You are on page 1of 21

School Psychology Review

ISSN: (Print) 2372-966x (Online) Journal homepage: https://www.tandfonline.com/loi/uspr20

Motivation and Self-Regulated Learning Influences


on Middle School Mathematics Achievement

Timothy J. Cleary & Anastasia Kitsantas |

To cite this article: Timothy J. Cleary & Anastasia Kitsantas | (2017) Motivation and Self-
Regulated Learning Influences on Middle School Mathematics Achievement, School Psychology
Review, 46:1, 88-107

To link to this article: https://doi.org/10.1080/02796015.2017.12087607

Published online: 14 Jan 2020.

Submit your article to this journal

View related articles

View Crossmark data

Citing articles: 3 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=uspr20
School Psychology Review,
2017, Volume 46, No. 1, pp. 88 –107

Motivation and Self-Regulated Learning Influences on


Middle School Mathematics Achievement

Timothy J. Cleary
Rutgers, The State University of New Jersey

Anastasia Kitsantas
George Mason University

Abstract. The primary purpose of the current study was to use structural equation
modeling to examine the relations among background variables (socioeconomic
status, prior mathematics achievement), motivation variables (self-efficacy, task
interest, school connectedness), self-regulated learning (SRL) behaviors, and
performance in middle school mathematics courses. Of particular interest was
examining the mediation roles of both self-efficacy and SRL behaviors. Data
about three types of motivation beliefs (self-efficacy, task interest, connectedness)
were obtained from 331 middle school students using self-report questionnaires,
while information regarding student SRL behaviors was obtained from teacher
ratings. Structural equation modeling analyses revealed an acceptable fit of the
data to the proposed model. In addition to the overall model explaining 51% of
the variance in mathematics performance, a key finding was that both cognitive
(i.e., self-efficacy) and behavioral (i.e., SRL) latent factors served as key medi-
ators in the model, with each of these factors exhibiting unique effects on
mathematics performance after controlling for prior achievement. Furthermore,
each of the three motivation beliefs played an important role in the model,
particularly regarding the explanation of SRL behaviors. Directions for future
research and implications for practice are discussed.

The extent to which students are capa- and school success (Boekaerts, Pintrich, &
ble of proactively and effectively regulating Zeidner, 2000; Zimmerman & Schunk, 2011).
their academic learning is an important area It is important to recognize, however, that
of interest among researchers and educators, SRL is a relatively broad construct that inte-
in part, because of the established empirical grates motivational processes, strategic behav-
link between self-regulated learning (SRL) iors, and metacognitive skills (Zimmerman,

We extend our gratitude to the students and administration at East Windsor Middle School, in particular
Lori Emmerson, for partnering with us on this project. We are also grateful for the valuable contributions
made by several doctoral research assistants including Caroline Kleeman, Audrey Lubin, Jaclyn Hogrebe,
Kristen Axelsen, Arielle Gartenberg, Bittany Velardi, Bracha Schnaidman, and Erica Pawlo.
Correspondence concerning this article should be addressed to Timothy J. Cleary, Rutgers, The State
University of New Jersey, 152 Frelinghuysen Road, Piscataway, NJ 08854; e-mail:timothy.cleary@
rutgers.edu
Copyright 2017 by the National Association of School Psychologists, ISSN 0279-6015, eISSN 2372-966x

88
Motivation and Self-Regulated Learning

2000). Furthermore, when one also considers spectives, SRL is conceptualized as a goal-
that the quality of one’s motivation and stra- directed process whereby individuals deploy
tegic behaviors will likely vary across contexts and refine their use of specific tactics or strat-
or learning situations (Cleary & Chen, 2009), egies to attain personal goals. It is important to
most researchers would agree that no single note that conceptions of academic SRL often
determinant can ever adequately explain stu- overlap to some degree with other constructs
dent achievement in school. frequently mentioned in the literature, includ-
Over the past decade, many studies have ing self-management (Briesch & Briesch,
employed path analyses and structural equa- 2015), metacognition (Flavell, 1979), and ex-
tion model (SEM) procedures to examine the ecutive functions (Barkley, 2012). For exam-
effects of motivation and regulatory processes ple, SRL and executive functions both empha-
on student outcomes and achievement (Fan- size goal-directed actions that are purposefully
tuzzo, LeBoeuf, Rouse, & Chen, 2012; Pajares used to attain those goals (Barkley, 2012;
& Graham, 1999; Sakiz, Pape, & Woolfolk Reddy, Newman, & Verdesco, 2015), whereas
Hoy, 2012). However, very few studies have SRL overlaps with metacognition because
concurrently examined a comprehensive set of both constructs underscore the relevance of
motivational variables and SRL strategic be- planning, monitoring, and reflection pro-
haviors, nor have studies examined the medi- cesses. However, SRL can be distinguished
ation role played by both student self-efficacy from these other constructs in important ways.
and SRL classroom behaviors. Addressing this Executive functions tend to represent some-
latter point is critical given the importance of what stable, higher order cognitive processes,
SRL and motivation issues in middle school whereas models of academic SRL tend to focus
contexts and because of the continued interest on malleable and highly contextualized skills,
in understanding how these factors intersect to cognitions, and behaviors (Reddy et al., 2015).
influence mathematics performance during In comparing SRL and metacognition, a few
these years (Sakiz et al., 2012). Furthermore, key differences are that models of SRL often
because motivation constructs and SRL be- underscore overt student actions, interactions
haviors are malleable in that they can be between student and environment, and the role
changed and improved, examining the direct of motivation beliefs.
and indirect effects of these multiple determi- In this study, we use a social-cognitive
nants is likely to have important implications framework as the conceptual foundation and
for instructional and intervention planning. thus focus on three core elements: (a) strategic
Thus, in this study we examined how several behaviors, (b) motivation beliefs, and (c) con-
motivational beliefs (e.g., self-efficacy, inter- textual factors. From a social-cognitive per-
est, school connectedness) and SRL behaviors spective, SRL is defined as a process involv-
displayed by middle school students in a math- ing self-initiated thoughts and behaviors that
ematics context intersected to predict student are planned and cyclically adapted to attain
performance in mathematics classrooms. personal goals (Zimmerman, 2000). Central to
this conceptualization is the focus on the qual-
CONCEPTUAL FOUNDATION ity of individuals’ strategic thinking and ac-
tions as they approach a learning activity (i.e.,
To guide the development of our struc- forethought phase), during learning (i.e., per-
tural model, we utilized social-cognitive and formance phase), and after learning has oc-
self-determination theoretical perspectives curred (i.e., reflection phase). A strong empha-
and considered the motivation and SRL lit- sis is also placed on the influence of task- or
erature. Several theoretical models concep- context-specific motivational beliefs, such as
tualize and delineate a host of processes self-efficacy perceptions (i.e., perceptions of
underlying academic SRL (Boekaerts & Ni- capability to perform specific actions) and task
emivirta, 2000; Efklides, 2011; Winne, interest (i.e., the level of interest in or enjoy-
2001; Zimmerman, 2000). From most per- ment for academic tasks), on one’s strategic
89
School Psychology Review, 2017, Volume 46, No. 1

and regulatory behaviors (Cleary, 2006; Zim- cognitive researchers but also to those adopt-
merman, 2000). Of all the potential motivation ing a self-determination theoretical perspec-
determinants, however, social-cognitive theo- tive (Ryan & Deci, 2002) or sociocultural per-
rists place primary importance on self-efficacy spective (Sakiz et al., 2012).
perceptions; these beliefs are theorized to be
the central determinant of human agency and VARIABLES IN THE SEM
have been shown to be a substantial predictor
of academic success among adolescents (Pa- In this section we provide an overview
jares & Urdan, 2006; Zimmerman, 2000). of the key variables to be included in the
Social-cognitive theorists also posit that structural model. The variables are subsumed
human functioning is determined by the recip- under the general categories of background,
rocal relations among personal (e.g., beliefs motivation beliefs, SRL, and achievement.
and perceptions), behavioral, and contextual
factors and that students’ strategic behaviors Background Variables
and motivational beliefs emerge and develop
within specific contexts (Bandura, 1997). Researchers have long been interested in
Thus, the actions students exhibit and the examining the effects of various types of con-
types of thoughts they have about learning will textual variables, such as SES and classroom
often vary across content areas or learning contexts, on student achievement and behav-
activities (Cleary & Chen, 2009; Hadwin, ior. Although some researchers have shown
Winne, Stockley, Nesbit, & Woszczyna, 2001; that controlling for biological or genetic vari-
Urdan & Midgley, 2003). To understand stu- ables will lessen the impact of SES on out-
dent achievement and performance through a comes (Bouchard & McGue, 2003), there is
social-cognitive lens, we developed a model much evidence demonstrating that family
that included contextualized motivation economic status relates directly to school
beliefs, perceptions (self-efficacy, interest, achievement and academic behaviors (By-
school connectedness), and behaviors (SRL rnes & Miller, 2006; Matthews, Kizzie,
actions), as well as other background (prior Rowley, & Cortina, 2010; Pagani, Boul-
achievement) or contextual determinants (so- erice, Vitaro, & Tremblay, 1999) although
cioeconomic status [SES]). not necessarily to motivation beliefs and stu-
Although social-cognitive theorists pay dent perceptions of context. For example,
particular attention to student beliefs about Eccles, Wong, and Peck (2006) found that
their own capabilities, they also believe that
SES had a significant effect on middle school
student perceptions of contextual factors, such
students’ academic achievement, but it did not
as how interesting a learning activity is or how
relate to students’ task values or self-concept
much support or connectedness they experi-
perceptions. For these reasons, we included
ence in school, are quite relevant in under-
SES as a background variable in our model but
standing student behavior and performance. In
fact, researchers using different theoretical or did not hypothesize that it would have a direct
conceptual paradigms have examined the role effect on student motivation processes (school
of student perceptions of various aspects of the connectedness, interest, and self-efficacy).
learning contexts on their motivated behaviors Specifically, we examined whether SES di-
and achievement, including classroom goal rectly affected mathematics achievement or if
structures (Urdan & Midgley, 2003), teacher its influence was mediated by students’ regu-
support (Patrick, Ryan, & Kaplan, 2007; Sakiz latory actions. We also included prior mathe-
et al., 2012), and connectedness or belonging matics achievement as a second background
in schools (Niehaus, Rudasill, & Rakes, 2012; variable to precisely determine the effects ex-
You et al., 2008). Thus, our inclusion of a hibited by motivational and SRL variables on
construct such as school connectedness in the student mathematics performance in school
structural model is relevant not only to social- after controlling for prior mathematics skill.
90
Motivation and Self-Regulated Learning

