age 12 to 13 years is a keydevelopmental time in adolescent drinking (e.g. NISRA2008), we were anxious to include participants of thisage in the present cohort. Given that adolescence issuggested to range from age 13 to 18 years, some ofthose participants fall outside the strict definition ofadolescence. It is most probable that it was for thereason of the age profile of the sample that the stressof Emerging Adult Responsibility was problematic. Itemsfocus on the stress of new family responsibilities,
employment and, in particular, how it interferes withsocial activities. In the United Kingdom, young peoplemay only be in paid employment from age 13 yearsonwards and then only in limited employment sectors.For many therefore, employment-related stress mightnot be salient. In the Norwegian sample, two out ofthe three items from Emerging Adult Responsibility didnot load satisfactorily. The authors speculated this toresult from the fact that, by virtue of their age, mostof the sample would still live with parents and beprovided for by them (Moksnes et al., 2010b). Thiswould likely be true of the present cohort also.
The modification indices suggested that the residualcovariance constraints be relaxed between items 9 and19 (stress of not having money to buy the things thatyou want and/or need), suggesting that there may bea clear enough differentiation between ‘want’ and‘need’. Moreover ,the cross-loading of three of theitems onto different factors suggests that stressesexperienced by adolescents in one domain are alsoexperienced in others. For example, ‘Lack of Freedom’,which originally loaded onto the School/Leisure Conflictdomain, also loaded onto the Home Life domainherein, suggesting that arguments at home may relateto a perceived lack of freedom. ‘Having too muchhomework’, originally loaded onto the School/LeisureConflict domain, also loaded onto the SchoolPerformance domain herein, suggesting a perceived rela-tionship between homework and school performance.Finally, ‘Abiding by Petty Rules at School’, whichoriginally loaded onto the Teacher Interaction domain,also loaded onto the School Attendance domain herein,suggesting that abiding by petty rules relates as much tothe school system as to the individual teacher.
Stress has been shown to impact males and femalesdifferently, and sex differences have been found tocontribute significantly to differences in the relationshipbetween stress and emotional outcomes (Rudolph &Hammen, 1999; Rudolph, 2002; Byrne et al., 2007;Moksnes et al., 2010a). Males did not report significantlyhigher levels of stress in any domain in the present study.However, in respect of the three stress domains wherethere were no significant sex differences, these could besaid to be non-interpersonal domains, namely the stressof School Attendance, School/Leisure Conflict and Emerg-ing Adult Responsibility, supporting work elsewhere(Rudolph & Hammen, 1999). Previous studies usingthe ASQ have reported significant sex differences forthe majority of stress domains. Byrne and colleagues(2007) reported significant sex differences on all domainswith the exception of Financial Pressure, Teacher Interac-tion and School Attendance. In the Norwegian studies,results revealed significant sex differences on all but twodomains (not teacher interaction or school attendance)in one study (Moksnes et al., 2010b), and significantdifferences on all domains in the other study (Moksneset al., 2010a). These combined results support the theorythat some stressors may be universal across cultures,
Stress and Health (2014) © 2014 John Wiley & Sons, Ltd.
M. T. McKay, A. Percy and D. G. Byrne Adolescent Stress Questionnaire, Alcohol Use
while others may be more culturally specific (Plunkett,Radmacher, & Moll-Phanara, 2000).
Significant negative relationships between ASQdomains and self-esteem were observed for all butone domain herein, therefore offering support for thescales’ validity in the present sample. Additionally, anumber of the correlations between ASQ domainsand self-efficacy measures support the structure of theASQ. This is especially true as follows: the medium-sized negative correlations between stress of bothSchool Performance and School Attendance withacademic self-efficacy; the correlations between thestress of Home Life, School Performance, Peer Pressureand Future Uncertainty with social self-efficacy; andthe significant negative correlation between the stressof Peer Pressure and emotional self-efficacy.
Results of alcohol-related analyses reveal a domain-specific relationship between six of the stress domainsand drinking behaviours. Three of the stress domains(School Attendance, Home Life and Peer Pressure) wereconsistent statistical predictors across all models.Explaining some of these relationships is not easy,although the results for School Attendance, School Perfor-mance and Peer Pressure fit with the broader literature.Academic failure and/or underachievement, high levelsof truancy and a low degree of commitment to schoolhave been cited as risk factors for problematic use ofsubstances in adolescence (e.g. Best, Manning, Gossop,Gross, & Strang, 2006, Fothergill & Ensminger 2006),and the present results support the idea that those whofind school attendance more stressful are also involvedwith alcohol in a potentially more problematic way.Studies involving those excluded from school inNorthern Ireland have found this group to be at anincreased risk of drug use and antisocial behaviour dur-ing adolescence and later marginalization and exclusionfrom society in adulthood (e.g. McCrystal, Higgins andPercy, 2006, McCrystal, Percy and Higgins, 2007).
