Running Head: Development Of The Parent -Adolescent Relationship . - QUT

Running head: Development of the Parent-Adolescent Relationship Scale6outcomes, including better academic outcomes (Chen & Gregory, 2009; Wang & Sheikh-Khalil, 2014),greater self-regulatory capacity (Purdie et al., 2004), emotional wellbeing (Bulanda & Majumdar,2009; Wang & Sheikh-Khalil, 2014), and reduced alcohol use (Nash et al., 2005) although the causalnature of these relationships is less well examined. One critical aspect of parental involvement is theextent to which a parent and their child spend time together in shared activities, that are enjoyed byboth members of the relationship (e.g., family mealtimes, family events or shared leisure activities;(Aquilino, 1997; Hair et al., 2008). The concept of shared activities may be particularly relevant acrossadolescence and emerging adulthood. Hair et al. (2008) found that adolescent perceptions of thequality of the relationship with their parent in early adolescence strongly predicted adolescentparticipation in shared activities in mid-adolescence, which in turn appeared to protect againstdelinquent behaviours and promote psychological wellbeing in late adolescence.Conversely, parental hostility, criticism or rejection in the parent-adolescent relationship isrelated to later emotional and behavioral difficulties (Fanti et al., 2008; Miranda et al., 2016; Sentseet al., 2009), alcohol use (Diggs et al., 2017) and suicidal ideation and attempts (Fotti et al., 2006).Importantly, several longitudinal studies have documented that when an adolescent feels distant,unloved and rejected by their parents, they are at increased risk of poor psychological adjustment inlate adolescence and adulthood, including depression, conduct behavior problems, substance abuseand suboptimal brain development (Rohner & Britner, 2002) and greater risk of depression. In onewell-controlled, longitudinal study of 1,247 families from nine countries, Putnick et al. (2015) foundthat maternal and paternal hostility, rejection and indifference predicted high risk of adolescentinternalizing and externalizing behaviors, and poorer school performance and social competencethree years later. While we acknowledge that there are likely to be reciprocal associations betweenparent and adolescent behaviour and subsequent outcomes for young people, improving the parentadolescent relationship may be effective at reducing the risk of poor outcomes even if the exactdirection of influence is unknown.

Running head: Development of the Parent-Adolescent Relationship Scale7In summary, there is overwhelming evidence that connectedness and involvement,specifically participating in shared activities, in the parent-adolescent relationship play a central rolein supporting adolescent development, whereas parent-adolescent hostility has detrimental effectson adolescent outcomes. However, there is variability in the quality of the measurement of theseconstructs, with a recent review highlighting there are serious limitations in the psychometricproperties of available measures in this area (Hurley et al., 2014). The parent-adolescent relationshipis often assessed from the perspective of either the parent or the adolescent, which is problematicgiven the increasing reciprocity in the parent-adolescent relationship over time, particularly asadolescents move into emerging adulthood (Aquilino, 1997). Recent work indicates that combinedparent and adolescent reports are the gold standard in providing the best source of informationregarding adolescent functioning in clinical and community samples (Kuhn et al., 2017). However,few measures allow multiple perspectives to be examined and those that do have other limitations.Existing Measures of Parent-Adolescent Relationships and Their LimitationsThere are a number of existing measures that have been used to assess the parentadolescent relationship, including the Inventory of Parent and Peer Attachment (IPPA, Armsden &Greenberg, 1987), the Parental Bonding Instrument (PBI, Parker et al., 1979), Adolescent FamilyProcesses scale (AFP, Vazsonyi et al., 2003) and the Parental Acceptance-Rejection Questionnaire(PARQ, Rohner & Khaleque, 2005). However, there are theoretical, psychometric and practicallimitations of these measures that mean they are less suitable for use in large-scale evaluationresearch, including: 1) they were developed as adolescent-report measures with no parallel parentreport form (e.g., PBI, Heider et al., 2005; AFP, Vazsonyi et al., 2003); 2) they are either costly (e.g.,the PARQ which is 5.68 per parent-child dyad administration) or lengthy and time intensive (e.g.,the PARQ has 60 items; 3) they do not assess emerging adults’ relationship with their parents (e.g.,PBI assesses the relationship for children under 16 years, Parker et al., 1979); 4) they have variablepsychometrics and/or inconsistent factor structures across different samples and studies (e.g., IPPA,Pace et al., 2011), or 5) use 4-point scales that are less effective at detecting subtle intervention