Relations Among Motivation Variables eral motivation variables (i.e., self-efficacy,


anxiety, task value), student effort, and prior
As indicated previously, our structural achievement and their relations to students’
model included three motivational processes: mathematics achievement. A key finding was
self-efficacy, task interest, and school connect- that after controlling for mathematics perfor-
edness. Prior research has shown that all three mance, self-efficacy emerged as a significant
motivational processes relate in meaningful predictor of mathematics outcomes but task
and important ways to each other and to values did not.
other critical outcomes in schools. In terms of Furthermore, and of particular relevance
school connectedness, relatively robust litera- to our study, is that self-efficacy beliefs often
ture has shown that students who feel con- act as a reliable mediator of the relations be-
nected to school have higher grades and tween other motivation beliefs, student
achievement (Catalano, Haggerty, Oesterle, achievement, and other important outcomes
Fleming, & Hawkins, 2004; Patrick et al., (Pajares & Kranzler, 1995; Sakiz et al., 2012).
2007), exhibit more positive emotions (Resnick For example, Sakiz et al. (2012) examined
et al., 1997), and exhibit higher levels of en- middle school students’ self-efficacy beliefs as
gagement (Wang & Eccles, 2013). Further- a mediator of the relation between motivation
more, Goodenow and Grady (1993) found that variables (enjoyment, hopelessness), sense of
school belonging accounted for a significant belonging (specific to mathematics contexts),
portion of the variance in students’ motiva- and self-reported academic effort. The authors
tion, expectancy for academic success, and found that academic self-efficacy not only had
academic task value. a direct effect on student self-reported effort in
Task interest has also been shown to be mathematics class but also mediated the im-
an important predictor of achievement in mid- pact of students’ sense of belonging and hope-
dle school and high school contexts (Cleary, lessness on self-reported effort. Furthermore,
2006; Cleary & Chen, 2009; Wolters & Pin- Wang and Eccles (2013) found that seventh-
trich, 1998). Cleary (2006) conducted a study grade students were more likely to behavior-
with 142 high school students and found that ally and emotionally engage in school if they
both task interest and instrumentality reliably had positive perceptions of relatedness (i.e.,
differentiated high and low achievers, al- support from teachers and peers). Similar to
though task interest was a more robust predic- the findings of the Sakiz et al. study, this
tor of achievement. Furthermore, in a study effect was mediated through motivation be-
with 880 middle school students, Cleary and liefs of academic self-concept and task values.
Chen (2009) showed that task interest rather In short, on the basis of the empirical literature
than task values was the strongest predictor of and the theoretical importance of self-efficacy
the frequency with which students reported perceptions in a social-cognitive framework,
using regulatory strategies during learning in we positioned self-efficacy as a central medi-
mathematics class. ator in our structural model. By doing so, we
One of the most consistent and strongest were able to examine whether efficacy medi-
findings in the literature, however, is that self- ated the effects of background variables, broad
efficacy is an extremely strong predictor of motivation perceptions (school connected-
academic achievement and behavior, even ness), and contextualized motivation beliefs
when compared to other types of self-beliefs, (math interest) on students’ SRL and mathe-
such as self-esteem, self-concept, and interest matics achievement.
(Bouffard-Bouchard, Parent, & Larivee, 1991; It is important to note, however, that
Pajares & Graham, 1999; Shell, Murphy, & task interest beliefs have also been shown to
Bruning, 1989; Zimmerman & Kitsantas, be an important predictor of academic behav-
2005). For example, Pajares and Graham iors and outcomes even after controlling for
(1999) conducted a study with middle school the effects of self-efficacy. Wolters and Pin-
students to examine the relations among sev- trich (1998) examined the relations between
91
School Psychology Review, 2017, Volume 46, No. 1

seventh- and eighth-grade students’ motiva- serve students in school across tasks and time
tion beliefs (e.g., self-efficacy, task values) and thus will often develop a strong under-
and SRL strategy use across different subject standing of students’ regulatory strengths and
areas (i.e., mathematics, social studies, and weaknesses. In addition to the studies support-
English). They reported that across all three ing the concurrent validity of teacher ratings
subject areas, the task value measure (which of student SRL, research has shown these
reflected both interest and perceived impor- measures to relate to motivation beliefs and to
tance) exhibited medium to large correlations reliably predict achievement (Cleary & Cal-
with self-efficacy and two measures of strat- lan, 2014; DiBenedetto & Zimmerman, 2013).
egy use (i.e., cognitive and regulatory strate- For these reasons, we used teacher ratings as
gies) in their study. Subsequent regression the source of student SRL behaviors in math-
analyses showed that both task value and self- ematics class.
efficacy measures accounted for unique vari- The importance of regulatory behaviors
ance in the strategy measures after controlling as a mediator has also been demonstrated
for gender and test anxiety. Thus, although across academic contexts (Fantuzzo et al.,
self-efficacy was the primary motivation me- 2012; Matthews et al., 2010; Shumow, Van-
diator in our model, we conceptualized task dell, & Posner, 1999). Fantuzzo et al. (2012)
interest as having unique effects on students’ showed that student engagement, which in-
strategy use and performance in mathematics cluded SRL indicators such as effort, positive
class. self-beliefs, and independent work habits, me-
diated the relation between risk factors (e.g.,
Self-Regulated Learning maternal education, poverty) and achievement
in both mathematics and reading. In the area
Self-regulated learners typically possess of literacy, Matthews et al. (2010) found that
a strong repertoire of cognitive and regulatory SRL-related skills, such as organization, flex-
strategies, including help seeking, elaboration, ibility, and persistence, accounted for signifi-
environmental structuring, and planning. Be- cant variance in students’ literacy skills after
cause of adaptive self-beliefs, such as high controlling for several background variables.
levels of self-efficacy and task interest, as well We extend this line of research by focusing on
as favorable perceptions of one’s connection mathematics contexts. Furthermore, on the ba-
to the learning environment (school connect- sis of these findings and the broader SRL
edness), self-regulated learners often feel em- literature, we positioned SRL behaviors as an-
powered to employ and flexibly adapt their use other central mediator in our model. We hy-
of strategies to achieve their goals. pothesized that SRL would exert a direct ef-
Much research has linked SRL strate- fect on student mathematics achievement and,
gies and achievement (Graham & Harris, along with student perceptions of efficacy,
2009; Zimmerman & Schunk, 2011), but some would serve as a robust mediator in our model.
SRL researchers have begun to question the
appropriateness of relying on self-report ques- PURPOSES
tionnaires as the primary approach for assess-
ing SRL-related behaviors and strategy use The broad objective of the current re-
(Winne & Jamieson-Noel, 2002). Because of search project was to examine the relations
potential problems with response biases and among background variables, motivational be-
inaccuracies, researchers have begun to con- liefs, SRL behaviors, and mathematics course
sider alternative SRL assessment tools includ- performance (see Figure 1). This study is im-
ing direct observations or traces, think-aloud portant because of our purposeful attempt to
protocols, and teacher rating scales (Cleary & include both motivation beliefs (i.e., self-effi-
Callan, 2014; Winne & Perry, 2000). Teacher cacy) and regulatory behaviors (i.e., SRL
ratings can be particularly valuable because strategy use in mathematics) as central medi-
teachers have repeated opportunities to ob- ators in the model. To date, very few studies in
92
Motivation and Self-Regulated Learning

Figure 1. Hypothetical Model


SES

SE

MGRADE
SRL
SC

TI

PRACH

Note. MGRADE ⫽ final mathematics grade; PRACH ⫽ prior achievement; SC ⫽ school connectedness; SE ⫽
self-efficacy; SES ⫽ socioeconomic status; SRL ⫽ self-regulated learning; TI ⫽ task interest.