Research has linked peer pressure to a variety ofpotential problems including substance abuse (e.g.Velleman, 2009). The results of the present study sug-gest that those who have a potentially more problem-atic relationship with alcohol are also those who findpeer pressure less stressful, a finding not necessarily atodds with this broader peer pressure literature. The re-sults could easily be explained by the fact that thosewho succumb to peer pressure to drink alcohol areno longer resistant to it and therefore do not find it a
Stress and Health (2014) © 2014 John Wiley & Sons, Ltd.
stressful experience. Accordingly, it is only those thatattempt to resist the pressure that would find it stress-ful. This same logic might explain the result for thestress of School Performance in the third alcohol usemodel. Namely, that those children who typically drinkmost per occasion are perhaps also those who are leastconcerned by their school performance and thereforefind it less stressful.
Research elsewhere has pointed to the relationshipbetween general communication processes betweenparents and children mediating the effects of riskfactors on problematic behaviour (Riesch, Andersen,& Krueger, 2006), and better family communicationprocesses have been shown to be protective againstnegative alcohol-related outcomes in young people(e.g. Nash, McQueen, & Bray, 2005). Results hereinsuggest a positive relationship between the experienceof Home Life stress and greater consumption ofalcohol per drinking occasion.
Limitations
Relative to some other previously discussed studies intothe factor structure of the ASQ, the sample size in thepresent study was relatively small. Additionally, all datarelating to stress and other variables in the presentstudy were gathered using self-report measures. How-ever, all possible precautions were taken to guaranteeand maintain participant anonymity.
ConclusionsThe present study demonstrates that the 10-factor,56-item version of the ASQ (Byrne et al., 2007) is stablefor use in samples of English-speaking adolescents. Tothe best of our knowledge, this is the first examinationof the structure of this scale in another English-speakingpopulation. Results of the alcohol-specific analyses sup-port the domain-specific nature of the scale. Moreover,the relationship between ASQ scores and sex and scoreson self-concept measures supports the construct validityof the scale. However, results also demand furtherexamination of the scale in samples varying in age andin location. In particular, the factor measuring the stressof Emerging Adult Responsibility appears to be unstable insamples including younger adolescents.
Conflict of interest
The authors have declared that there is no conflict ofinterest.
REFERENCES
Allison, K. R., Adlaf, E. M., Ialomiteanu, A., & Rehm, J.
(1999). Predictors of health risk behaviors among
among young adults: Analysis of the National Popu-
lation Health Survey. Canadian Journal of Public
Health, 90, 85–89.
Best, D., Manning, V., Gossop, M., Gross, S., &
Strang, J. (2006). Excessive drinking and
other problem behaviors among 14–16 year
old schoolchildren. Addictive Behaviors, 31,
1424–1435.
Bray, R. M., Fairbank, J. A., & Marsden, M. E. (1999).
Stress and substance use among military women
and men. American Journal of Drug and Alcohol
Abuse, 25, 239–256.
Byrne, B. M. (2008). Testing for multigroup
equivalence of a measuring instrument: A
walk through the process. Psichothema, 20,
872–882.
Byrne, D. G., & Mazanov, J. (2001). Self-esteem, stress
and cigarette smoking in adolescents. Stress and
Health, 17, 105–110.
Byrne, D. G., Davenport, S. C., & Mazanov, J. (2007).
Profiles of adolescent stress: The development of the
adolescent stress questionnaire (ASQ). Journal of
Adolescence, 30, 393–416.
Adolescent Stress Questionnaire, Alcohol Use M. T. McKay, A. Percy and D. G. Byrne
Charbonneau, A. M., Mezulis, A. H., & Hyde, J. S.
(2009). Stress and emotional reactivity as explana-
tions for gender differences in adolescents’ depressive
symptoms. Journal of Youth and Adolescence, 38, 1050–
1058.
Deardorff, J., Gonzales, N. A., & Sandler, I. N. (2003).
Control beliefs as a mediator of the relationship
between stress and depressive symptoms among
inner city adolescents. Journal of Abnormal Child
Psychology, 31, 205–217.