Running head: Development of the Parent-Adolescent Relationship Scale8effects due to respondents’ tendency to avoid endorsing scale anchors (PARQ, Gomez & Rohner,2011; PBI, Parker et al., 1979).Aims of the Two Studies Presented in this PaperThis paper reports two studies that describe the development and validation of a newmeasure of the parent-adolescent relationship, the Parent-Adolescent Relationship Scale (PARS). Theaim of this process was to develop a measure that: a) assessed both parent and adolescentperspectives with appropriate invariance established; b) is brief and easy to administer; c) is freelyavailable; d) has strong psychometric properties including reliability, validity and change sensitivity,and e) can be used in both research and clinical settings. Study 1 reports on the development, initialvalidation and psychometric properties of the parent-report version of the PARS. Study 2 comprisestwo parts. Part 1 used a separate sample of parents to undertake a confirmatory factor analysis ofthe draft PARS to further establish its construct validity. Part 2 assessed the factor structure andvalidity of the PARS for use by adolescents and emerging adults to describe the relationship withtheir parent. Part 2 also assessed the equivalence of the three forms of the PARS (parent, adolescentand emerging adult) via invariance testing.Study 1: Development and Validation of the Parent-Report Version of the PARSAims and HypothesesStudy 1 describes the scale construction process and initial validation of the parent-reportversion of the PARS. Specifically, we hypothesized a 3-factor structure with scores possessing goodinternal consistency (H index .70). We predicted that the scale would comprise factors that assessthe extent to which parents report: 1) experiencing a warm, loving and accepting relationship withtheir adolescent (connectedness); 2) being actively and practically involved in their adolescent’s life(shared activities and involvement); and 3) adverse or negative behaviors directed towards theadolescent, such as criticism or complaining (hostility). We tested the convergent and concurrentvalidity of the interpretations of PARS scores by examining its associations with existing measures ofthe parent-adolescent relationship and parenting practices. We hypothesized that scores on positive

Running head: Development of the Parent-Adolescent Relationship Scale9dimensions of the PARS would be positively correlated with scores on tests of family cohesion(measured by the Family Environment Scale [FES]) and positive parenting (measured by the AlabamaParenting Questionnaire [APQ]) and, negatively correlated with scores on tests of family conflict(measured by the FES), parent-adolescent conflict (measured by the Conflict Behavior Questionnaire[CBQ]) and poor supervision and inconsistent parenting (measured by the APQ). We also predictedthat scores on the negative dimension of the PARS would be positively correlated with family andparent-adolescent conflict, poor supervision and inconsistent parenting, and negatively correlatedwith family cohesion and positive parenting. Finally, to assess discriminant validity of PARS scoreinterpretations, we predicted that there would be no correlation between scores on subscales of thePARS and parental technology use.MethodParticipantsIn total, 256 parents of young people aged 11 to 18 years completed the draft PARS in twoseparate studies conducted from March 2014 to June 2015: 1) a cross-sectional survey of familyrelationships and adolescent adjustment (n 152); and 2) baseline data from a randomizedcontrolled trial (RCT) of a 2-hour group on family conflict for parents of adolescents aged 11 to 16years (n 104; Dittman et al., 2020). The use of multiple data sources ensured the sample comprisedboth community and support-seeking parents. Recruitment for both studies involved the distributionof flyers to secondary schools and community agencies (e.g., General Practice Medical Centres,community centres, local business, family support services). In addition, organic and paid posts wereused to boost recruitment (e.g., posts on a dedicated Facebook page, requests to parent forums andbloggers). Eligibility criteria for both studies required participants to be the parent of an adolescent(11-18 years for the cross-sectional study; 11-16 years for the RCT) and have sufficient English toparticipate. In addition, to be eligible for the RCT study, parents needed to indicate that they wereconcerned about conflict with their adolescent during an initial screening telephone call, while

