Journal of Adolescence 51 (2016) 50e57

Contents lists available at ScienceDirect

Journal of Adolescence
journal homepage: www.elsevier.com/locate/jado

Development of the Personal Responsibility Scale for

adolescents
Amanda Mergler*, Paul Shield
Queensland University of Technology, Australia

a r t i c l e i n f o a b s t r a c t

Article history: Increasing interest by policy makers and educators in enhancing the non-cognitive factors
Available online 11 June 2016 that impact on the wellbeing of young people across the school years requires valid and
reliable measurement scales. This paper reports the psychometric development of the
Keywords: Personal Responsibility Scale drawing on a preliminary 34-item measure developed using
Personal responsibility data from focus groups with secondary students in Australia. This scale was then
Adolescence
administered to 513 students across Years 9e12. Exploratory factor analysis identified
Exploratory factor analysis (EFA)
three factors, ‘personal accountability’, ‘behavioural and emotional control’, and ‘cognitive
Structural equation modelling (SEM)
Non-cognitive skills
control’. Structural equation modelling was used to examine the congeneric measurement
models for these factors. The multidimensional nature of the personal responsibility
construct was then examined through multilevel confirmatory factor analysis. These an-
alyses supported a three-factor structure to propose a final 15-item scale. Additional
validation requirements, future research possibilities, and implications of this study are
addressed.
© 2016 The Foundation for Professionals in Services for Adolescents. Published by Elsevier
Ltd. All rights reserved.

Introduction

Internationally, interest in the social and emotional development of adolescent students has received increased attention
in recent years because of increased awareness of the complexity of the worlds in which young people currently live (Glass,
2013; Trosset, 2013; West et al., 2014). Educational policy has increasingly emphasised the importance of students' abilities to
be resilient, demonstrate self-control, and demonstrate social and personal responsibility (ACARA, 2015; Lovat, Clement,
Dally, & Toomey, 2010; MCEETYA, 2008). These abilities are often labelled non-cognitive skills in the research literature
(Duckworth & Yeager, 2015) and there is growing evidence that these skills can be taught to young people (O'Neill, 2012;
Piquero, Jennings, & Farrington, 2010). However, Duckworth and Yeager (2015) argue that robust measures of non-
cognitive skills are lacking and this hampers research efforts in this important area. Without the development of robust
measures to quantify such constructs as personal responsibility, researchers and schools are limited in their ability to
demonstrate the effectiveness (or otherwise) of their intervention efforts to develop these skills. This foundational study
details the development of a measure of personal responsibility for adolescents.

* Corresponding author. Faculty of Education, Queensland University of Technology, Victoria Park Road, Kelvin Grove, 4059 Brisbane, QLD, Australia. Tel.:
þ61 7 3138 3308; fax: þ61 7 3138 3987.
E-mail address: a.mergler@qut.edu.au (A. Mergler).

http://dx.doi.org/10.1016/j.adolescence.2016.05.011
0140-1971/© 2016 The Foundation for Professionals in Services for Adolescents. Published by Elsevier Ltd. All rights reserved.
 A. Mergler, P. Shield / Journal of Adolescence 51 (2016) 50e57 51

