INTEREST, SELF-EFFICACY AND PRIOR MATHEMATICS ACHIEVEMENT – CARMICHAEL ET AL 83

Australian Journal of Educational & Developmental Psychology. Vol 10, 2010, pp. 83-93

Statistical literacy in the middle school: The relationship between interest,

self-efficacy and prior mathematics achievement1

Colin Carmichael
Rosemary Callingham
Ian Hay
Jane Watson2
University of Tasmania

ABSTRACT

This paper examines the relationship between key predictors of middle school
students’ interest for statistical literacy, including their self-efficacy for statistical
literacy and their prior mathematics achievement. Given the lack of research in the
statistical literacy domain, a review of the mathematics education literature confirms
the importance of self-efficacy and achievement in predicting interest and describes
the nature of their influence on interest. The paper then reports on an empirical study
involving a sample of 438 Australian middle school students from 8 schools across 3
states. Data from this study are used in a series of statistical models to confirm the
influence of identified predictors and to study their inter-relationship. Results from
the study suggest that the influence of prior achievement on interest is completely
mediated by self-efficacy. In addition to this, the relationship between interest and
self-efficacy appears to be quadratic, in that students with high levels of self-efficacy
may not have corresponding high levels of interest. Implications for this research are
then discussed.

INTRODUCTION

It is imperative that citizens of the 21st century have the necessary skills to interact in a
meaningful way with the increasing volumes of information that they encounter; information
that in many instances contains statistical elements such as graphs and tables. Yet in this time
of rapid change, researchers report that Australia is unable to educate a sufficient number of
skilled graduates in mathematics (Australian Academy of Science, 2006) and statistics
(Trewin, 2005). Such skill shortages are influenced by the reported declines in the numbers of
students in senior secondary school who opt to take the higher level mathematics courses that
lead into tertiary study (Forgasz, 2006). Students’ choice of mathematics course is influenced
by a number of factors that include: their interest and liking for mathematics, their previous
achievement in mathematics, their mathematics self-concept, and their perceptions regarding

1
Acknowledgements
The research was funded by Australian Research Council grant number LP0669106, with support from
Linkage partner, the Australian Bureau of Statistics.
2
Contact
Professor Jane Watson
Faculty of Education
University of Tasmania
Email [email protected]

ISSN 1446-5442 Website: www.newcastle.edu.au/journal/ajedp/

INTEREST, SELF-EFFICACY AND PRIOR MATHEMATICS ACHIEVEMENT – CARMICHAEL ET AL 84

the usefulness and difficulty of mathematics (McPhan, Morony, Pegg, Cooksey, & Lynch,
2008). Of these factors, evidence suggests that interest is dominant (Watt, 2005; Wigfield,
Tonks, & Eccles, 2004): Students are more likely to re-engage in any activity if they are
interested in it. This study seeks to address the reported skill shortages in statistics through an
investigation of middle school students’ interest for statistical literacy. Such literacy is
defined as an ability to interact critically with messages containing statistical elements (Gal,
2003) and the underlying skills of this literacy are developed in the chance and data strand of
all Australian P-10 mathematics syllabi.
Although the underlying concepts of statistical literacy are embedded in the mathematics
curriculum, it is assumed that students encounter statistical messages in a wide variety of
contexts both in and out of school. Consequently statistical literacy is more than a topic of the
mathematics curriculum. In fact statistics itself is regarded as a methodological discipline that
is distinct from mathematics (Moore, 1988). The importance of statistical literacy is reflected
in the proposed Australian National Mathematics Curriculum (National Curriculum Board,
2009), which identifies chance and data as one of three content strands. It is no surprise,
therefore, that researchers have explored the development of statistical literacy in both school
students (Watson, 2006) and adults (Gal, 2003; 2004). Indeed Callingham and Watson
(2005) have used students’ performance in statistical literacy related tasks to map out a
detailed hierarchy for the domain. Although these studies note the importance of affective
elements, none have explored their influence in this domain. This study seeks to fill this gap
in the literature through an investigation of middle school students’ interest.
A review of the literature suggests that middle school students’ interest for statistical
literacy should be influenced by a number of factors that can broadly be classified as those
relating to the individual and those relating to the actual learning situation (Carmichael,
Callingham, Watson, & Hay, 2009). In this study, the focus is primarily on the former,
although the influence of the teacher is also explored. Given this focus, the research question
is: How do variables such as age, prior achievement, gender, and self-competency beliefs,
contribute to middle school students’ interest for statistical literacy?