the SRL literature have included both motiva- motivation beliefs as predictors of SRL behav-
tion beliefs and regulatory behaviors as medi- iors. That is, social-cognitive theories tend to
ators in the same model. This component of place much emphasis on the causal role of
our model allowed us to investigate whether human cognition on student behavior. How-
cognition (i.e., beliefs), regulatory behaviors, ever, on the basis of Bandura’s (1997) notion
or both accounted for unique variance in of reciprocal determinism, we also wanted to
achievement. test a model that reversed the sequence of the
Furthermore, because most path analy- cognition– behavior link. By testing this alter-
ses and SEMs have relied primarily on stu- native model, we would be able to determine
dents as the data source for their motivation whether the cognition-to-behavior sequence or
beliefs and regulatory behaviors (Sakiz et al., behavior-to-cognition sequence was the more
2012; Wang & Eccles, 2013), our measure- viable one. Furthermore, our inclusion of a
ment model was unique because of its multi- measure of school connectedness was impor-
source assessment focus. That is, we gathered tant from a theoretical perspective. In addition
data about student motivation directly from to school connectedness being relevant to dif-
the students but used teacher ratings to obtain ferent theoretical paradigms, this construct fo-
information about student SRL behaviors. Us- cused on the overall school context rather than
ing external sources to report on student be- the mathematics classroom context, as was the
havior is an important feature of the study case with most other variables. This method-
given that self-reported behavior often does ological feature enabled us to examine the
not correspond with what students actually do nature of the relations between a global moti-
(Winne & Jamieson-Noel, 2002). Further- vation belief (connectedness) and other con-
more, this multisource approach reduces the tent area–specific motivation beliefs (i.e.,
likelihood that an observed relation was sim- self-efficacy, task interest) along with the col-
ply a function of method-specific variance. lective influence of these different levels of
This study is also important from a the- motivation on student SRL and mathematics
oretical perspective. In developing our struc- achievement.
tural model through a social-cognitive lens, Ultimately, we were interested in exam-
we were primarily interested in examining ining both direct and indirect effects in this
93
School Psychology Review, 2017, Volume 46, No. 1

model. The primary hypotheses for the direct ers were primarily female (90%) and White
effects were guided by prior research and were (100%). It is of importance to note that the
as follows: (a) School connectedness would target middle school was selected for this proj-
relate in a positive direction to task interest, ect, in part, because of the interest expressed
self-efficacy, and SRL behaviors; (b) task in- by school administrators in further under-
terest would be associated positively with standing how motivation and SRL processes
self-efficacy, student SRL, and mathematics per- relate to mathematics performance. We also
formance; (c) self-efficacy would relate posi- selected this school because of its sociocul-
tively to student SRL and mathematics perfor- tural diversity and because we were interested
mance; and (d) student SRL would predict in specifically targeting a middle school pop-
mathematics performance. ulation. The target middle school’s perfor-
Relative to the indirect effects, we were mance on standardized mathematics and read-
primarily interested in illustrating that stu- ing examinations was comparable to state
dents’ self-efficacy and their classroom-re- averages.
lated SRL behaviors served as important me-
diators in the model. We hypothesized that (a) Motivation Measures
SES would relate to mathematics performance We used three measures of motivation:
through student SRL, (b) school connected- self-efficacy, task interest, and school connect-
ness would be positively related to self-effi- edness. All measures were pre-existing scales
cacy through task interest and positively re- established in prior research.
lated to student SRL through self-efficacy, (c)
task interest would be positively related to Self-Efficacy for Self-Regulated Learning
student SRL through self-efficacy and posi- Scale
tively related to mathematics performance The Self-Efficacy for Self-Regulated
through self-efficacy, and (d) self-efficacy Learning Scale is a seven-item measure that
would be predictive of mathematics perfor- assesses student perceptions of efficacy to reg-
mance through its relations with student SRL. ulate their behaviors during mathematics ac-
tivities conducted at home and school (Pajares
METHOD & Graham, 1999; Usher & Pajares, 2008). All
Participants items begin with the phrase “How well can
you . . .” and address various types of regula-
Three hundred sixty-three students from tory actions, such as structuring an effective
one middle school in a northeastern city par- home study environment, self-motivating,
ticipated in the study (response rate of 48.9%). managing time, and directing attention. For
Of the 363 who completed the surveys, 32 the purposes of this study, all items were cus-
were removed from the final sample as a result tomized to mathematics contexts, such as “fin-
of missing data for SES or entire scales from ish your math homework on time” and “par-
their records. The final student sample in the ticipate in math classroom discussions.” Stu-
study consisted of 331 sixth-grade students (n ⫽ dents rated their efficacy using a 5-point Likert
213, 64.4%) or seventh-grade students (n ⫽ scale ranging from 1 (not well at all) to 5 (very
118, 35.6%). Approximately 59% of the sample well). In addition to adhering to guidelines for
was female, and 24.9% of students were eligible constructing self-efficacy measures and hav-
for free or reduced-price lunch. The participating ing strong content validity (Bandura, 2006),
sample was diverse and fairly representative of this measure has been shown to exhibit large
state demographics: White, n ⫽ 142 (42.9%); positive relations with other motivation beliefs
Hispanic, n ⫽ 76 (23.0%); Asian, n ⫽ 70 (i.e., mathematics self-efficacy, task values,
(21.1%); Black, n ⫽ 17 (5.1%); and other, self-concept) and engagement in mathematics,
n ⫽ 26 (7.9%). as well as large negative relations with anxiety
Eleven mathematics teachers from the (Pajares & Graham, 1999). There is also evi-
same middle school also participated. Teach- dence for the unidimensional nature of the
94
Motivation and Self-Regulated Learning

measure and evidence that this structure is in- for the validity of inferences based on scores
variant across gender and grade level (elemen- from this measure, specifically regarding its
tary school, middle school, and high school; role as a mediator among contextual, motiva-
Usher & Pajares, 2008). The ␣ coefficient for tion, and well-being measures. That is, consis-
this scale in the current study is .78, which is tent with theoretical predictions, the School
comparable to estimates reported from prior re- Connectedness Scale has been shown to serve
search (␣ ⫽ .81; Pajares & Graham, 1999). as a mediator between school context vari-
ables and behavioral problems (Loukas et al.,
Task Interest Inventory 2006), as well as between hope and feelings of
The Task Interest Inventory is a six-item life satisfaction (You et al., 2008). The current
self-report measure designed to examine stu- estimate of internal consistency is ␣ ⫽ .76.
dents’ interest and enjoyment in specific con-
tent areas, such as science (Cleary, 2006) and SRL Measure: Self-Regulation Strategy
mathematics (Cleary & Callan, 2014; Cleary Inventory–Teacher Rating Scale
& Chen, 2009). The mathematics version was
used in this study. Students used a 5-point The Self-Regulation Strategy Invento-
Likert scale ranging from 1 (strongly dis- ry–Teacher Rating Scale (SRSI-TRS) is a 13-
agree) to 5 (strongly agree) to convey their item teacher rating scale designed to examine
perceptions. Example items include “I like the frequency with which students exhibit mo-
learning about math even when it is difficult” tivation and regulatory behaviors in classroom
and “I think math is boring.” This measure has contexts (Cleary & Callan, 2014), such as
been shown to exhibit adequate internal con- help-seeking behaviors, self-monitoring, and
sistency (␣ ⫽ .87; Cleary & Chen, 2009). self-motivation tactics. In this study, however,
Consistent with the theoretical premise and we elected to use seven items that were con-
prior research that task interest measures ceptually linked to specific types of SRL be-
should correlate with measures of SRL and haviors, such as looking for important infor-
performance, research has shown that this mation to optimize learning and seeking out
measure is a significant predictor of adaptive
help during mathematics activities when con-
and maladaptive student SRL behaviors after
fused. Example items include “The student
controlling for gender and other motivation
asks questions in class when he or she does not
beliefs (Cleary & Chen, 2009), and it reliably
understand something” and “The student asks
differentiates high- and low-achieving stu-
about the format of upcoming tests.” The
dents in middle school and high school con-
SRSI-TRS uses a 5-point Likert scale ranging
texts (Cleary, 2006; Cleary & Callan, 2014;
Cleary & Chen, 2009). The current ␣ for this from 1 (almost never) to 5 (almost always).
measure is .89. Prior research has shown the SRSI-TRS to
have high internal consistency (Cleary & Cal-
School Connectedness Scale lan, 2014; ␣ ⫽ .97). In terms of validity evi-
The School Connectedness Scale is a dence, the SRSI-TRS has been shown to con-
five-item measure designed to assess the ex- verge in expected directions with other regu-
tent to which students feel connected, sup- latory measures and motivation beliefs and has
ported, and safe in their schools (Resnick been shown to reliably predict student perfor-
et al., 1997). A 5-point Likert scale ranging mance on mathematics course examinations
from 1 (strongly disagree) to 5 (strongly (Cleary & Callan, 2014). Given that we used a
agree) was used. Example items include “I variation of the original SRSI-TRS, we con-
feel close to people at this school” and “I feel ducted confirmatory factor analysis to demon-
like I am part of this school.” This measure strate that this measure was represented by a
has been shown to have adequate internal con- single factor (see Results). In the current
sistency (␣ ⫽ .82; Loukas, Suzuki, & Horton, study, the internal consistency coefficient is
2006). There is also some empirical evidence considered high (␣ ⫽ .90).
95
School Psychology Review, 2017, Volume 46, No. 1