Definition and measurement of personal responsibility

Within the educational and psychological literature, the concept of personal responsibility has been discussed since at
least the 1950s. Despite this, a clear definition of the construct has been difficult to determine, largely due to the disparate
nature of earlier work. An early theoretical framework within which to understand personal responsibility was developed by
Heider (1958), outlining that personal responsibility is determined on the basis of causality (what was done) and expectations
(what should have been done). More recently, a new perspective was outlined by Schlenker, Britt, Pennington, Murphy, and
Doherty (1994), who conceptualised the Triangle Model of Responsibility which highlighted that the construct involved
combining the strength of three elements being prescriptions (rules for conduct), events (units of action), and identity (actor's
roles, qualities, commitments, and pretentions). Alternatively, within the context of enhancing personal and social re-
sponsibility in American school students through physical education classes, Hellison (1985, 2003) proposed a values-based
model called the Teaching Personal and Social Responsibility model (TPSRM). Within this model, personal responsibility
included demonstrating self-control, participating and persisting (effort) in familiar and new activities, and investing and
taking increased responsibility for one's physical activity levels and overall wellbeing.
Definitions of personal responsibility have also varied across time and research studies, making comparisons difficult. For
example, Mergler (2007, p.66) defined personal responsibility as “the ability to regulate one's own thoughts, feelings and
behaviour, along with a willingness to hold oneself accountable for the choices made and the social and personal outcomes
generated”. In earlier work in education, Lickona (1991, p. 68) defined personal responsibility broadly, outlining that the
construct was about “taking care of self and others, fulfilling our obligations, contributing to our communities, alleviating
suffering, and building a better world”. Focussing on more immediate classroom behaviours, Lewis (2004) defined personal
responsibility as the frequency with which students engaged in responsible classroom behaviours including protecting
students' and teachers' rights associated with learning and emotional and physical safety. More recently, Smithkirai,
Longthong, and Peijsel (2015) defined personal responsibility as a person's belief that one is the master of one's own life;
one is aware of and chooses one's own choices and goals; and one is willing to hold oneself accountable for one's behaviour
and its consequences.
The empirical research examining personal responsibility is sparse (Linley & Maltby, 2009). Of the scales that do exist,
most have been created for use with adults (Martel, McKelvie, & Standing, 1987; Singg & Ader, 2001; Smithkirai, Longthong, &
Peijsel, 2015). A number of efforts to develop measures have been within the context of physical education (Lee, Kim, & Kim,
2012; Li, Wright, Rukavina, & Pickering, 2008; Watson, Newton, & Kim, 2003). In addition, most researchers have viewed
personal responsibility as a unidimensional construct (Martel et al., 1987; Singg & Ader, 2001). Swaner (2005) argued that
personal responsibility needs to be viewed as a multidimensional construct that includes behavioural, cognitive, emotional
and moral components. Advancement of this field of inquiry can only occur when researchers clearly identify what aspects of
personal responsibility they are seeking to measure, and develop a consistent definition and a strong measure that allow for
cross-study comparisons of findings across contexts that involve different populations.
Within the Australian education system, government policy demands that schools seek to enhance the personal re-
sponsibility of students (ACARA, 2015; MCEETYA, 2008). Indeed, it is suggested that a central focus of schooling should be the
development of young people who demonstrate self-awareness, social consideration and the willingness to take re-
sponsibility for the choices they make. A focus in education on the importance of personal responsibility is welcome,
however, we can only measure, understand and develop these capacities of young people when we have a clear under-
standing of what the construct involves as well as reliable and valid measures.

The current study

The current study extends our earlier work (Mergler, 2007; Mergler & Patton, 2007) to develop a Personal Responsibility
Scale for use with adolescents. While there are increasing expectations that schools can develop such non-cognitive skills in
young people, the outcomes of their efforts to do so can only be assessed when well developed and empirically robust
measurement scales are available. The current study documents the psychometric properties of the Personal Responsibility
Scale.

Method and results

Item generation

The initial set of items created for the measure were drawn from focus group data of 20 Australian Grade 11 students, of
which 13 were female and 7 were male, and 10 secondary school teachers, of which 7 were female and 3 were male. Targeted
questions were used to generate discussion including, ‘If you broke personal responsibility down into its key parts, what would
they be?’ and ‘What in your life do you take personal responsibility for?’ Each focus group was audiotaped, and the lead author of
the current paper transcribed each focus group, and used Krueger’s (1998) content analysis continuum model to group
common notions and identify key themes. Five key themes were identified, indicating that for these participants personal
responsibility included internal accountability, a consideration of consequences, self-management and self-control, cognitive
awareness, and social responsibility. As these areas were emphasised by the participants they were targeted as key areas from
52 A. Mergler, P. Shield / Journal of Adolescence 51 (2016) 50e57

which to draw items for the current measure, and some quotes offered by the participants were used as items. It was expected
that doing so would assist the measure in being relevant and applicable to the lives of adolescents.

Expert panel review

A draft 34-item measure was submitted to an expert panel of three for consideration, all of whom were experienced
researchers in adolescent developmental psychology, and two of whom had taught previously in secondary schools. Feedback
from this panel led to minor adjustments to the wording of various items in order to make the vocabulary used more un-
derstandable for adolescents.

Draft questionnaire administration

Participants
Participants were 513 predominantly middle-class Western secondary school students enrolled in Grades 9 to 12 in two
public schools in Queensland, Australia. Roughly equal numbers of female (n ¼ 263) and male (n ¼ 250) students were
included. The schools were suburban and situated in a medium-sized city. The students had a mean age of 14.74 years
(SD ¼ 1.17 years).