Theoretical review
The discussion in this section commences with a theoretical review of the interest
construct. It then identifies key predictors of interest that are unique to the individual and in
particular establishes the importance of students’ ages, self-competency beliefs and prior
mathematics achievement on their assessments of interest. The discussion concludes with an
examination of the influence of the teacher on students’ interest.
Interest is an affect with both trait and state properties. At the trait level individual
interest is a valuing of an interest object, so much so that it becomes identified with the self
(Dewey, 1910). Having such interest is considered to be essential for psychological health
(Hunter & Csikszentmihalyi, 2003). At the state level, situational interest is identified as one
of several basic human emotions (Izard, 1984) and is considered to be similar to the
motivational state of curiosity (Silvia, 2001). Theories of interest development suggest that
situational interest can emerge into individual interest (Hidi & Renninger, 2006; Krapp,
2007).
In an educational setting, the interest assessed through pen and paper instruments is
regarded as a form of individual interest, where it is defined as a student’s “relatively
enduring predisposition to reengage particular content over time” (Hidi & Renninger, 2006, p.
113). Students with such an interest for a particular subject are known to display deeper levels
of cognitive processing, greater use of self-regulatory learning strategies and provide more
positive ratings on the quality of the learning experience, than their disinterested peers
(Schiefele, 1991).
The adolescent students that inhabit middle schools find themselves undergoing
significant physical and psychological changes. In regard to the latter, researchers argue the
primary role of adolescence is that of identity formation (Hay & Ashman, 2003). Such
students typically become more attuned to the interests of their peers and use these to help

ISSN 1446-5442 Website: www.newcastle.edu.au/journal/ajedp/

INTEREST, SELF-EFFICACY AND PRIOR MATHEMATICS ACHIEVEMENT – CARMICHAEL ET AL 85

define the self (Krapp, 2007). It is not surprising, therefore, that researchers report declines in
the levels of interest that students have for learning in general (Dotterer, McHale, & Crouter,
2009) and for specific domains such as mathematics (Watt, 2008). It is therefore expected that
age will have a negative influence on middle school students’ interest for statistical literacy.
Krapp (2007) argued that the development of interest is directed by two psychological
control systems, the cognitive and the emotional. In regards to the former, he argued that
students consciously direct their interest towards certain objects in order to achieve goals. In a
school setting many students have achievement related goals and their interest for attaining
these goals will be influenced by their self-competency beliefs. This association between
interest and self-competency beliefs, in the form of mathematics self-concept, is confirmed in
the secondary mathematics context (Trautwein, Ludtke, Köller, Marsh, & Baumert, 2006).
Self-concept, however, is a reflective construct (Bong & Skaalvik, 2003) that is primarily
based on students’ perceptions of their past achievement experiences. Self-efficacy, which is
defined as “beliefs in one’s capabilities to organize and execute the courses of action required
to produce given attainments” (Bandura, 1997, p.3), is a forward-looking construct. Apart
from past mastery experiences, self-efficacy is also influenced by the vicarious experiences of
others, the persuasion of significant others and even immediate physiological factors
(Bandura, 1997). Of all the psychosocial factors, self-efficacy is reported to be the best
predictor of achievement in an educational context (Robbins et al., 2004).
The expected relationship between self-efficacy and interest may be more complex than
the reported linear association between self-concept and interest (Trautwein et al., 2006). It is
thought that one of the mechanisms for evoking the interest state is a person interaction with a
collative variable such as novelty or uncertainty (Berlyne, 1960). In relation to this, Silvia
(2003) argued that there is a degree of uncertainty associated with students’ self-efficacy
assessments and that this uncertainty will manifest itself in a quadratic way on their
associated interest. More specifically, students with low levels of self-efficacy for a given task
are certain that they cannot master it and as a result will lose interest. On the other hand,
students with high levels of self-efficacy are certain that they can master the task and as a
result will also lose interest. Students with mid-levels of self-efficacy, those with sufficient
uncertainty regarding their mastery of the task, are expected to report the most interest.
In regards to prior achievement, the “Model of Domain Learning (MDL)” (Alexander,
2003) predicts that the acquisition of knowledge in a domain will be accompanied by
increased levels of individual interest. This association between prior achievement and
interest has been confirmed in the mathematics domain (Köller, Baumert, & Schnabel, 2001).
More recently, however, Trautwein et al. (2006) reported that students’ achievement relative
to their peers is a stronger predictor of interest than their absolute achievement. They also
reported that the influence of students’ self-concept almost entirely mediates the relationship
between their achievement and interest.
Research suggests that gender will also influence interest. For example Köller et al.
(2001) reported that boys have higher levels of interest for mathematics than girls. In addition
to this, Schiefele, Krapp, and Winteler (1992) reported that the association between
achievement and interest is stronger for boys than for girls. It is not clear, however, whether
such reported gender differences will occur in this context, especially given reported gender
differences in students’ interest for some of the topics and activities associated with the
acquisition of statistical literacy (Carmichael & Hay, 2009b).
As mentioned, features of the learning situation are thought to contribute to students’
situational interest and ultimately their individual interest. The class teacher is the primary
architect of what happens in the learning environment and therefore should exert some
influence on students’ interest. For example, teachers’ provision of structure during a lesson is
known to have a positive impact on students’ interest (Kunter, Baumert, & Köller, 2007).
Given the identified predictors of middle school students’ interest for statistical literacy,
the discussion in the next section describes methods for obtaining relevant measures and the
subsequent modelling of these measures in order to confirm their influence on interest and to
explore their inter-relationship.