Mathematics Classroom Performance: practices and was provided to the primary


Mathematics Course Grade author by school administrators. Percentile
scores were used to ensure comparability of
Students’ numerical grades in their performance across grades.
mathematics class were used as an indicator of
mathematics success. Although students could Socioeconomic Status
technically earn grades from 0 to 100 on their SES was assessed using a dichotomous
report card, the actual range observed in this variable, free or reduced-price lunch and no
sample was from 56 to 100. We calculated an free or reduced-price lunch (Fantuzzo et al.,
overall numerical grade representing the aver- 2012), and students’ scores were provided to
age of the students’ quarterly report card the primary author by the school district.
grades from the third and fourth semesters Procedures
given that the primary measures were admin-
istered in the third quarter of the school year. Students completed all instruments dur-
We elected to use mathematics course grade ing a single 20- to 25-minute testing session in
rather than a standardized measure of mathe- their social studies classrooms over a 3-week
matics achievement for a couple of reasons. period during the third quarter of the school
First, given that most of our motivation and year. Students were surveyed in their social
SRL measures targeted students’ motivation studies classrooms rather than their mathemat-
and regulatory beliefs and behaviors relative ics classrooms in order to reduce potential
to their specific mathematics classes, we contextual effects or biases. Trained graduate
wanted to use an outcome measure that re- research assistants administered the surveys in
flected performance in that context. Further- each classroom. They read instructions aloud
more, because motivation and SRL variables and answered questions as needed. Teachers
are likely to influence other variables that are completed the SRSI-TRS within 2 weeks of
factored into student course grades (home- students being surveyed.
work, participation, effort), we felt that course
grades were appropriate (Bembenutty, 2013; RESULTS
Kitsantas, Cheema, & Ware, 2011; Sakiz Multilevel structural equation modeling
et al., 2012). The course grades were provided (MSEM) was used in the present research
to the authors by the participating school dis- study to account for students nested within
trict at the end of the school year. classes and measurement error (Preacher, Zy-
Background Variables phur, & Zhang, 2010). A two-phase procedure
was used: (a) A confirmatory factor analysis
Two background variables were in- was used to verify the measurement model and
cluded in the structural model: mathematics to test whether items of a construct were con-
prior achievement and socioeconomic status. sistent with the theoretical approach; and (b)
Both variables were provided to the authors by MSEM was conducted to examine the fit of
the participating school district. the complex structural model to the data using
Mplus software, version 6.12 (Muthén &
Mathematics Prior Achievement Muthén, 1998 –2015). Several indexes were
Prior mathematics skills were assessed used to evaluate the goodness of fit of the
using STAR Math Enterprise, a computer model to the data, including the ␹2 test, which
adaptive test in mathematics (Renaissance can examine the difference in fit between an
Learning, 2014). This measure was developed overidentified model and an identified model
for students in Grades 1 to 12 and targets five by taking the degrees of freedom into consid-
key domains: numbers and operations, alge- eration (Kline, 2011). Comparative fix index
bra, geometry and measurement, data analysis, (CFI) values of 0.95 are deemed good,
and statistics. This measure was given to all whereas values between 0.90 and 0.95 are
participants as part of district-wide assessment considered acceptable. Finally, the root mean
96
Motivation and Self-Regulated Learning

Table 1. Means, Standard Deviations, and Zero-Order Correlations Among


All Measures

Measures M SD 1 2 3 4 5 6 7

1. SES 0.25 0.43 —


2. PRACH 78.0 25.0 .30** —
3. SC 4.07 0.74 .06 .08 —
4. SE 4.01 0.67 .12* .23** .43** —
5. SRL 3.23 0.96 .20** .31** .10 .28** —
6. TI 3.76 0.96 .01 .06 .43** .61** .14** —
7. MGRADE 87.47 7.19 .35** .55** .20** .41** .45** .27** —

Note. MGRADE ⫽ final mathematics grade; PRACH ⫽ prior achievement; SC ⫽ school connectedness; SE ⫽
self-efficacy; SES ⫽ socioeconomic status; SRL ⫽ self-regulated learning; TI ⫽ task interest.
*p ⬍ .05. **p ⬍ .01.

square error of approximation (RMSEA), yielded an acceptable fit, ␹2(269) ⫽ 709.2,


which is the average of the residuals between p ⫽ .001, CFI ⫽ .90, Tucker–Lewis index
the observed covariance or correlation from (TLI) ⫽ .90, RMSEA ⫽ .07. However, on the
the sample and the expected model estimated basis of information provided from modifica-
from the population (Kline, 2011), provides tion indexes, we dropped two items from the
valuable information about the fit of a model. SRL subscale and one item from the self-
According to Kline (2011), an RMSEA value efficacy scale. The new model yielded an im-
lower than 0.05 indicates an excellent fit while proved fit, ␹2(203) ⫽ 413.7, p ⫽ .001, CFI ⫽
a value between 0.05 and 0.08 shows an ac- .94, TLI ⫽ .94, RMSEA ⫽ .05. The hypoth-
ceptable error of approximation. esized constructs were verified as single and
Preliminary analyses showed that de- discrete latent variables. The standardized
spite the data being multilevel (i.e., teacher loadings ranged from .47 to .92, and they were
and school), all variables were normally dis- all statistically significant at p ⬍ .01.
tributed in terms of skewness and kurtosis
(please see Table 1 for a summary of descrip- Testing the Hypothetical Structural
tive statistics for all measures). Multivariate Model
normality was also achieved as Mardia’s co- Our proposed structural model demon-
efficient indicated nonsignificant kurtosis. strated an acceptable fit to the data, ␹2(203) ⫽
Confirmatory Factor Analysis of the 445.3, p ⫽ .001, CFI ⫽ .93, TLI ⫽ .93,
Measurement Model RMSEA ⫽ .05, with the 90% confidence in-
terval (CI) lower bound of .05 and upper
Before testing the structural model, we bound of .06. Four of the paths were nonsig-
used confirmatory factor analysis to examine nificant in this model. The paths between
the validity of the latent constructs (i.e., self- school connectedness and SRL behaviors
efficacy, school connectedness, task interest, (␤ ⫽ .07, p ⫽ .35) and between prior achieve-
and SRL). All variables exhibited low to mod- ment and task interest (␤ ⫽ .05, p ⫽ .41) were
erate correlations, which eliminated any mul- not statistically significant. Furthermore, the
ticollinearity concerns. Furthermore, in our hypothesized paths between task interest and
model all factors were allowed to intercorre- both mathematics performance (␤ ⫽ .01, p ⫽
late simultaneously. The measurement model .95) and SRL behaviors (␤ ⫽ .09, p ⫽ .35)
with all the key latent constructs (self-efficacy, were not statistically significant. The final
school connectedness, task interest, and SRL) model accounted for a total of 51% of the
97
School Psychology Review, 2017, Volume 46, No. 1

variance in mathematics performance, with Table 2. Unstandardized Estimates


four variables contributing unique variance. and Direct Effect Pathways
The R2 values for each of the latent con-
structs were as follows: self-efficacy, R2 ⫽ Standard
.43; SRL behaviors, R2 ⫽ .11; and task inter- Pathway B Error ␤
est, R2 ⫽ .26.
SC 3 TI 0.20 0.03 0.50***
Because the motivation and SRL mea-
SC 3 SE 0.16 0.04 0.22***
sures were administered at a single time TI 3 SE 1.05 0.14 0.60***
point, it is possible that the flow of causality PRACH 3 SE 0.01 0.001 0.25***
between these constructs is bidirectional. PRACH 3 SRL — — 0.13**
Within the context of the larger model, we SES 3 SRL 0.15 0.004 0.13**
were specifically interested in examining SE 3 SRL 0.01 0.003 0.22***
whether SRL behavior (behavioral dimen- SES 3 MGRADE 2.30 0.67 0.14***
sion) can be a cause and an outcome of PRACH 3
MGRADE 0.11 0.003 0.45***
motivation beliefs (personal dimension). To
SE 3 MGRADE 2.24 0.48 0.26**
test this model, we reversed the positioning SRL 3 MGRADE 68.10 1.61 0.25***
of SRL and the two motivational beliefs
(i.e., self-efficacy and task interest) relative Note. MGRADE ⫽ final mathematics grade; PRACH ⫽
to the original model so that SRL behavior prior achievement; SC ⫽ school connectedness; SE ⫽
self-efficacy; SES ⫽ socioeconomic status; SRL ⫽ self-
was hypothesized to impact the motivation regulated learning; TI ⫽ task interest.
beliefs. We found that most of the path **p ⬍ .01. ***p ⬍ .001.
coefficients in the alternative model were
substantially lower than the original path
coefficients and that the new alternative
model did not adequately fit the data, Indirect Effects of Self-Efficacy and Self-
Regulatory Behaviors
␹2(203) ⫽ 465.5, p ⫽ .001, CFI ⫽ .89,
TLI ⫽ .88, RMSEA ⫽ .07, thereby provid- Of particular importance to our objec-
ing further support for our original hypo- tives, however, was the role of self-efficacy
thetical model. and self-regulatory behaviors as mediators.
Table 3 lists all indirect total effects, which
Direct Effects on Student Mathematics indicate that mediation was likely between
Performance these variables of interest. First, in terms of
self-efficacy, the indirect effect analyses
We present the standardized path coef- showed that self-efficacy was a significant me-
ficients in Table 2. In terms of mathematics diator between prior achievement and student
performance, four variables emerged as SRL behaviors (95% CI [.027, .093]), as well
unique predictors: SES (␤ ⫽ .14, p ⬍ .001); as between task interest and SRL behaviors
prior academic achievement (␤ ⫽ .45, p ⬍ (95% CI [.11, .169]). The important mediation
.001); self-efficacy beliefs (␤ ⫽ .26, p ⬍ role of SRL behaviors was also confirmed.
.001); and SRL behaviors (␤ ⫽ .25, p ⬍ .001). That is, SRL behaviors mediated the relation
That is, background variables (i.e., SES, prior between self-efficacy and mathematics perfor-
achievement), personal factors (i.e., efficacy mance (95% CI [.021, .119]), as well as be-
beliefs), and behaviors (i.e., SRL) all exerted a tween SES and mathematics performance
direct effect on mathematics performance (see (95% CI [.006, .075]). Also consistent with
Figure 2). However, on the basis of the size of expectations was that task interest mediated
the standardized regression coefficients, the the relation between school connectedness and
prior achievement variable contributed most self-efficacy (95% CI [.206, .394]).
strongly to the prediction of mathematics The importance of self-efficacy and
course grades. SRL as mediators was further illustrated when
98
Motivation and Self-Regulated Learning