Running head: Development of the Parent-Adolescent Relationship Scale10parents were excluded if they were seeking psychological or medical assistance for mental health orpartner-related issues, or who had a child with intellectual or developmental disorders.The combined sample ranged in age from 26 to 67 years (M 45.49; SD 6.02) and weremostly mothers (84%). Target adolescents were aged between 11 and 18 years (M 13.82; SD 1.82)and 51% were female. Most parents identified their cultural background as Australian (74%), with thelargest other cultural representations being European (15%) and New Zealander (5%). In terms ofhighest education level, 6% of parents had not completed secondary school, 12% had a secondaryschool education, 22% had a trade or technical/vocational college qualification, and 60% wereUniversity-educated. Adolescents mostly lived in two-parent (63%) or sole-parent (23%) families.Data from 40 parents in the waitlist condition from the RCT were used to assess test-retest reliability.These parents completed the PARS before randomization and again 6 to 8 weeks later, before theparenting group was offered.ProcedureEthics approval was obtained from the authors’ University Human Research EthicsCommittee. Parents provided consent and completed the survey online. Participants with more thanone child in the target age range responded in relation to the adolescent with the next birthday. Inthe RCT, interested parents contacted the researchers and participated in a telephone screeninginterview before providing consent and completing the survey online. Parents who chose tocomplete the survey in hard copy had the assessment sent to them for completion and returnedcompleted consent forms and surveys by mail. RCT parents completed questionnaires in relation tothe adolescent for whom they were seeking support.Scale ConstructionDevelopment of the draft PARS was conducted according to the Standards for Educationaland Psychological Testing (American Educational Research Association et al., 2014): 1) definition ofpurpose of the measure; 2) review of theory and literature related to parenting and the parentadolescent relationship to ascertain key constructs for measurement; 3) item generation of

Running head: Development of the Parent-Adolescent Relationship Scale11positively- and negatively-worded items based on the literature review and clinical experience of theauthors; 4) determining measure structure (e.g., item complexity, response format); 5) expert andconsumer review of draft items; 6) integration of feedback and assembly of draft measure.Following generation of a set of 54 items by the authors, 13 parents and seven teenagerswere recruited independently from the two main studies to provide feedback on item relevance,clarity and difficulty. Parents and teenagers were recruited via advertising in the authors’ universitynewsletter. In addition, six experts in adolescence and/or parenting (clinical and research) providedfeedback on face validity, theoretical coherence, wording and item clarity with an additional eightprofessionals with expertise in parenting and/or adolescence from diverse cultural backgrounds (i.e.,Panama, El Salvador, Switzerland, Germany, China, India) also providing feedback on cross-culturalrelevance. Items that were too complex, not relevant to the hypothesised constructs or not crossculturally appropriate were removed, and retained items were refined for clarity. This processresulted in 45 items hypothesised to assess three constructs: connectedness (behaviors reflective ofcloseness and intimacy, warmth and affection, and acceptance and approval); shared activities(behaviors related to reciprocal activities between parents and adolescents); and hostility (behaviorssuggestive of negativity, rejection and criticism in the relationship).Scale ValidationThe 45-item PARS was included in a questionnaire battery. Parents rated each statement as itapplied to their relationship with their adolescent on a 6-point Likert scale ranging from 0 (not at alltrue) to 5 (nearly always or always true). The survey also contained demographic items (i.e., parentage, gender, educational level, employment status, country of birth; family structure and income;and target adolescent age and gender) and the measures described below to assess PARS validity.Validation MeasuresValidity of the PARS score interpretations were investigated using several published parentreport measures designed to assess aspects of the parent-adolescent relationship and parenting