De Vriendt, T., Moreno, L. A., & De Henauw, S.
(2009). Chronic stress and obesity in adolescents:
Scientific evidence and methodological issues for
epidemiological research. Nutrition, Metabolism, and
Cardiovascular Diseases, 33, 412–425.
De Vriendt, T., Clay, E., Moreno, L. A., Bergman, P.,
Vincente-Rodriguez, G., Nagy, E., Dietrich, S.,
Manios, Y., & De Henauw, S. (2011). Reliability and
validity of the Adolescent Stress Questionnaire in a
sample of European adolescents—The HELENA
study. BMC Public Health, 11, 717.
Diaz, A., Symantov, E., & Rickert, V. I. (2002). Effect of
abuse on health: Results of a National survey. Archives
of Pediatric and Adolescent Medicine, 156, 811–817.
Field, A. (2005).Discovering statistics using SPSS (2nd Ed.).
London: Sage Publications.
Flora, D. B., & Curran, P. J. (2004). An empirical evalua-
tion of alternative methods of estimation for confirma-
tory factor analysis with ordinal data. Psychological
Methods, 9, 466–491.
Fothergill, K. E., & Ensminger, M. E. (2006). Childhood
and adolescent antecedents of drug and alcohol
problems: A longitudinal study. Drug and Alcohol
Dependence, 82, 61–76.
Grant, K. E., Compas, B. E., Thurm, A. E., McMahon,
S. D., & Gipson, P.Y. (2004). Stressors and child
and adolescent psychopathology: Measurement
issues and prospective effects. Journal of Clinical
Child and Adolescent Psychology, 33, 412–425.
Gray-Little, B., Williams, V. S. L., & Hancock, T. D.
(1997). An item response theory analysis of the
Rosenberg Self-Esteem Scale. Personality and Social
Psychology Bulletin, 23, 443–451.
Gullette, L., & Lyons, M. A. (2006). Sensation seeking,
self esteem and unprotected sex in college students.
Journal of the Association of Nurses in AIDS Care, 17,
23–31.
Hall, A. S., & Torres, I. (2002). Partnerships in
preventing adolescent stress: Increasing self-esteem,
coping and support through effective counseling.
Journal of Mental Health Counseling, 24, 97–109.
Henderson, A. S., Byrne, D. G., & Duncan-Jones, P.
(1981). Neurosis and the social environment. Sydney:
Academic Press.
Horton, N. J., Kim, E., & Saitz, R. (2007). A cautionary
note regarding count models of alcohol consumption
in randomized controlled trials. BMC Medical
Research Methodology, 7, 9.
Hu, L., & Bentler, P. M. (1999). Cut-off criteria for fit
indexes in covariance structure analysis: Conventional
criteria versus new alternatives. Structural Equation
Modelling, 6, 1�55.
Jackson, S., & Goossens, L. (Eds.). (2006). Handbook
of adolescent development. Hove, UK: Psychology
Press.
King, K. M., Molina, B. S. G., & Chassin, L. (2009). Pro-
spective relations between growth in drinking and fa-
milial stressors across adolescence. Journal of
Abnormal Psychology, 118, 610–622.
Liu, Y., & Lu Z. (2011). The Chinese high school stu-
dent’s stress in the school and academic achievement.
Educational Psychology, 31, 27–35.
McCrystal, P., Higgins, K., & Percy, A. (2006). Brief re-
port: School exclusion drug use and delinquency in
adolescence. Journal of Adolescence, 29, 829–836.
McCrystal, P., Percy, A., & Higgins, K. (2007). Exclu-
sion and marginalisation in adolescence: The experi-
ence of school exclusion on drug use and antisocial
behaviour. Journal of Youth Studies, 10, 35–54.
Moksnes, U. K., Moljord, I. E. O., Espnes, G. A., &
Byrne, D. G. (2010a). The association between stress
and emotional states in adolescents: The role of
gender and self-esteem. Personality and Individual
Differences, 49, 430–435.
Moksnes, U. K., Byrne, D. G., Mazanov, J., &
Espnes, G. A. (2010b). Adolescent stress: Evaluation
of the factor structure of the Adolescent Stress
Questionnaire (ASQ-N). Scandinavian Journal of
Psychology, 51, 203–209.
Moksnes, U. K., & Espnes, G. A. (2011). Evaluation of
the Norwegian version of the Adolescent Stress Ques-
tionnaire (ASQ-N): Factorial validity across samples.
Scandinavian Journal of Psychology, 52, 601–608.