Measure
The draft Personal Responsibility Scale comprised 34 items that measure aspects of personal responsibility and utilised a
Likert scale with 1 being strongly disagree and 4 being strongly agree. Participants were asked to circle the response that
appears true of them most of the time. Examples of items include, ‘I often lash out when I am all stirred up’, and ‘I choose how to
respond in situations’. While it is recognised the Likert scale used is technically an ordinal scale, in this context it is treated as
being of interval level of measurement where the distributional properties allow which is in line with common practice in
educational research (Lehman, 1991).

Procedure
The 34-item questionnaire and demographic questions (e.g., age and gender) were administered to students by classroom
teachers, who had been provided with instructions regarding the protocol for administration. Parents and students received
an information sheet outlining the nature of the study. Students signed their consent to participate and were informed that
they were free to withdraw at any time without penalty. All students chose to participate in the study. Ethical approval was
obtained from Education Queensland and the Queensland University of Technology before the study commenced. .

Data examination and descriptives

A missing value analysis conducted on the 513 questionnaires identified that less than 1% of the data set was missing. No
patterns were observed in the missing data and as a consequence missing data values were not imputed (Pallant, 2013) and
where necessary items were removed listwise. As 10 items on the Personal Responsibility Scale were negatively scored, these
were recoded into new variables to reflect a positive emphasis in line with the remainder of the item set (Coakes & Steed,
2003). The minimum and maximum values for each variable lie in the scale range of 1e4 respectively, suggesting the
absence of data entry anomalies. The mean scale scores on each variable ranged from 2.34 to 3.20 on a 4 point scale with
standard deviations ranging from .599 to 1.55, suggesting range of response was adequate across the item set. Three variables
were shown to have issues with skewness and kurtosis (items 25, 30, and 19) (West, Finch, & Curran, 1995). Tabachnick and
Fidell (2007) indicate that when factor analysis is exploratory in nature, and will be followed by a CFA, it is acceptable to retain
items showing skewness and kurtosis. As such the decision was made to retain the items in the initial analysis in the current
study.

Exploratory factor analysis (EFA)

Data analysis
In SPSS version 21, a Maximum Likelihood Analysis (MLA) was used as the method of extraction, with a varimax rotation
(Tabachnick & Fidell, 2007). Examining the correlation matrix indicated that many correlations were above .3 and were
statistically significant, and most items correlated with three or more other items.
Barlett's test value for the correlation matrix was 4812.394 with a significance level of <.001. This suggests that there are
large correlations among the variables. The Kaiser-Meyer-Olkin measure verified the sampling adequacy for this analysis,
KMO ¼ .89, indicating that there is likely a factor structure underlying the items. An examination of the squared multiple
correlation coefficients (R2) or communalities indicated a reasonable level of variation in the items was being explained by the
latent factors with fifty-three percent of communalities >.4.
The factor analysis identified eight factors with eigenvalues greater than 1 accounting for 43% of the variance. Examination
of the scree plot indicated three distinct factors. Horn’s (1965) parallel analysis indicated that four factors should be retained,
however examination of the rotated factor matrix showed that only two items loaded on the fourth factor. Thus three factors
were retained which accounted for 29% of the variance. Table 1 shows each item and its factor loadings for factors 1, 2 and 3.
 A. Mergler, P. Shield / Journal of Adolescence 51 (2016) 50e57 53

Table 1
Questionnaire items and their factor loading on factors 1, 2 and 3.

Items Factor 1 Factor 2 Factor 3

34. I treat others with respect because that is how I would like to be treated. .557
32. I have set goals and believe in working hard to meet them .502
24. I think of the consequences of my actions before doing something. .499
11. When it comes to my behaviour I have set guidelines that I expect myself to follow. .496
3. I want my actions to help other people. .495
2. When I have done the wrong thing, I accept the punishment. .482
27. I often think about what events are coming up and ensure .407
I have everything I need to do well in these events.
33. If I have said or done something that has hurt someone else, .406
I am prepared to put things right
4. I am aware of how my behaviour impacts on other people. .363
6. I often lose my temper and am unable to control my behaviour. .740
17. Sometimes people make me so mad that I cannot control my behaviour. .729
5. I often lash out when I am all stirred up. .586
15. I cannot control my behaviour. .550
20. My emotions come out whenever they feel like it and there is little I can do about it. .521
23. I am always getting in trouble for things that aren't my fault. .459
12. If someone upsets me, it is not my fault if I am mean to them. .418
7. People often make me feel emotions like anger or sadness. .395
13. Friends can often talk me into doing things that I know may not be right. .357
14. It is up to me to control my behaviour. .494
10. I choose how to respond in situations. .482
16. When I am angry or sad I can usually work out why. .456
9. I can choose how I behave. .456
1. When making decisions, I decide for myself what is the best thing to do. .429