Figure 2. SEM Results

SES

.13 .14
SE
2
R = .43

.22 .22 .26


.60
.03
SRL MGRADE
2 2
R =.11 R = .51
.50 .25
.30 SC
TI .09
2
R = .26
.01
.03
.25
.13 .45

PRACH

Note. Multilevel structural equation model (SEM) results are shown for prior achievement, socioeconomic status (SES),
motivational variables, self-regulated learning (SRL), and student achievement. MGRADE ⫽ final mathematics grade;
PRACH ⫽ prior achievement; SC ⫽ school connectedness; SE ⫽ self-efficacy; TI ⫽ task interest.

we examined paths including both factors as DISCUSSION


mediators. For example, in regard to the rela-
tion between prior achievement and mathe- The core objective of this study was to
matics performance, the total indirect effect examine the role of student self-efficacy and
(.12) showed that student performance in SRL behaviors as mediators within a complex
mathematics was simultaneously affected by model of background variables, motivation be-
the two mediators (self-efficacy and SRL be- liefs, and mathematics achievement. We also
haviors); however, the stronger mediator was focused on the relations among the three moti-
student self-efficacy beliefs, with a specific vation variables (i.e., self-efficacy, task interest,
indirect effect of .06. Furthermore, two spe- and school connectedness) and considered how
cific indirect paths emerged between task in- these measures collectively and separately pre-
terest and mathematics performance, with a dicted students’ SRL behaviors and achieve-
total indirect effect of .19. The first path ex- ment. The overall fit of our proposed model to
tended from task interest to mathematics per- the sample data was acceptable, and the majority
formance through both self-efficacy and SRL of our hypotheses were confirmed. The overall
behaviors, a finding that further illustrates structural model explained a robust amount of
the importance of both self-efficacy and the variance in mathematics performance (51%);
SRL behaviors. The second path indicated prior achievement appeared to contribute most to
that self-efficacy alone also served as a me- course grades, with self-efficacy and SRL be-
diator between task interest and mathemat- haviors contributing a lesser but comparable
ics performance. amount. Of particular theoretical relevance from
99
School Psychology Review, 2017, Volume 46, No. 1

Table 3. Analysis of Magnitude and Statistical Significance of Indirect


Effects

95% CI for
Indirect Effects
Total Indirect
Paths for Total and Specific Indirect Effects Effects LB UB

1. SES 3 MGRADE .04* 0.006 0.075


● SES 3 SRL 3 MGRADE
2. PRACH 3 SRL .06** 0.027 0.093
● PRACH 3 SE 3 SRL
3. SC 3 SE .30** 0.206 0.394
● SC 3 TI 3 SE
4. TI 3 SRL .14** 0.11 0.169
● TI 3 SE 3 SRL
5. SE 3 MGRADE .07** 0.021 0.119
● SE 3 SRL 3 MGRADE
6. TI 3 MGRADE .19** 0.112 0.268
● TI 3 SE 3 SRL 3 MGRADE (.04)
● TI 3 SE 3 MGRADE (.15)
7. SC 3 SRL .12** 0.069 0.171
● SC 3 SE 3 SRL (.05)
● SC 3 TI 3 SE 3 SRL (.07)
8. PRACH 3 MGRADE .12** 0.083 0.157
● PRACH 3 SE 3 SRL 3 MGRADE (.02)
● PRACH 3 SE 3 MGRADE (.06)
● PRACH 3 SRL 3 MGRADE (.04)
9. SC 3 MGRADE .16** 0.074 0.246
● SC 3 TI 3 SE 3 SRL 3 MGRADE (.02)
● SC 3 SE 3 SRL 3 MGRADE (.02)
● SC 3 SE 3 MGRADE (.05)
● SC 3 TI 3 SE 3 MGRADE (.07)

Note. Discrepancies in number calculations are due to rounding errors. CI ⫽ confidence interval; LB ⫽ lower bound;
MGRADE ⫽ final mathematics grade; PRACH ⫽ prior achievement; SC ⫽ school connectedness; SE ⫽ self-efficacy;
SES ⫽ socioeconomic status; SRL ⫽ self-regulated learning; TI ⫽ task interest; UB ⫽ upper bound.
*p ⬍ .05. **p ⬍ .01.

a social-cognitive perspective was that at least focused on college and high school students and
one variable from distinct domains of function- because the transition to middle school repre-
ing (i.e., personal, environment [SES], and sents a tremendous challenge for many students
behavior) exhibited a direct influence on perfor- (Grolnick & Raftery-Helmer, 2015), our results
mance even after controlling for prior mathemat- are timely and of particular value to middle
ics skills. When also considering the unique and school personnel.
combined mediation roles exhibited by self-effi-
cacy and SRL strategies in the model, our find- The Proposed Model
ings underscore the importance of simultane-
ously considering the influence of both cognitive A primary finding of this study was that
and behavioral influences on achievement. Fi- self-efficacy beliefs and SRL behaviors were
nally, because much of the SRL literature has fairly robust mediators in the structural model.
100
Motivation and Self-Regulated Learning

Self-efficacy beliefs mediated (i.e., alone or in Kitsantas et al. (2011) used regression analysis
conjunction with SRL behaviors) the relations to reveal a standardized ␤ weight of .46 for
between two other motivation beliefs (task self-efficacy after controlling for ethnicity,
interest, school connectedness) with both stu- gender, and homework quality. A potential
dent SRL behaviors and final mathematics reason for these discrepant findings is that four
grades. In fact, self-efficacy fully mediated the variables in our model explained a total of
effects of task interest on SRL behaviors and 51% of the variance in mathematics perfor-
the effects of task interest on final mathemat- mance. Because we included multiple predic-
ics grades. The fact that self-efficacy, but not tors of mathematics performance that cut
task interest, accounted for unique variation in across different dimensions of functioning, it
SRL and mathematics grades supports the so- is possible that variance typically accounted
cial-cognitive viewpoint that self-efficacy acts for by self-efficacy was lost to other predic-
as a critical factor in understanding academic tors. Another possible explanation pertains to
outcomes (Bandura, 1997; Pajares & Kranzler, the nature of the self-efficacy measure. A large
1995; Usher & Pajares, 2008). proportion of self-efficacy studies employing
However, a couple of qualifying points mediation analyses have used measures that
about this finding should be emphasized. First, assess perceptions of competence to attain par-
just because self-efficacy beliefs played a key ticular outcomes, such as receiving an A in
role in the model does not suggest that the class, or to perform particular academic skills,
other motivation beliefs, such as task interest such as solving specific types of mathematics
and school connectedness, were irrelevant or problems. In our study, the self-efficacy mea-
less important. In fact, both of these motiva- sure targeted students’ perceptions of capabil-
tion beliefs were strong predictors of students’ ity to regulate their learning behaviors in re-
efficacy beliefs to regulate their behaviors. lation to mathematics activities. We purpose-
Furthermore, task interest was observed to fully used this measure because of our interest
work with self-efficacy beliefs in mediating in examining the link between students’ com-
the relations between school connectedness petency perceptions of regulation and their
and both SRL behaviors and mathematics actual regulatory behaviors reported by teach-
grades. That is, one of the mediation pathways ers. In short, it is possible that self-efficacy
from connectedness to SRL and achievement measures of performance or outcomes are
involved both self-efficacy and task interest more robust predictors than self-efficacy mea-
(see Table 3). Thus, in our study we showed sures of regulation because the former are
that the effects of a broad motivation variable more closely intertwined with performance
(i.e., student connection focused on the school outcomes. A few prior studies have supported
level) on students’ SRL behaviors in mathe- this claim (Pajares & Graham, 1999; Zimmer-
matics contexts were mediated by self-efficacy man, Bandura, & Martinez-Pons, 1992).
and interest motivation beliefs linked to a In addition to examining the role of self-
mathematics context. efficacy, we were interested in determining
A second qualifying point about the cen- whether SRL behaviors in mathematics served
tral role of self-efficacy is that although it as a reliable predictor and mediator of math-
emerged as a significant predictor of mathe- ematics class performance. We found that the
matics performance (.26), the size of this path influence of regulatory behaviors on mathe-
coefficient was lower than what is typically matics course grades was comparable to that
reported in the self-efficacy literature. For ex- of the self-efficacy latent factor; a finding that
ample, Pajares and Kranzler (1995) and Pa- parallels SRL research across intervention
jares and Miller (1994) reported path coeffi- studies (Butler, 1998; Graham & Harris,
cients ranging from about .35 to .55 between 2009), path or structural equation models
self-efficacy and mathematics performance af- (Fantuzzo et al., 2012; Matthews et al., 2010),
ter controlling for the effects of both general and ex post facto designs (Cleary, 2006;
cognitive ability and prior achievement, while Cleary & Chen, 2009). As a mediator, SRL
101
School Psychology Review, 2017, Volume 46, No. 1