ISSN 1446-5442 Website: www.newcastle.edu.au/journal/ajedp/

INTEREST, SELF-EFFICACY AND PRIOR MATHEMATICS ACHIEVEMENT – CARMICHAEL ET AL 86

METHOD
The data collection in this study commenced with the selection of a representative sample
of Australian middle school students. From these students a number of data were then
collected that included measures of their interest, self-efficacy and prior achievement. These
data were then analysed using a series of regression equations. The discussion in this section
describes each of these stages.

Sample selection
The sample of 438 students used in this study was a sub-sample of 791 students
participating in an interest study for whom prior mathematics achievement data were
available. The larger sample represents those from a total of 1384 invited students who agreed
to participate in the interest study, a response rate of 57%. It was a convenience sample3 of
the Australian middle school population, which can be assumed to consist of approximately
equal proportions of each gender, equal proportions of students in each of years 7, 8, and 9,
and a range of school types including both Government and Independent schools.
The 438 students reported in this study attended six independent and two government
schools from three Australian states. Of the students in this sample, 54% were from
Tasmania, 29% from Victoria and the remainder from Queensland. Most students (69%)
attended independent schools and the majority (85%) were in Years 7, 8 or 9. The mean age
of students in this study was 13.6 years and just over one half (53%) were male.

Instruments used and data collected

All students were asked to complete the Statistical Literacy Interest Measure (SLIM), a
previously validated interest instrument (Carmichael, 2008) that contains 16 self-descriptions,
which students respond to using a five-point Likert scale. An interval measure of interest was
constructed using the Rasch Rating Scale Model (Andrich, 1978) with the software package
Winsteps (Linacre, 2006) and all students were assigned an interest score. For this sample of
students, SLIM explained 67% of the variance in their responses and reported an estimated
internal reliability coefficient of α = 0.91 .
Students also completed the Self-efficacy for Statistical Literacy (SESL) scale, also
previously validated (Carmichael & Hay, 2009a), that contains nine self-descriptions with the
same five-point Likert scale. As with the interest measure, a Rasch measurement paradigm
was adopted for the construction of the self-efficacy measure, which was used to assign each
student a self-efficacy score. For this sample of students, SESL explained 70% of the variance
in their responses and reported an estimated internal reliability coefficient of α = 0.93 .
Teachers of students provided estimated ratings of their students’ prior achievement in
mathematics, with such ratings ranging from E, the lowest grade, to A. These grades reflect
the reporting measures used in Australia, which were mandated by the Commonwealth
Government (Department of Education, Science and Training, 2005). Given that grades of E
and D represent less than satisfactory achievement, these categories were collapsed into one
denoted F. Consequently, of the 438 students in the study, 36 were assigned a grade of F, 106
a grade of C, 182 a grade of B and 114 a grade of A. Given the hypothesised importance of
students’ prior achievement relative to their class, an additional four category prior
achievement variable was constructed in which students’ achievement grades were compared
to the median grade of their class. Of the 438 students in the study, 27 had an achievement
two grades below the class median, 89 had an achievement one grade below the median, 219
achieved the median grade, and the remainder had an achievement one grade above the
median.
In addition to prior achievement, interest and self-efficacy data, demographic data
including students’ ages, gender and school attended were collected. Information regarding
the class and teacher each student was assigned was only available for 411 students.