Internal consistency assessment of factors

Reliability was plumbed using Cronbach's alpha (Tabachnick & Fidell, 2007). For factors 1, 2, and 3, Cronbach's alpha was
shown to be .81, .81, and .71, respectively, and was considered adequate (George & Mallery, 2014). A contextual analysis of the
item sets showed that the items in factor 1 grouped around a common theme of personal accountability, items in factor 2
grouped around a common theme of behavioural and emotional control, and items in factor 3 grouped around a common
theme of cognitive control.

Confirmatory factor analysis (CFA)

Chi square goodness of model fit statistics were generated for each model and used as the prime decision mechanism for
model acceptance or rejection. The decision to accept or reject the model was also informed by the examination of absolute fit
and comparative fit indices, including the Root-Mean-Square Error of Approximation (RMSEA) and the Comparative Fit Index
(CFI). The following cut-offs were applied when judging model fit: Chi Squared Goodness of Fit (p > .05), Bollen-Stein
bootstrap (p > .05), RMSEA (<.05), CFI (>.95).

Participants
As the chi-square test is sensitive to large sample size, a smaller sample of cases were randomly selected from the larger
participant pool to serve as the calibration sample upon which the model would be built. A random sample of 250 cases
constituting the calibration group were selected for each factor. The remaining participants were used in a validation sample
to test the robustness of the model developed.

Data analysis and results

Before conducting the Maximum Likelihood CFA in AMOS version 22, testing for multivariate normality was conducted by
calculating Mardia's coefficient, a measure of multivariate kurtosis (Mardia, 1970). The sample for each factor showed the
assumption of multivariate normality to be violated (factor 1: Mardia's coefficient ¼ 42.137; c.r. ¼ 32.769; factor 2: Mardia's
coefficient ¼ 16.839; c.r. ¼ 13.286; factor 3: Mardia's coefficient ¼ 16.494; c.r. ¼ 22.194). As a consequence, a Bollen-Stine
bootstrap was applied to estimate bias free parameters (Bollen & Stine, 1992). In order to further refine each factor and
demonstrate unidimensionality (Anderson & Gerbing, 1982; Hunter & Gerbing, 1982), three single factor congeneric mea-
surement models were tested. The process of conducting congeneric measurement models after an EFA, and before a second-
level CFA, is essential for valid and reliable measurement development (Anderson & Gerbing, 1988).

CFA for factor 1

The model for factor 1 converged and was a good fit, Bollen-Stine bootstrap p ¼ .507, c2 (27) ¼ 38.152, p ¼ .075,
RMSEA ¼ .029 (.000, .049), CFI ¼ .988. The standardised residual covariances were examined, and it was revealed that item 4
54 A. Mergler, P. Shield / Journal of Adolescence 51 (2016) 50e57

was not fitting overly well with the other items, and that it had the lowest regression weight (.45). A contextual analysis of the
items that comprised factor 1 showed that item 4 (I am aware of how my behaviour impacts on other people) seemed
theoretically at odds with the remaining items, which focused on the notion of personal accountability. For these reasons,
item 4 was removed and the model was tested again. The new model was found to have a better fit than the original, Bollen-
Stine bootstrap p ¼ .449, c2 (20) ¼ 28.682, p ¼ .094, RMSEA ¼ .030 (.000, .053), CFI ¼ .989. Construct reliability was .53 and
the variance extracted was .44 (44%). Coefficient H was calculated to be .810 and the critical ratio of the parameter estimates
were all significant, indicating convergent validity. Based on the context underpinning the eight items that comprise this
factor, this factor was named “personal accountability”.