behaviors mediated the relations between two couple of ways. Unlike Sakiz et al. (2012), we
variables (i.e., self-efficacy and SES) on final included a measure of connectedness that op-
mathematics grades (Fantuzzo et al., 2012; erated at the school level. This enabled us to
Matthews et al., 2010) and, together with self- examine whether a more global type of moti-
efficacy, mediated the effects of task interest vation belief (school connectedness) impacted
on final grades. In short, the strategic quality content-specific motivation beliefs (efficacy,
of students’ regulatory behaviors exhibited interest) and subsequent regulatory behaviors.
during classroom activities, such as seeking Given our results that school connectedness
out help from others when confused, not only exerted a direct effect on specific, contextual-
served as a direct predictor of their mathemat- ized motivation beliefs and an indirect effect
ics outcomes but also mediated the effects of on SRL behaviors, it appears that feelings of
how students think and feel about themselves connectedness, regardless of whether they are
as learners and the contexts in which they directed at the school level or at a specific
learn. classroom level, are very important when at-
A few additional points should be made tempting to understand student motivation and
regarding the observed relations within the behaviors within that classroom context. Fur-
model. One critical issue pertains to the em- thermore, the fact that we gathered data about
pirical link between a cognitive dimension of student SRL behaviors from an external
SRL (self-efficacy for SRL), a behavioral di- source rather than from student self-reports
mension (classroom SRL behaviors), and provides confirming evidence that how stu-
achievement. We demonstrated that students’ dents think about learning and themselves is a
perceptions of efficacy to regulate and their strong predictor of the things they do in
actual regulatory behaviors (as reported by school.
teachers) in the classroom exerted unique ef-
fects on their achievement. Thus, beliefs and Limitations and Areas of Future
actions were both found to be important for Research
understanding their achievement. Further-
more, our test of an alternative structural Although the findings of this study are
model enabled us to determine whether the important, there are a few limitations. First,
sequential link between motivation beliefs and we administered several measures at a single
regulatory behaviors was best represented as a point in time and we did not manipulate any
cognition-to-behavior pathway or vice versa. variables. Thus, we were not able to make
Our analyses revealed that the cognition-to- causal claims regarding the relations among
behavior causal sequence was more credible; the three dimensions of personal, behavior,
it resulted in stronger observed relations and and environment. Given that the transition
an overall better model fit. Although this find- to middle school is often problematic for
ing is theoretically important in that it supports many students from a motivational and regu-
the social-cognitive perspective that cognition latory perspective (Grolnick & Raftery-Hel-
has a key deterministic influence on behavior, mer, 2015), it is critical that future research
the finding also has important implications for longitudinally investigates how motivation be-
educators (see Implications for Educators and liefs and SRL behaviors vary over time and
School Psychologists subsection). across situations and contexts. These types of
Another important point involved the longitudinal analyses may prove useful in
link between school connectedness, self-effi- more directly assessing the causal links among
cacy and interest, and student behaviors. Re- different types of motivation processes, regu-
cent research has shown that a sense of con- latory behaviors, and achievement.
nectedness to a specific classroom context is Second, although the observed fit in-
an important predictor of other motivational dexes for our proposed model were adequate
processes and effort within that context (Sakiz and stronger than an alternative model, we
et al., 2012). We complemented this study in a make no claims that our model is the best
102
Motivation and Self-Regulated Learning

possible one or that it includes all potential Implications for Educators and School
variables relevant to understanding mathemat- Psychologists
ics performance. Clearly, a variety of cogni-
tive abilities, such as working memory, pro- Even though our research design did not
cessing speed, and visual–spatial skills, have enable us to draw causal inferences regarding
been linked to different types of mathematics the link among context, motivation, regula-
outcomes (Mather & Wendling, 2005). We tion, and performance, the observed findings
also did not include many of the motivation are relevant to school-based personnel and
processes emphasized across different theo- practices. First, it is important for school per-
retical paradigms, such as autonomy, growth sonnel in middle school contexts to recognize
mindset, perceived instrumentality, or goal that academic performance is directly influ-
orientation, and did not delve into issues enced by many different variables, such as
such as task or content area difficulty (Sc- students’ background and prior knowledge,
hunk & Zimmerman, 2008). Future research the quality of their perceptions of classroom
should consider including such processes to contexts and their own skills and capabilities,
further enhance our understanding of how and the quality of their engagement and regu-
different motivation processes and contex- latory behaviors.
tual factors intersect to promote adaptive Although most school personnel recog-
student behavior. nize the importance of using behavioral prin-
Although we intentionally used teacher ciples and techniques, such as antecedent con-
ratings of student SRL behaviors to avoid trol, reinforcement, and functional behavior
overreliance on student reports, it is possible assessments, to optimize student behavior and
that the teacher ratings were biased because of achievement (Akin-Little, Eckert, Lovett, &
their prior knowledge about student skill or Little, 2004; Kruger et al., 2016), our findings
performance. That is, some teachers may have suggest that it is also relevant to directly assess
simply provided higher ratings for more capa- and target students’ self-perceptions and be-
ble or skilled students and lower ratings for liefs. However, because various self-percep-
less skilled students. From our perspective, it tions and beliefs are often intertwined and
is important for researchers to continue to use connected, educators who are interested in
different types of measures to assess student empowering youth should not simply target
behavior and SRL so that we can enhance the one type of belief or self-perception; they
validity of inferences that are made regarding need to concurrently examine multiple be-
observed results or findings from a particular liefs such as how confident students feel
study. about performing a specific task, their inter-
The use of students’ report card grades est in that activity, and their perceptions of
as an indicator of mathematics performance value or importance of that activity, as well
can also be viewed as a limitation. Although as their perceptions about the level of sup-
we were specifically interested in students’ ports, encouragement, and connection they
performance in mathematics classes rather experience in school. These recommenda-
than a broad measure of mathematics skill, the tions are particularly relevant to middle
use of a report card grade is fairly broad and school contexts because of the enhanced rigor
may be influenced by variables other than and demands that middle school students ex-
mathematics skills (e.g., class participation, perience and their corresponding vulnerability
homework completion). Thus, course grades to frustration, poor motivation, and ultimately,
may not be representative of students’ actual low levels of achievement (Byrnes & Miller,
mathematics skills. Researchers who are spe- 2006; Cleary & Chen, 2009; Grolnick & Raf-
cifically interested in examining mathematics tery-Helmer, 2015; Urdan & Midgley, 2003).
skill rather than classroom performance Furthermore, because the key variables
should use standardized measures of mathe- in our study (connectedness, self-efficacy, task
matics as their primary academic outcome. interest, and SRL behaviors) have been shown
103
School Psychology Review, 2017, Volume 46, No. 1