3
Since the sample was non-random, cited p-values are notional, as are tests of statistical significance.

ISSN 1446-5442 Website: www.newcastle.edu.au/journal/ajedp/

INTEREST, SELF-EFFICACY AND PRIOR MATHEMATICS ACHIEVEMENT – CARMICHAEL ET AL 87

Analysis
Bivariate relationships between interest and its hypothesised predictors were initially used
to identify their statistical significance. Latent regression models (Adams, Wilson, & Wu,
1997) were then used to analyse the inter-relationship between interest, self-efficacy, prior
mathematics achievement and other individual factors. Such models allow for the inherent
measurement error in the response variable, in this case the interest and/or self-efficacy
measures, and were implemented using the software package Conquest (Wu, Adams, Wilson,
& Haldane, 1998). They do not allow for the inherent measurement error in predictor
variables, however in such instances coefficient estimates can be attenuated for error (Aiken
& West, 1991).
As no teacher specific variables were available in this study, the influence of the teacher
was examined through the use of mixed effects models (Faraway, 2006), which cater for any
dependency between subjects in given groups. In this study students were grouped by class
and it assumed that any dependency is attributed, in part, to the influence of their teacher.
These models were applied using the software package R (R Development Core Team, 2009).

RESULTS
The discussion in this section commences with a report of significant bivariate
relationships between interest and other measured variables. It then reports the application of
latent regression models to students’ interest responses and also their self-efficacy responses.
The discussion concludes with a report of the application of mixed effects models to students’
interest scores.
Students’ interest scores were moderately associated with their self-efficacy
scores (r = .63, p = .00) . The strength of this association is similar in magnitude to the
average value of .59 reported by Rottinghaus, Larson, and Borgen (2003) in their meta-
analysis of 60 self-efficacy/interest studies. Students’ relative prior achievement in
mathematics influenced their interest scores ( F = 3.85, p = .01) , although the greatest
differences in group means occurred between students with an achievement two grades below
median and the other groups of students. Interestingly, students’ absolute prior mathematics
achievement had a greater influence on their interest ( F = 11.47, p = .00) than their relative
prior achievement. There was no significant linear association between the ages of students
and their interest scores.
A series of latent regression models were then used to analyse the inter-relationship
between these significant predictors of interest. Of the available predictors of interest, only
students’ self-efficacy scores and ages were statistically significant. In the presence of self-
efficacy the reported influence of prior achievement was not significant. In order to assess the
proposed quadratic association between self-efficacy and interest, the square of the former
was also included in the model. This term was found to contribute significantly and reduced
model deviance by 33.44. This reduction is approximately chi-square distributed on one
degree of freedom and hence represents a statistically significant, albeit small, improvement
in model fit. Coefficients and their standard errors for the resulting model are reported in
Table 1 as Model 1. This table also includes the estimated R-square and reported deviance for
the model.
Given that prior mathematics achievement was not a predictor of interest; a model was
created with self-efficacy as the response variable. In this instance, prior mathematics
achievement contributed to explaining self-efficacy, as did age and interest. Gender, however,
was not a significant factor. The details of this model are reported in Table 1 as Model 2. The
influence of prior achievement was modelled using contrasts so that compared to students
with an achievement two grades below the class median an achievement of one grade below
the median will contribute an additional 0.160 logits of self-efficacy, a median grade
achievement will contribute an additional 0.454 logits, and an above median grade
achievement will contribute an additional 0.614 logits.

ISSN 1446-5442 Website: www.newcastle.edu.au/journal/ajedp/

INTEREST, SELF-EFFICACY AND PRIOR MATHEMATICS ACHIEVEMENT – CARMICHAEL ET AL 88

Table 1: Details for Models 1 and 2

Variable B se( B)
Model 1
Constant 0.762 0.509
Age (years) -0.065 0.028
Self-efficacy (logits) 0.424 0.020
Square of self-efficacy -0.045 0.006
R2 = .50, Deviance = 19935

Model 2
Constant -2.524 0.860
Age 0.169 0.034
Prior achievement contrast 1 0.160 0.090
Prior achievement contrast 2 0.454 0.100
R2 = .51, Deviance = 11326

A series of mixed effects models were applied to students’ interest scores using the
significant predictors identified in Model 1 and with students’ class membership as the
grouping variable. In order to model any dependence between students in the one class, a
random effect in the constant term was assessed for significance. In this instance, however,
the influence of age in the model became insignificant, presumably because students within
the same class have similar ages. Re-applying the mixed effects model without age as a fixed
effect and with a random effect in the constant term generated a model with two significant
predictors, namely self-efficacy and its square. The specified model is:
Interest = (−0.16 + b) + 0.44SE − 0.05SE 2 + ε , where b represents the random effect in
the constant term, SE the student’s self-efficacy score, and ε the residual error. The variance
associated with the random effect b is reported as σ b = 0.03 , while the residual variance is
2

σ ε 2 = 0.63 . Further details of the model are reported in Table 2.