CFA for factor 2

The model for factor 2 converged but did not fit the data well, Bollen-Stine p ¼ .006, c2 (27) ¼ 62.256, p ¼ .000,
RMSEA ¼ .052 (.035, .068), CFI ¼ .966. The standardised residual covariances were explored, and indicated item 13 (Friends
can often talk me into doing things that I know may not be right) was not fitting well contextually with the other items. In
addition this item had the lowest regression weight (.38). Item 13 was deleted and the model was tested again.
Once again the model converged but did not fit the data well, Bollen-Stine p ¼ .016, c2 (20) ¼ 46.879, p ¼ .001,
RMSEA ¼ .052 (.033, .072), CFI ¼ .972. The standardised residual covariances were explored, and it was shown that item 23 (I
am always getting in trouble for things that aren't my fault) was not fitting well contextually with the other items and had a
low regression weight (.47). Item 23 was deleted and the model was tested again.
The testing of the model with the remaining seven items found that the model converged and fit the data well, Bollen-
Stine p ¼ .321, c2 (14) ¼ 21.147, p ¼ .098, RMSEA ¼ .032. (.000, .059), CFI ¼ .992. Construct reliability was .55 and the vari-
ance extracted was .47 (47%). Coefficient H was calculated to be .837 and the critical ratio of the parameter estimates were all
significant, indicating convergent validity. Based on the context underpinning the seven items that comprise this factor, this
factor was named “behavioural and emotional control”.

CFA for factor 3

The model for factor 3 converged but did not fit the data well, Bollen-Stine p ¼ .012, c2 (5) ¼ 20.081, p ¼ .001,
RMSEA ¼ .032 (.000, .099), CFI ¼ .997. The standardised residual covariances were explored, and it was shown that item 1
(When making decisions, I decide for myself what is the best thing to do) was not fitting well with the other items and had the
lowest regression weight (.40). While this item did fit contextually, an analysis of its underlying distributional properties
showed the item to be heavily skewed (.520; Zs ¼ 3.35) and kurtosed (2.054; Zk ¼ 6.63). In addition, both the Kolmogrov-
Smirnov (<.001) and Spapiro-Wilk (<.001) tests were significant, indicating non-normality. For these reasons, item 1 was
deleted and the model was tested again. The testing of the model with the remaining four items found that the model
converged and fit the data well, Bollen-Stine p ¼ .493, c2 (2) ¼ 2.074, p ¼ .354, RMSEA ¼ .012 (.000, .127), CFI ¼ 1.00.
Construct reliability was .79 and the variance extracted was .64 (64%). Coefficient H was calculated to be .815 and the
critical ratio of the parameter estimates were all significant, indicating convergent validity. Based on the context under-
pinning the four items that comprise this factor, this factor was named “cognitive control”.

Validation sample
To test the models of each factor further, a random sample of 200 participants was chosen from those participants not
included in the calibration sample. The model fit for factor 1, Bollen-Stine p ¼ .522, c2 (20) ¼ 26.399, p ¼ .153, RMSEA ¼ .040
(.000, .077), CFI ¼ .985, and factor 3, Bollen-Stine p ¼ .353, c2 (2) ¼ 2.495, p ¼ .287, RMSEA ¼ .035 (.000, .150), CFI ¼ .998.
The validation model did not fit for factor 2, Bollen-Stine p ¼ .100, c2 (14) ¼ 30.742, p ¼ .006, RMSEA ¼ .078 (.040, .115),
CFI ¼ .959. The standardised residual covariances were explored, and it was shown that item 17 (Sometimes people make me
so mad that I can't control my behaviour) was not fitting well contextually with the other items. This item was dropped and
the model was tested again and found to fit well, Bollen-Stine p ¼ .683, c2 (9) ¼ 9.280, p ¼ .412, RMSEA ¼ .012 (.000, .081),
CFI ¼ .999.
To further test the robustness of this change to Factor 2, the original calibration sample was retested with item 17 removed.
The model converged and fit the data well, Bollen-Stine p ¼ .156, c2 (9) ¼ 16.331, p ¼ .060, RMSEA ¼ .057 (.000, .101),
CFI ¼ .975, and demonstrated good reliability and validity statistics (construct reliability ¼ .59, variance extracted ¼ .45 (45%),
Coefficient H ¼ .786). As such, a decision was made to drop item 17 from the second factor.