to be malleable in nature (i.e., they can be Hughes, 2000). Not surprisingly, in both lines
modified and changed over time), our results of research, the participants reported that pro-
have implications for both assessment and in- fessional development training in SRL-related
tervention planning and development. In the processes was a top priority and was of critical
SRL literature, researchers have increasingly value for helping them to refine and develop
emphasized the use of dynamic, contextual- their teaching and professional competencies
ized forms of assessment to track student be- (Cleary et al., 2010; Coalition for Psychology
liefs and self-perceptions as they engage in in Schools and Education, 2006; Wehmeyer
particular learning activities and to use that et al., 2000). From our perspective, profes-
information to inform intervention planning sional development training and workshops
(Cleary & Platten, 2013; Winne & Perry, are needed to help educators and school psy-
2000). For example, Cleary & Platten (2013) chologists become more proficient in under-
used structured interviews to assess academi- standing the interplay among student beliefs,
cally at-risk high school students’ perceptions regulatory behaviors, and learning contexts, as
about the causes of their academic struggles well as how this interplay can be addressed via
(i.e., attributions) and used that information as academic interventions within problem-solv-
part of an intervention to help students adapt ing frameworks.
and improve their strategic behaviors. Further-
more, as part of the Self-Regulated Strategy REFERENCES
Development intervention program for im-
Akin-Little, K. A., Eckert, T. L., Lovett, B. J., & Little,
proving written expression skills, Graham and S. G. (2004). Extrinsic reinforcement in the classroom:
Harris (2009) have repeatedly demonstrated Bribery or best practice. School Psychology Review,
that instruction that seeks to simultaneously 33(3), 344 –362.
Bandura, A. (1997). Self-efficacy: The exercise of control.
enhance students’ motivation beliefs and stra- New York, NY: W.H. Freeman.
tegic behaviors within the context of writing Bandura, A. (2006). Guide for creating self-efficacy
will lead to substantial improvements in per- scales. In F. Pajares & T. Urdan (Eds.), Self-efficacy
beliefs of adolescents (pp. 307–338). Charlotte, NC:
formance. In short, directly assessing how stu- Information Age.
dents think and reflect on their learning is a Barkley, R. A. (2012). Executive functions: What they are,
core element of many SRL intervention pro- how they work, and why they evolved. New York, NY:
Guilford Press.
grams designed to empower students to take Bembenutty, H. (2013). The triumph of homework com-
greater control over their success in school pletion through a learning academy of self-regulation.
(Butler, 1998; Cleary & Platten, 2013; Gra- In H. Bembenutty, T. J. Cleary, & A. Kitsantas (Eds.),
ham & Harris, 2009). Applications of self-regulated learning across diverse
disciplines: A tribute to Barry J. Zimmerman (pp.
Finally, our findings are also in line with 153–196). Charlotte, NC: Information Age.
recent survey research illustrating a gap in Boekaerts, M., & Niemivirta, M. (2000). Self-regulated
practice regarding the frequency with which learning: Finding a balance between learning goals and
ego-protective goals. In M. Boekaerts, P. Pintrich, &
teachers and school psychologists incorpo- M. Zeidner (Eds.), Handbook of self-regulation (pp.
rate SRL and motivation principles into their 417– 450). Cambridge, MA: Academic Press.
professional activities. For example, Cleary Boekaerts, M., Pintrich, P. R., & Zeidner, M. (Eds.).
(2000). Handbook of self-regulation. Cambridge, MA:
(2009) and Cleary, Gubi, and Prescott Academic Press.
(2010) showed that school psychologists do Bouchard, T. J., Jr., & McGue, M. (2003). Genetic and
not routinely evaluate SRL processes nor do environmental influences on human psychological dif-
ferences. Journal of Neurobiology, 54(1), 4 – 45. doi:
they systematically implement motivation 10.1002/neu.10160
and SRL interventions for youth with aca- Bouffard-Bouchard, T., Parent, S., & Larivee, S. (1991).
demic or learning problems. Similar surveys Influence of self-efficacy on self-regulation and perfor-
mance among junior and senior high-school age students.
with teachers have revealed comparable find- International Journal of Behavioral Development, 14(2),
ings; teachers often do not routinely incorpo- 153–164. doi:10.1177/016502549101400203
rate SRL-related principles into their instruc- Briesch, A. M., & Briesch, J. M. (2015). Self-management
interventions to reduce disruptive classroom behavior.
tion and curriculum (Grigal, Neubart, Moon, In T. J. Cleary (Ed.), Self-regulated learning interven-
& Graham, 2003; Wehmeyer, Agran, & tions with at-risk youth: Enhancing adaptability, per-

104
Motivation and Self-Regulated Learning

formance, and well-being (pp. 45– 65). Washington, resilience. Journal of School Psychology, 50(5), 559 –
DC: APA Publication Press. 579. doi:10.1016/j.jsp.2012.04.004
Butler, D. (1998). The strategic content learning approach Flavell, J. H. (1979). Metacognition and cognitive moni-
to promoting self-regulated learning: A report of three toring: A new area of cognitive-developmental inquiry.
studies. Journal of Educational Psychology, 90(4), American Psychologist, 34(10), 906 –911. doi:10.1037/
682– 697. doi:10.1037/0022-0663.90.4.682 0003-066x.34.10.906
Byrnes, J. P., & Miller, D. C. (2006). The relative Goodenow, C., & Grady, K. E. (1993). The relationship of
importance of predictors of math and science school belonging and friends’ values to academic mo-
achievement: An opportunity-propensity analysis. tivation among urban adolescent students. The Journal
Contemporary Educational Psychology, 32(4), 599 – of Experimental Education, 62(1), 60 –71. doi:10.1080/
629. doi:10.1016/j.cedpsych.2006.09.002 00220973.1993.9943831
Catalano, R. F., Haggerty, K. P., Oesterle, S., Fleming, Graham, S., & Harris, K. R. (2009). Almost 30 years of
C. B., & Hawkins, J. D. (2004). The importance of writing research: Making sense of it all with the Wrath
bonding to school for health development: Findings of Khan. Learning Disabilities Research and Practice,
from the social development research group. Journal 24(2), 58 – 68. doi:10.1111/j.1540-5826.2009.01277.x
of School Health, 74(7), 252–261. doi:10.1111/j.1746- Grigal, M., Neubart, D. A., Moon, S. M., & Graham, S.
1561.2004.tb08281.x (2003). Self-determination for students with disabili-
Cleary, T. J. (2006). The development and validation of ties: Views of parents and teachers. Exceptional Chil-
the self-regulation strategy inventory—Self-report. dren, 70(1), 97–112.
Journal of School Psychology, 44(4), 307–322. doi: Grolnick, W. S., & Raftery-Helmer, J. N. (2015). Contexts
10.1016/j.jsp.2006.05.002 supporting self-regulated learning at school transitions.
Cleary, T. J. (2009). School-based motivation and self- In T. J. Cleary (Ed.), Self-regulated learning interven-
regulation assessments: An examination of school tions with at-risk youth: Enhancing adaptability, per-
psychologist beliefs and practices. Journal of Ap- formance, and well-being (pp. 251–276). Washington,
plied School Psychology, 25(1), 71–94. doi:10.1080/ DC: APA Publication Press.
15377900802484190 Hadwin, A. F., Winne, P. H., Stockley, D. B., Nesbit,
Cleary, T. J., & Callan, G. L. (2014). Student self-regu- J. C., & Woszczyna, C. (2001). Context moderates
lated learning in an urban high school: Predictive va- students’ self-reports about how they study. Journal of
lidity and relations between teacher ratings and student Educational Psychology, 93(3), 477– 487. doi:10.1037/
self-reports. Journal of Psychoeducational Assessment, 0022-0663.93.3.477
32(4), 295–305. doi:10.1177/0734282913507653 Kitsantas, A., Cheema, J., & Ware, H. W. (2011). Math-
Cleary, T. J., & Chen, P. (2009). Self-regulation, motiva- ematics achievement: The role of homework and self-
tion, and math achievement in middle school: Varia- efficacy beliefs. Journal of Advanced Academics,
tions across grade level and math context. Journal 22(2), 310 –339. doi:10.1177/1932202x1102200206
of School Psychology, 47(5), 291–314. doi:10.1016/ Kline, R. B. (2011). Principles and practice of structural
j.jsp.2009.04.002 equation modeling (3rd ed.). New York, NY: Guilford
Cleary, T. J., Gubi, A., & Prescott, M. V. (2010). Moti- Press.
vation and self-regulation assessments in urban and Kruger, A. M., Strong, W., Daly, E. J., O’Connor, M.,
suburban schools: Professional practices and needs of Sommerhalder, M. S., Holtz, J., . . . Heifner, A. (2016).
school psychologists. Psychology in the Schools, Setting the stage for academic success through ante-
47(10), 985–1002. doi:10.1002/pits.20519 cedent intervention. Psychology in the Schools, 53(1),
Cleary, T. J., & Platten, P. (2013). Examining the corre- 24 –38. doi:10.1002/pits.21886
spondence between self-regulated learning and aca- Loukas, A., Suzuki, R., & Horton, K. D. (2006). Examining
demic achievement: A case study analysis. Education school connectedness as a mediator of school climate
Research International. 2013, 1–18. doi:10.1155/2013/ effects. Journal of Research on Adolescence, 16(3), 491–
272560 502. doi:10.1111/j.1532-7795.2006.00504.x
Coalition for Psychology in Schools and Education. Mather, N., & Wendling, B. J. (2005). Linking cognitive
(2006, August). Report on the teacher needs survey. assessment results to academic interventions for stu-
Washington, DC: American Psychological Associa- dents with learning disabilities. In D. P. Flanagan &
tion, Center for Psychology in Schools and Education. P. L. Harrison (Eds.), Contemporary intellectual as-
DiBenedetto, M. K., & Zimmerman, B. J. (2013). Con- sessment: Theories, tests, and issues (pp. 269 –294).
struct and predictive validity of microanalytic mea- New York, NY: Guilford Press.
sures of students’ self-regulation of science learning. Matthews, J. S., Kizzie, K. T., Rowley, S. J., & Cortina,
Learning and Individual Differences, 26, 30 – 41. doi: K. (2010). African Americans and boys: Under-
10.1016/j.lindif.2013.04.004 standing the literacy gap, tracing academic trajecto-
Eccles, J. S., Wong, C. A., & Peck, S. C. (2006). Ethnicity ries, and evaluating the role of learning-related
as a social context for the development of African- skills. Journal of Educational Psychology, 102(3),
American adolescents. Journal of School Psychology, 757–771. doi:10.1037/a0019616
44(5), 407– 426. doi:10.1016/j.jsp.2006.04.001 Muthén, L. K., & Muthén, B. O. (1998 –2015). Mplus
Efklides, A. (2011). Interactions of metacognition with user’s guide (7th ed.) Los Angeles, CA: Author.
motivation and affect in self-regulated learning: The Niehaus, K., Rudasill, K. M., & Rakes, C. R. (2012). A
MASRL model. Educational Psychologist, 46(1), longitudinal study of school connectedness and academic
6 –25. doi:10.1080/00461520.2011.538645 outcomes across sixth grade. Journal of School Psychol-
Fantuzzo, J., LeBoeuf, W., Rouse, H., & Chen, C. (2012). ogy, 50(4), 443– 460. doi:10.1016/j.jsp.2012.03.002
Academic achievement of African American boys: A Pagani, L., Boulerice, B., Vitaro, F., & Tremblay, R. E.
city-wide, community-based investigation of risk and (1999). Effects of poverty on academic failure and