Table 2: Details for Model 3

Variable B se( B)
Constant -0.16 0.06
Self-efficacy (logits) 0.44 0.03
Square of self-efficacy -0.05 0.01
Deviance = 990

Compared to a simple linear regression model with age, self-efficacy and its square as
predictors, this model reduced deviance by 4.3. This reduction in deviance is chi-square
distributed on one degree of freedom and is statistically significant at the 5% level.
Consequently the model improves fit on a standard linear model, but only slightly. It does not,
however, provide useful information regarding the influence of the teacher. This is because
the random effect in the constant term also models the influence of students’ ages. In general,
classes with lower average levels of interest were those with older students. In addition to
this, the variance associated with the class grouping (σ b = 0.03) , is much smaller than the
2

ISSN 1446-5442 Website: www.newcastle.edu.au/journal/ajedp/

INTEREST, SELF-EFFICACY AND PRIOR MATHEMATICS ACHIEVEMENT – CARMICHAEL ET AL 89

variance associated with the residual error (σ ε = 0.63) , suggesting that compared to the
2

noise in the model, the influence of class grouping is relatively minor.

DISCUSSION
The analysis confirms that students’ self-efficacy for statistical literacy is a key predictor
of their interest and that the relationship is quadratic. The analysis also confirms the findings
of Trautwein et al. (2006) that self-competency beliefs, in this case self-efficacy, completely
mediate the influence of prior achievement on interest. The quadratic relationship supports
Silvia’s (2003) contention that more confident students lose interest in tasks that they are
certain they can successfully complete. As expected, age did have a negative influence on
interest, but only when controlling for self-efficacy.
The lack of a gender effect in predicting both interest and self-efficacy is noteworthy,
especially in light of reports that boys show higher levels of interest (Köller et al., 2001) and
self-efficacy (Hyde & Durik, 2005) for mathematics than girls. This may point to distinctive
differences between the mathematics and statistical literacy domains, in that the latter also
involves a facility with language where girls are known to display higher levels of self-
efficacy than boys (Hyde & Durik, 2005). In addition to this, there are contexts associated
with statistical literacy that girls tend to find more interesting than boys (Carmichael & Hay,
2009b).
Although it is expected that the teacher will have an influence on students’ interest, the
results of the mixed effects models were inconclusive. In the first instance a random class
effect, so presumably a teacher related effect, was significant in the constant term. This only
occurred, however, when age was removed as a fixed effect from the model and included as
random effect in the constant term. In the second instance, the variation associated with the
class grouping term was relatively small when compared with the residual error suggesting
that its influence on interest is relatively minor.
Given the strong influence of interest on self-efficacy, a reciprocal effects model might be
appropriate in that students with high levels of interest for statistical literacy will be likely to
engage in the domain, gain competence and thus self-efficacy. Increased levels of self-
efficacy, in turn, are then likely to promote further engagement and thus interest. In a study
involving the self-concept construct, Marsh, Trautwein, Ludtke, Köller, and Baumert (2005)
found some support for a reciprocal effects model, although they reported stronger support for
self-concept predicting later interest than the reverse.

Limitations of the study

There were limitations associated with the methodology. In the first instance it was not
possible to select students randomly. Given that participation was voluntary, students in this
sample may have displayed different interest characteristics than those who chose not to
participate.
A second limitation of the study concerns the use of latent variables and the inherent
measurement error associated with such variables. This was overcome in Models 1 and 2
through the use of latent regression models, although these did include predictor variables that
contained measurement error. The aim of this study, however, was to investigate the inter-
relationship between interest, achievement and self-efficacy and not specifically the size of
effects. The two latent regression models, therefore, served this purpose. The mixed effects
models did not account for measurement error in any variable and this could have contributed
to the inclusive results obtained from these models.