Second level confirmatory factor analysis

Theoretical debates as to the underlying structure of the personal responsibility construct indicate that the construct is
most likely to be multidimensional (Mergler, 2007; Swaner, 2005). To empirically test this possibility, a second level
confirmatory factor analysis model was conducted. In this model, personal responsibility was viewed as a first level construct,
with factors 1, 2 and 3 being dimensions of this latent construct. The initial sample of 513 participants that was used for the
EFA was used for this analysis, minus the 47 cases with missing data, leaving a total of 466 cases. Testing for multivariate
normality was conducted by calculating Mardia's coefficient and the sample showed the assumption of multivariate
normality to be violated (Mardia's coefficient ¼ 75.397; c.r. ¼ 36.036). Therefore the Bollen-Stine bootstrapping procedure
was applied (Bollen & Stine, 1992).
 A. Mergler, P. Shield / Journal of Adolescence 51 (2016) 50e57 55

In an ordered, sequential manner, the model was scrutinised to determine goodness of fit. The standardised residual
covariances and regression weights were explored to determine which items were not fitting well with other items, and the
single worst fitting item was removed at each step. At the end of this systematic process (step 4), the model converged and
was a good fit, Bollen-Stine p ¼ .643, c2 (87) ¼ 107.266, p ¼ .069, RMSEA ¼ .022 (.000, .035). CFI ¼ .988. Fig. 1 shows the final
multi-dimensional model for the Personal Responsibility construct.

Discussion

This critical, foundational step towards the development of a multi-dimensional measure of personal responsibility in
adolescents will support future research in this area. It paves the way for a consistent empirical use and understanding of this
construct in research across different contexts and with different populations. The identification of the underlying factors of
personal responsibility allows researchers to advance understanding of personal responsibility and conduct research that
explores its impact on the life choices and education of young people. In the current study, three factors were identified as
being critical determinants of personal responsibility in adolescents, including personal accountability, behavioural and
emotional control, and cognitive control. Understanding the dimensionality of personal responsibility generated through the
data informing these analyses is in line with the dimensionality of this construct that was proposed by Swaner (2005). As
such, this knowledge can deepen our understanding of the ways in which personal responsibility is constructed and the
factors that comprise it, as well as how it may interact with other non-cognitive skills that affect adolescent development.
In particular, the literature surrounding the self-control construct provides valuable insights into the control dimension of
personal responsibility. Previous research exploring personal responsibility, be it qualitative or quantitative studies, has
included an aspect of self-control as an important component of personal responsibility (Hellison, 2003; Martel et al., 1987).
In the current study self-control emerged as an important theme in the focus group data, and while not emerging as a distinct
factor in the quantitative analysis, elements of self-control are reflected in many items retained. Researchers can explore the
similarities, differences, and possible interplay between these constructs and design intervention studies that test and
measure the importance of self-control and personal responsibility for adolescent developmental outcomes.
While the five key themes from the focus group data did not hold as five distinct factors of personal responsibility, the final
15 items retained aspects from all five areas. Items from a number of themes merged into a single factor, while items reflective
of the cognitive awareness theme were separated into those that demonstrated “behavioural and emotional control” and
those that demonstrated “cognitive control”. The fact that behavioural and emotional control items were found to load onto a

Fig. 1. The second level confirmatory factor model for the personal responsibility construct.
56 A. Mergler, P. Shield / Journal of Adolescence 51 (2016) 50e57

single factor, while cognitive control items loaded onto a separate factor, offers valuable information to help clarify the
underlying multi-dimensionality of the personal responsibility construct. Earlier scholarly work exploring the personal re-
sponsibility construct with young adolescents found that adolescents may lack the cognitive sophistication required to
identify the differences between behavioural, emotional and cognitive components of personal responsibility (Lee et al., 2012;
Li et al., 2008; Watson et al., 2003). Future research needs to replicate this structure of the personal responsibility construct,
and explore potential differences in this structure based on developmental differences.