105
School Psychology Review, 2017, Volume 46, No. 1

delinquency in boys: A change and process model mentary school children. Merrill-Palmer Quarterly,
approach. Journal of Child Psychology and Psychiatry, 45, 309 –331.
40(8), 1209 –1219. doi:10.1017/s0021963099004783 Urdan, T., & Midgley, C. (2003). Changes in the per-
Pajares, F., & Graham, L. (1999). Self-efficacy, motiva- ceived classroom goal structure and pattern of adap-
tion constructs, and mathematics performance of en- tive learning during early adolescence. Contempo-
tering middle school students. Contemporary Educa- rary Educational Psychology, 28(4), 524 –551. doi:
tional Psychology, 24(2), 124 –139. doi:10.1006/ceps. 10.1016/s0361-476x(02)00060-7
1998.0991 Usher, E. L., & Pajares, F. (2008). Self-efficacy for self-
Pajares, F., & Kranzler, J. (1995). Self-efficacy beliefs and regulated learning: A validation study. Educational
general mental ability in mathematical problem-solv- and Psychological Measurement, 68(3), 443– 463. doi:
ing. Contemporary Educational Psychology, 20(4), 10.1177/0013164407308475
426 – 443. doi:10.1006/ceps.1995.1029 Wang, M., & Eccles, J. S. (2013). School context,
Pajares, F., & Miller, M. D. (1994). Role of self-efficacy achievement motivation, and academic engagement: A
and self-concept beliefs in mathematics problem-solv- longitudinal study of school engagement using a mul-
ing: A path analysis. Journal of Educational Psychol- tidimensional perspective. Learning and Instruction,
ogy, 86(2), 193–203. doi:10.1037/0022-0663.86.2.193 28, 12–23. doi:10.1016/j.learninstruc.2013.04.002
Pajares, F., & Urdan, T. (Eds.). (2006). Self-efficacy be- Wehmeyer, M. L., Agran, M., & Hughes, C. A. (2000).
liefs of adolescents. Charlotte, NC: Information Age. National survey of teachers’ promotion of self-de-
Patrick, H., Ryan, A. M., & Kaplan, A. (2007). Early termination and student-directed learning. The Jour-
adolescents’ perceptions of the classroom social envi- nal of Special Education, 34(2), 58 – 68. doi:
ronment, motivational beliefs, and engagement. Jour- 10.1177/002246690003400201
nal of Educational Psychology, 99(1), 83–98. doi: Winne, P. H. (2001). Self-regulated learning viewed from
10.1037/0022-0663.99.1.83 models of information processing. In B. J. Zimmerman
Preacher, K. J., Zyphur, M. J., & Zhang, Z. (2010). A & D. H. Schunk (Eds.), Self-regulated learning and
general multilevel SEM framework for assessing mul- academic achievement: Theoretical perspectives (2nd
tilevel mediation. Psychological Methods, 15(3), 209 – ed., pp. 153–189). Mahwah, NJ: Lawrence Erlbaum.
233. doi:10.1037/a0020141 Winne, P. H., & Jamieson-Noel, D. L. (2002). Explor-
Reddy, L. A., Newman, E., & Verdesco, A. (2015). Use of ing students’ calibration of self-reports about study
self-regulated learning for children with ADHD: Re- tactics and achievement. Contemporary Educational
search and practice opportunities. In T. J. Cleary (Ed.), Psychology, 27(4), 259 –276. doi:10.1016/s0361-
Self-regulated learning interventions with at-risk 476x(02)00006-1
Winne, P. H., & Perry, N. E. (2000). Measuring self-
youth: Enhancing adaptability, performance, and well-
regulated learning. In M. Boekaerts, P. Pintrich, & M.
being (pp. 15– 43). Washington, DC: APA Publication
Zeidner (Eds.), Handbook of self-regulation (pp. 531–
Press.
566). Cambridge, MA: Academic Press.
Renaissance Learning. (2014). The research foundation
Wolters, C. A., & Pintrich, P. R. (1998). Contextual
for STAR assessment. Wisconsin Rapids, WI: Author.
differences in student motivation and self-regulated
Resnick, M. D., Bearman, P. S., Blum, R. W., Bauman,
learning in mathematics, English, and social studies
K. E., Harris, K. M., Jones, J., . . . Udry, J. R. (1997).
classrooms. Instructional Science, 26, 27– 47.
Protecting adolescents from harm. Findings from the
You, S., Furlong, M. J., Felix, E., Sharkey, J. D., Tani-
national longitudinal study on adolescent health. Jour- gawa, D., & Green, J. G. (2008). Relations among
nal of the American Medical Association, 278(10), school connectedness, life satisfaction, and bully vic-
823– 832. doi:10.1001/jama.1997.03550100049038 timization. Psychology in the Schools, 45(5), 446 – 460.
Ryan, R. M., & Deci, E. L. (2002). An overview of Zimmerman, B., & Schunk, D. (Eds.). (2011). Handbook
self-determination theory: An organismic-dialectical of self-regulation of learning and performance. New
perspective. In E. L. Deci & R. M. Ryan (Eds.), Hand- York, NY: Routledge.
book of self-determination research (pp. 3–36). Roch- Zimmerman, B. J. (2000). Attaining self-regulation: A
ester, NY: University of Rochester Press. social-cognitive perspective. In M. Boekaerts, P. Pin-
Sakiz, G., Pape, S. J., & Woolfolk Hoy, A. (2012). trich, & M. Zeidner (Eds.), Handbook of self-regula-
Does perceived teacher affective supports matter for tion (pp. 13–39). Cambridge, MA: Academic Press.
middle school students in mathematics classrooms? Zimmerman, B. J., Bandura, A., & Martinez-Pons, M.
Journal of School Psychology, 50(2), 235–255. doi: (1992). Self-motivation for academic attainment.
10.1016/j.jsp.2011.10.005 American Educational Research Journal, 29(3), 663–
Schunk, D. H., & Zimmerman, B. J. (Eds.). (2008). Mo- 676. doi:10.3102/00028312029003663
tivation and self-regulated learning: Theory, research, Zimmerman, B. J., & Kitsantas, A. (2005). Homework
and applications. Mahwah, NJ: Lawrence Erlbaum. practices and academic achievement: The mediating
Shell, D. F., Murphy, C. C., & Bruning, R. H. (1989). role of self-efficacy and perceived responsibility be-
Self-efficacy and outcome expectancy mechanisms in liefs. Contemporary Educational Psychology, 30(4),
reading and writing achievement. Journal of Educa- 397– 417. doi:10.1016/j.cedpsych.2005.05.003
tional Psychology, 81(1), 91–100. doi:10.1037/0022-
0663.81.1.91
Shumow, L., Vandell, D. L., & Posner, J. (1999). Risk Date Received: April 14, 2016
and resilience in the urban neighborhood: Predictors Date Accepted: August 3, 2016
of academic performance among low-income ele- Associate Editor: Erin Dowdy 䡲

106
Motivation and Self-Regulated Learning

Timothy J. Cleary is an associate professor in the Department of Applied Psychology in


the Graduate School of Applied and Professional Psychology at Rutgers, The State
University of New Jersey. His primary research interests include the development and
application of self-regulated learning (SRL) and motivation assessment and intervention
practices across academic, athletic, and clinical contexts. Dr. Cleary also provides
coaching and professional development training to school districts regarding best prac-
tices in motivation and SRL.

Anastasia Kitsantas is professor of educational psychology in the College of Education


and Human Development at George Mason University. Her research focuses on the role
of self-regulation on learning and performance across diverse areas of functioning,
including academics, athletics, and health. She is particularly interested in the develop-
ment of student SRL and motivation.

107

You might also like