Implications and future research

The research reported in this paper suggests that students’ competency valuations are a
strong predictor of interest. This result is not unexpected with cited papers reporting such a
relationship. Indeed studies involving the concept of flow (Csikszentmihalyi, 2002), which is
an extreme form of interest where total involvement with an activity occurs, report that

ISSN 1446-5442 Website: www.newcastle.edu.au/journal/ajedp/

INTEREST, SELF-EFFICACY AND PRIOR MATHEMATICS ACHIEVEMENT – CARMICHAEL ET AL 90

positive competency valuations are a necessary precondition for flow. Further, in order for
flow to occur, students must perceive that a balance exists between the perceived challenge of
the task and their perceived competence. Students who do not perceive sufficient challenge in
the task or who see the task as too challenging are unlikely to experience flow. The quadratic
relationship reported in this study supports this notion of an optimum level of challenge in
order for interest or flow to occur. More recently, Csikszentmihalyi, Abuhamdeh, and
Nakamura (2005) argued that some students may need external support during the initial
stages of an activity in order to gain sufficient competence valuations. That given this support
and sufficient challenge in the activity an emergent motivation may develop, making it more
likely that the students will experience flow. Given the key role that competence valuations
appear to have on interest and the importance of it as an outcome variable, further research
needs to occur at the task level with a view to ascertaining how support can increase middle
school students’ levels of self-efficacy during their interaction with specific tasks.
The MDL predicts a positive association between students’ knowledge and interest. In
this study no measures of statistical literacy knowledge were available, yet students’ prior
mathematics achievement did not have a direct influence on their interest for statistical
literacy. This lack of a direct relationship could be due to differences in the domain, although,
as discussed, studies in mathematics have reported similar findings. Much of the reported
empirical testing of the MDL has occurred in an undergraduate context where students are
older and likely to have more stable emotions than middle school students (Larson, Moneta,
Richards, & Wilson, 2002). Further research in the statistical literacy domain with a larger
randomised sample and with statistical literacy knowledge measures could provide valuable
information regarding the efficacy of the MDL in the adolescent context.
As discussed, the lack of reported gender effects on both interest and self-efficacy could
be associated with gender differences in students’ achievement goal orientations, in that girls
are likely to show higher levels of mastery and performance goals than boys in stereotypically
female domains such as language (Hyde & Durik, 2005). More recent research has
established a link between students’ use of mastery achievement goals and increased levels of
their interest (Hulleman, Durik, Schweigert, & Harackiewicz, 2008). Further research in the
statistical literacy domain should consider the goal orientation of students and the impact that
gender has on this and subsequent interest.
The regression models reported in this paper explained at most 51% of the variance in
measured variables. Consequently there is quite a large proportion of unexplained variance.
Apart from demographic variables, only measures of students’ self-efficacy and prior
mathematics achievement were available. Given the earlier discussion regarding the influence
of the teacher and the suggestion that external support needs to be provided for some students
in order to experience the interest state, further research in this area would benefit from the
inclusion of teacher level variables, especially that of teacher support.

REFERENCES
Adams, R. J., Wilson, M., & Wu, M. J. (1997). Multilevel item response models: An
approach to errors in variables regression. Journal of Educational and Behavioral
Statistics, 22(1), 47-76.
Aiken, K., & West, S. (1991). Multiple regression: Testing and interpreting interaction.
Newbury Park: SAGE Publications.
Alexander, P. A. (2003). The development of expertise: The journey from acclimation to
proficiency. Educational Researcher, 32(8), 10-14.
Andrich, D. (1978). A rating formulation for ordered response categories. Psychometrika,
43(4), 561-573.
Australian Academy of Science. (2006). Mathematics and statistics: Critical skills for
Australia's future. Retrieved from: http://www.review.ms.unimelb.edu.au/Report.html.
Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman.
Berlyne, D. E. (1960). Conflict, arousal, and curiosity. New York: McGraw-Hill.