Running head: Development of the Parent-Adolescent Relationship Scale12practices, along with a measure of adolescent media use. All alphas reported below are based on thecurrent data.Criterion Validity (Concurrent). Two established measures of the parent-adolescentrelationship were used. The Family Environment Scale (FES; Moos & Moos, 1994) assessed overallrelationship functioning in the family. In the present sample, both the 9-item Conflict (α .80) andCohesion (α .75) subscales had adequate internal consistency. The 20-item Conflict BehaviorQuestionnaire (CBQ; Robin & Foster, 1989) assessed the level of conflict in the parent-adolescentrelationship, with strong internal consistency for the total score using the present data, α .94. Bothquestionnaires asked parents to rate whether each item was true or false.Construct Validity (Convergent). Along with the FES and CBQ, Alabama ParentingQuestionnaire’s 9-item short form (APQ; Elgar et al., 2007) was used to assess the construct validityof interpretations of scores on the PARS. Adapted from the original APQ (Shelton et al., 1996), theshort form assessed effective and ineffective parenting practices via three, 3-item subscales. Eachitem described a parenting practice. Parents rated how often they typically used each practice on a5-item scale from 1 (never) to 5 (always). In the present sample, all subscales possessed adequateinternal consistency; Positive Parenting (α .80); Inconsistent Discipline (α .78) and PoorSupervision (α .72).Construct Validity (Discriminant). Four subscales from the Media and Technology Usage andAttitudes Scale (MTUA; Rosen et al., 2013) were used. The subscales assessed frequency of use offour modes of technology on a scale from 1 (never) to 10 (all the time). All four subscales had goodinternal consistency in the present sample: email (α .80); smart phone (α .92); internet searching(α .89) and general social media (α .95).Data AnalysisDescriptive statistics were examined to ensure items had enough variability to be suitable forfactor analysis (Tabachnick & Fidell, 2007). Exploratory factor analysis (EFA) using principal axisfactoring was used to identify the factor structure of the PARS and reduce the draft version to a

Running head: Development of the Parent-Adolescent Relationship Scale13smaller, more practical number of items. Suitability of the data for EFA was determined viainspection of the correlation matrix for values greater than .30 (i.e., Factorability of R), inspection ofthe Kaiser-Meyer-Olkin (KMO) Measure of Sampling Adequacy for a value greater than .5, and asignificant (p .05) Bartlett's Test of Sphericity (Tabachnick & Fidell, 2007; Williams et al., 2010).Because it was assumed that the factors would be correlated, oblique (Direct Oblimin) rotation wasperformed and the rotated solution interpreted. Determination of the number of factors to extractwas done using a combination of Kaiser’s criterion (identification of Eigenvalues greater than one),visual inspection of the scree plot (Cattell, 1966) and parallel analysis (Horn, 1965). Parallel analysis(PA) was conducted using the Monte Carlo PCA for Parallel Analysis for-Parallel-Analysis/) for testing the probability thata factor is due to chance (Wood et al., 2015). PA requires that a set of random correlation matricesbe generated based upon the same number of variables and participants as the experimental data.The output is then compared to the eigenvalues produced by the experimental data, and thecriterion for number of factors extracted is the number at which the eigenvalues generated byrandom data exceed the eigenvalues produced by the experimental data (Watkins, 2006).Pearson’s correlations assessed the criterion and construct validity of PARS test scoreinterpretations and the test-retest reliability of PARS scores. Due to limitations associated withCronbach’s alpha when the assumptions of tau-equivalence and/or uncorrelated errors are violated,internal consistency was assessed using the H coefficient (Hancock & Mueller, 2001), the range andinterpretation of which is the same as for Cronbach’s alpha. Descriptive statistics were calculated forthe full sample and by age group (i.e., 11 to 14 years and 15 to 18 years).ResultsExploratory Factor AnalysisMissing data were imputed using the Expectation-Maximisation method as there was veryminimal missing data ( 3%). Since many correlations were greater than .3, the initial KMO was .93and Bartlett’s Test of Sphericity was significant (χ2 (946) 7211.92, p .001), the data were deemed

Running head: Development of the Parent-Adolescent Relationship Scale14suitable for factor analysis. EFA on the 45 draft items produced seven factors with eigenvalues above1. However, the Pattern Matrix failed to converge, inspection of the Scree plot (see Figure 1)suggested a 3- or 4-factor solution, and parallel analysis supported a 4-factor solution (Tabachnick &Fidell, 2007; Watkins, 2006). A forced 4-factor solution expl

child relationship into emerging adulthood (Nelson and Padilla-Walker (2013). Parental . involvement. in an adolescent's life, characterised by investment of time and resources into an adolescent's school and leisure activities , is predictive of positive adolescent