Limitations and future research

While the current study offers a short and psychometrically sound measure of personal responsibility, additional testing is
required before wide adoption of the measure occurs. In particular, the scale needs to be validated across a range of pop-
ulations to ensure that it is appropriate for use in such contexts. A limitation with the current research is that the scale has
been tested on a predominantly white, middle class adolescent population, and as such the results may not generalise to other
populations. In addition, it is imperative that the Personal Responsibility Scale is used alongside well established measures of
related constructs (such as self-control) in order to demonstrate construct validity (Strauss & Smith, 2009). To establish the
long-term validity of the scale, measurement invariance analyses need to be conducted across items and multiple groups.
Continuing the robust statistical development of the current Personal Responsibility Scale in future work will allow for a
consistent approach to measuring this construct across varying contexts.
Future research that applies the scale across different samples of participants to identify cultural, age and socioeconomic
status related correlates of personal responsibility would be valuable. For culture in particular, an argument could be made
that the way in which personal responsibility has been operationalised in the current paper highlights a western individu-
alistic focus, one that may not be borne out by more collectivist cultures (Lee et al., 2012). Intervention studies that seek to
alter levels of personal responsibility would provide necessary insight into how the construct can be manipulated, and studies
that test for the strength of one's commitment to being personally responsible would indicate the range of factors that may
undermine or enhance the likelihood of choosing personally responsible behaviour. As we begin to understand and measure
personal responsibility effectively, it is an exciting time to be engaged in this fundamentally important research area to
support the healthy social and emotional development of young people.

Conclusion

Personal responsibility in adolescents involves behavioural, emotional and cognitive control, alongside a willingness to
hold oneself accountable for one's choices. This Personal Responsibility Scale enables the multi-dimensional nature of per-
sonal responsibility to be measured. For researchers, the measure allows for a consistent approach to measuring personal
responsibility across different populations and in varying contexts. With consistent application of this measure, researchers
can perform essential validation studies, and extend our understanding of the personal responsibility construct. In doing so,
we get closer to developing targeted interventions to be used in schools to enhance this non-cognitive and increasingly
important skill in young people.

Acknowledgements

The authors would like to thank Professor Donna Berthelsen for her considered feedback on earlier versions of this
manuscript, and the former Executive Dean of the Faculty of Education at QUT, Professor Wendy Patton, for her support of the
Quantitative Research Fellowship program that allowed this work to be undertaken.

References

ACARA. (2015). Personal and social capability Accessed 07.07.15 http://www.australiancurriculum.edu.au/generalcapabilities/personal-and-social-capability/

introduction/introduction.
Anderson, J. C., & Gerbing, D. W. (1982). Some methods for respecifying measurement models to obtain unidimensional construct measurement. Journal of
Marketing Research, 19, 453e460.
Anderson, J. C., & Gerbing, D. W. (1988). Structural equation modeling in practice: A review and recommended two-step approach. Psychological Bulletin,
103(3), 411e423.
Bollen, K. A., & Stine, R. A. (1992). Bootstrapping goodness-of-fit measures in structural equation models. Sociological Methods & Research, 21, 205e229.
Coakes, S. J., & Steed, L. G. (2003). SPSS: Analysis without anguish: Version 11.0 for windows. Queensland: John Wiley & Sons.
Duckworth, A. L., & Yeager, D. S. (2015). Measurement matters: Assessing personal qualities other than cognitive ability for educational purposes.
Educational Researcher, 44(4), 237e251.
George, D., & Mallery, P. (2014). IBM SPSS statistics 21 step by step: A simple guide and reference (13th ed.). London, U.K: Pearson.
Glass, C. R. (2013). Strengthening and deepening education for personal and social responsibility. New Directions for Higher Education, 164, 83e94.
Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley.
Hellison, D. (1985). Goals and strategies for physical education. Champaign, IL: Human Kinetics.
Hellison, D. (2003). Teaching personal and social responsibility in physical education. In S. Silverman, & C. Ennis (Eds.), Students learning in physical edu-
cation: Applying research to enhance instruction (pp. 241e254). Champaign: Human Kinetics.
Horn, J. L. (1965). A rationale and test for the number of factors in factor analysis. Psychometrika, 30(2), 179e185.
Hunter, J. E., & Gerbing, D. W. (1982). Unidimensional measurement, second-order factor analysis, and causal models. In B. M. Staw, & L. L. Cummings (Eds.),
Vol. 4. Research in organizational behavior (pp. 267e299). Greenwich, CT: JAI Press.
 A. Mergler, P. Shield / Journal of Adolescence 51 (2016) 50e57 57