ISSN 1446-5442 Website: www.newcastle.edu.au/journal/ajedp/

INTEREST, SELF-EFFICACY AND PRIOR MATHEMATICS ACHIEVEMENT – CARMICHAEL ET AL 91

Bong, M., & Skaalvik, E. M. (2003). Academic self-concept and self-efficacy: How different
are they really? Educational Psychology Review, 15(1), 1-40.
Callingham, R., & Watson, J. M. (2005). Measuring statistical literacy. Journal of Applied
Measurement, 6(1), 1-29.
Carmichael, C. S. (2008). The development and validation of the Statistical Literacy Interest
Measure (SLIM). In T. Hays & R. Hussain (Eds.), Proceedings of the Third Annual
Postgraduate Research Conference: Bridging the gap between ideas and doing research
(pp. 35-39). Armidale: University of New England.
Carmichael, C. S., Callingham, R., Watson, J. M., & Hay, I. (2009). Factors influencing the
development of middle school students' interest in statistical literacy. Statistics Education
Research Journal, 8(1), 62-81.
Carmichael, C. S., & Hay, I. (2009a). The development and validation of the Students' Self-
Efficacy for Statistical Literacy Scale. In R. Hunter, B. Bicknell & T. Burgess (Eds.),
Proceedings of the 32nd Annual Conference of the Mathematics Education Research
Group of Australasia (Vol. 1, pp. 97-104). Wellington: MERGA Inc.
Carmichael, C. S., & Hay, I. (2009b). Gender differences in middle school students' interests
in a statistical literacy context. In R. Hunter, B. Bicknell & T. Burgess (Eds.),
Proceedings of the 32nd Annual Conference of the Mathematics Education Research
Group of Australasia (Vol. 1, pp. 89-96). Wellington: MERGA Inc.
Csikszentmihalyi, M. (2002). Flow: The classic work on how to achieve happiness. London:
Rider.
Csikszentmihalyi, M., Abuhamdeh, S., & Nakamura, J. (2005). Flow. In A. J. Elliot & C. S.
Dweck (Eds.), Handbook of competence and motivation (pp. 598-608). New York:
Gilford Press.
Department of Education, Science and Training (2005). Schools assistance regulations.
Retrieved from Department of Education, Science and Training website:
http:/www.dest.gov.au
Dewey, J. (1910). Interest in relation to training of the will. In J. J. Findlay (Ed.), Educational
essays (pp. 73-132). Bath: Cedric Chivers Ltd.
Dotterer, A. M., McHale, S. M., & Crouter, A. C. (2009). The development and correlates of
academic interests from childhood through adolescence. Journal of Educational
Psychology, 101(2), 509-519.
Faraway, J. (2006). Extending the linear model with R: Generalized linear, mixed effects and
non-parametric regression models. New York: Chapman and Hall/CRC.
Forgasz, H. (2006). Australian year 12 "Intermediate" level mathematics enrolments 2000-
2004: Trends and patterns. In P. Grootenboer, R. Zevenbergen & M. Chinnappan (Eds.),
Proceedings of the 29th Annual Conference of the Mathematics Education Research
Group of Australasia (Vol. 1, pp. 211-220). Canberra: MERGA.
Gal, I. (2003). Teaching for statistical literacy and services of statistics agencies. The
American Statistician, 57(2), 80-84.
Gal, I. (2004). Statistical literacy. In D. Ben-Zvi & J. B. Garfield (Eds.), The challenge of
developing statistical literacy, reasoning and thinking (pp. 47-78). Dordrecht: Kluwer
Academic Publishers.
Hay, I., & Ashman, A. F. (2003). The development of adolescents' emotional stability and
general self-concept: The interplay of parents, peers, and gender. International Journal of
Disability, Development and Education, 50(1), 77-91.
Hidi, S., & Renninger, K. A. (2006). The four-phase model of interest development.
Educational Psychologist, 41(2), 111-127.
Hulleman, C. S., Durik, A. M., Schweigert, S. A., & Harackiewicz, J. M. (2008). Task values,
achievement goals, and interest: An integrative analysis. Journal of Educational
Psychology, 100(2), 398-416.
Hunter, J. P., & Csikszentmihalyi, M. (2003). The positive psychology of interested
adolescents. Journal of Youth and Adolescence, 32(1), 27-35.