Murberg, T. A., & Bru, E. (2005). The role of coping
styles as predictors of depressive symptoms among
adolescents: A prospective study. Scandinavian Jour-
nal of Psychology, 46, 385–393.
Muris, P. (2001). A brief questionnaire for measuring
self-efficacy in youths. Journal of Psychopathology
and Behavioral Assessment, 23, 145–149.
Murray, K. M., Byrne, D. G., & Rieger, E. (2011). Inves-
tigating adolescent stress and body image. Journal of
Adolescence 34, 269–278.
Nash, S., McQueen, A., & Bray, J. H. (2005). Pathways
to adolescent alcohol use: Family environment, peer
influence and parental expectations. Journal of
Adolescent Health, 37, 19–28.
Natvig, G. K., Albrektsen, G., Anderssen, N., &
Qvarnstrom, U. (1999). School-related stress and
psychosomatic symptoms among school adolescents.
Journal of School Health, 69, 362–368.
Nicolai, K. A., Laney, T., & Mezulis, A. H. (2013).
Different stressors, different strategies, different
outcomes: How domain-specific stress responses
differentially predict depressive symptoms among
adolescents. Journal of Youth and Adolescence, 42,
1183–1193.
Northern Ireland Statistics and Research Agency
(NISRA) (2008). Young person’s behaviour and
attitudes survey. Belfast: NISRA.
Stress an
Nurmi, J. -E. (2002). An introduction: Thinking about,
preparing for and negotiating the future. In J.
Trempala & L-E. Malmberg (Eds.), Adolescents’
future-orientation. Oxford: Peter Lang.
Oppedal, B., & Røysamb, E. (2004). Mental health, life
stress and social support among young Norwegian
adolescents with immigrant and host national
background. Scandinavian Journal of Psychology,
45, 131–144.
Pasupathi, M., Staudinger, U. M., & Baltes, P. B. (2001).
Seeds of wisdom: Adolescents’ knowledge and judge-
ment about difficult life problems. Developmental
Psychology, 37, 351–361.
Perreira, K. M., & Sloan, F. A. (2001). Life events and
alcohol consumption among mature adults: A
longitudinal analysis. Journal of Studies on Alcohol,
62, 501–508.
Plunkett, S. W., Radmacher, K. A., & Moll-Phanara, D.
(2000). Adolescent life events, stress and coping: A
comparison of communities and genders. Professional
School Counseling, 3, 356–366.
Rappange, D. R., Brouwer, W. B. F., & van Exel, N. J. A.
(2009). Back to the Consideration of Future Conse-
quences Scale: Time to Reconsider? The Journal of
Social Psychology, 149, 562–584.
Riesch, S. K., Anderson, L. S., & Krueger, H. A. (2006).
Parent–child communication processes: Preventing
children’s health-risk behavior. Journal of the Society
of Pediatric Nurses, 11, 41–56.
Rosenberg, M. (1965). Society and the adolescent
self-image. Princeton, NJ: Princeton University
Press.
Rudolph, D. R. (2002). Gender differences in emotional
responses to interpersonal stress during adolescence.
Journal of Adolescent Health, 30, 3–13.
Rudolph, K. D., & Hammen, C. (1999). Age and gender
as determinants of stress exposure, generation, and
reactions in youngsters: A transactional perspective.
Child Development, 70, 660–677.
Skaff, M. M., Finney, J. W., & Moos, R. H. (1999).
Gender differences in problem drinking and depres-
sion: Different vulnerabilities? American Journal of
Community Psychology, 27, 25–54.
Suldo, S. M., & Shaffer, E. J. (2007). Evaluation of
the Self-Efficacy Questionnaire for Children in
two samples of American adolescents. Journal of
Psychoeducational Assessment, 25, 341–355.
Udry, J. R., Li, R. M., & Hendrickson-Smith, J. (2003).
Health and behaviour risks of adolescents with mixed
race identity. American Journal of Public Health, 93,
1865–1870.
Velleman, R. (2009). How do children and young
people learn about alcohol: A major review of the
literature. Joseph Rowntree Foundation. Retrieved from
http://www.bath.ac.uk/health/mhrdu/
RVJRFreviewfinalpdf2009.pdf.
West, P., & Sweeting, H. (2003). Fifteen, female and
stressed: Changing patterns of psychological distress
over time. Journal of Child Psychology and Psychiatry,
44, 399–411.
d Health (2014) © 2014 John Wiley & Sons, Ltd.