Krueger, R. A. (1998). The focus group kit. Thousand Oaks, CA: Sage.
Lee, O., Kim, Y., & Kim, B. J. (2012). Relations of perception of responsibility to intrinsic motivation and physical activity among Korean middle school
students. Perceptual & Motor Skills: Exercise & Sport, 115(3), 944e952.
Lehman, R. S. (1991). Statistics and research design in the behavioral sciences. Belmont, California: Wadsworth.
Lewis, R. (2004). Kids behaving badly, or responsibly? Helping teachers help students to act responsibly. Professional Educator, 3(4), 17e19.
Lickona, T. (1991). Educating for character: How our schools can teach respect and responsibility. New York: Bantam Books.
Linley, P. A., & Maltby, J. (2009). Personal responsibility. In S. J. Lopez (Ed.), Encyclopedia of positive psychology (pp. 685e689). Hoboken: Wiley.
Li, W., Wright, P. M., Rukavina, P. B., & Pickering, M. (2008). Measuring students' perceptions of personal and social responsibility and the relationship to
intrinsic motivation in urban physical education. Journal of Teaching in Physical Education, 27, 167e178.
Lovat, T., Clement, N., Dally, K., & Toomey, R. (2010). Values education as holistic development for all sectors: Researching for effective pedagogy. Oxford
Review of Education, 36(6), 713e729.
Mardia, K. V. (1970). Measures of multivariate skewness and kurtosis with applications. Biometrika, 57, 519e530.
Martel, J., McKelvie, S. J., & Standing, L. (1987). Validity of an intuitive personality scale: Personal responsibility as a predictor of academic achievement.
Educational and Psychological Measurement, 47, 1153e1163.
MCEETYA. (2008). The Melbourne declaration on the educational goals for young Australians Accessed 07.07.15 http://www.curriculum.edu.au/verve/_
resources/National_Declaration_on_the_Educational_Goals_for_Young_Australians.pdf.
Mergler, A. (2007). The creation, implementation and evaluation of school-based personal responsibility program (PhD dissertation). Queensland University of
Technology.
Mergler, A. G., & Patton, W. (2007). Adolescents talking about personal responsibility. Journal of Student Wellbeing, 1(1), 57e70.
O'Neill, N. (2012). Promising practices for personal and social responsibility: Findings from a national research collaborative. Washington, DC.: Association of
American Colleges and Universities Accessed 13.07.15 http://www.aacu.org/sites/default/files/files/core_commitments/promising_practices_rc2012.pdf.
Pallant, J. (2013). SPSS survival manual: A step by step guide to data analysis using IBM SPSS (4th ed.). Crows Nest, NSW: Allen & Unwin.
Piquero, A. R., Jennings, W. G., & Farrington, D. P. (2010). Self-control interventions for children under age 10 for improving self-control and delinquency and
problem behaviours. Oslo, Norway: The Campbell Collaboration.
Schlenker, B. R., Britt, T. W., Pennington, J., Murphy, R., & Doherty, K. (1994). The triangle model of responsibility. Psychological Review, 101, 632e652.
Singg, S., & Ader, J. A. (2001). Development of the student Personal Responsibility Scale-10. Social Behavior and Personality, 29(4), 331e336.
Smithkirai, C., Longthong, N., & Peijsel, C. (2015). Effect of using movies to enhance personal responsibility of university students. Asian Social Science, 11(5),
1e9.
Strauss, M. E., & Smith, G. T. (2009). Construct validity: Advances in theory and methodology. Annual Review of Clinical Psychology, 27(5), 1e25.
Swaner, L. E. (2005). Educating for personal and social responsibility: A review of the literature. Liberal Education, 91(3), 14e21.
Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Boston: Allyn and Bacon.
Trosset, C. (2013). Broadening our understanding and assessment of personal and social responsibility: A challenge to researchers and practitioners. New
Directions for Higher Education, 164, 23e30.
Watson, D. L., Newton, M., & Kim, M. (2003). Recognition of values-based constructs in a summer physical activity program. The Urban Review, 35(3),
217e231.
West, S. G., Finch, J. F., & Curran, P. J. (1995). Structural equation models with non-normal variables: Problems and remedies. In R. Hoyle (Ed.), Structural
equation modeling: Concepts, issues, and applications (pp. 56e75). Thousand Oaks, CA: Sage.
West, M. R., Kraft, M. A., Finn, A. S., Martin, R., Duckworth, A. L., Gabrieli, C. F. O., et al. (2014). Promise and paradox: Measuring students' non-cognitive skills
and the impact of schooling. In CESinfo Area Conference on Economics and Education, 12e13 September, Munich, Germany.