ISSN 1446-5442 Website: www.newcastle.edu.au/journal/ajedp/

INTEREST, SELF-EFFICACY AND PRIOR MATHEMATICS ACHIEVEMENT – CARMICHAEL ET AL 92

Hyde, J. S., & Durik, A. M. (2005). Gender, competence, and motivation. In A. J. Elliot & C.
S. Dweck (Eds.), Handbook of competence and motivation (pp. 375-391). New York:
Gilford Press.
Izard, C. E. (1984). Emotion-cognition relationships and human development. In C. E. Izard,
J. Kagan & R. B. Zajonc (Eds.), Emotions, cognition, and behavior (pp. 17-37). New
York: Cambridge University Press.
Köller, O., Baumert, J., & Schnabel, K. (2001). Does interest matter? The relationship
between academic interest and achievement in mathematics. Journal for Research in
Mathematics Education, 32(5), 448-470.
Krapp, A. (2007). An educational-psychological conceptualisation of interest. International
Journal for Educational and Vocational Guidance, 7(1), 5-21.
Kunter, M., Baumert, J., & Köller, O. (2007). Effective classroom management and the
development of subject-related interest. Learning and Instruction, 17(5), 494-509.
Larson, R. W., Moneta, G., Richards, M. H., & Wilson, S. (2002). Continuity, stability, and
change in daily emotional experience across adolescence. Child Development, 73(4),
1151-1165.
Linacre, J. M. (2006). WINSTEPS Rasch measurement computer program (Version 3.61.2)
[Computer Software]. Chicago: Winsteps.com.
Marsh, H. W., Trautwein, U., Ludtke, O., Köller, O., & Baumert, J. (2005). Academic self-
concept, interest, grades, and standardized test scores: Reciprocal effects models of causal
ordering. Child Development, 76(2), 397-416.
McPhan, G., Morony, W., Pegg, J., Cooksey, R., & Lynch, T. (2008). Maths? Why not?
Retrieved from Australian Department of Education, Employment and Workplace
Relations website: http://www.dest.gov.au.
Moore, D. S. (1988). Should mathematicians teach statistics? The College Mathematics
Journal, 19(1), 3-7.National Curriculum Board. (2009). Shape of the Australian
curriculum: Mathematics. Retrieved from Australian Curriculum Assessment and
Reporting Authority website: http://www.acara.edu.au
R Development Core Team. (2009). R 2.10.1. Vienna, Austria: R Foundation for Statistical
Computing.
Robbins, S. B., Lauver, K., Le, H., Davis, D., Langley, R., & Carlstrom, A. (2004). Do
psychosocial and study skill factors predict college outcomes? A meta-analysis.
Psychological Bulletin, 130(2`), 261-288.
Rottinghaus, P. J., Larson, L. M., & Borgen, F. H. (2003). The relation of self-efficacy and
interests: A meta-analysis of 60 samples. Journal of Vocational Behavior, 62(4), 221-236.
Schiefele, U. (1991). Interest, learning, and motivation. Educational Psychologist, 26(3 & 4),
299-323.
Schiefele, U., Krapp, A., & Winteler, A. (1992). Interest as a predictor of academic
achievement: A meta-analysis of research. In K. A. Renninger, S. Hidi & A. Krapp
(Eds.), The role of interest in learning and development (pp. 183-212). New Jersey:
Lawrence Erlbaum Associates.
Silvia, P. J. (2001). Interest and interests: The psychology of constructive capriciousness.
Review of General Psychology, 5(3), 270-290.
Silvia, P. J. (2003). Self-efficacy and interest: Experimental studies of optimal incompetence.
Journal of Vocational Behavior, 62(4), 237-249.
Trautwein, U., Ludtke, O., Köller, O., Marsh, H. W., & Baumert, J. (2006). Tracking,
grading, and student motivation: Using group composition and status to predict self-
concept and interest in ninth-grade mathematics. Journal of Educational Psychology,
98(4), 788-806.
Trewin, D. (2005). Improving statistical literacy: The respective roles of schools and the
National Statistical Offices. In M. Coupland, J. Anderson & T. Spencer (Eds.), Twentieth
Biennial Conference of the Australian Association of Mathematics Teachers (pp. 11-19).
Adelaide: AAMT.
Watson, J. M. (2006). Statistical literacy at school: Growth and goals. Mahwah, New Jersey:
Lawrence Erlbaum Associates.

ISSN 1446-5442 Website: www.newcastle.edu.au/journal/ajedp/

INTEREST, SELF-EFFICACY AND PRIOR MATHEMATICS ACHIEVEMENT – CARMICHAEL ET AL 93

Watt, H. M. G. (2005). Explaining gendered math enrolments for NSW Australian secondary
school students. New Directions for Child and Adolescent Development, 110, 15-29.
Watt, H. M. G. (2008). A latent growth curve modeling approach using an accelerated
longitudinal design: The ontogeny of boys' and girls' talent perceptions and intrinsic
values through adolescence. Educational Research and Evaluation, 14(4), 287-304.
Wigfield, A., Tonks, S., & Eccles, J. S. (2004). Expectancy value theory in cross-cultural
perspective. In D. M. McInerney & S. Van Etten (Eds.), Big theories revisited (pp. 165-
198). Greenwich (Conn): Information Age Publishing Inc.
Wu, M. J., Adams, R. J., Wilson, M. R., & Haldane, S. (1998). ACER Conquest. Generalised
item response modelling software manual. Melbourne: ACER Press.

ISSN 1446-5442 Website: www.newcastle.edu.au/journal/ajedp/