Sunteți pe pagina 1din 10

Learning and Individual Differences 21 (2011) 337346

Contents lists available at ScienceDirect

Learning and Individual Differences


j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / l i n d i f

A meta-analytic review of the Motivated Strategies for Learning Questionnaire


Marcus Cred a,, L. Alison Phillips b
a
Department of Psychology, University at Albany, State University of New York, 1400 Washington Avenue, Albany, NY 12222, United States
b
Department of Psychology, Rutgers, The State University of New Jersey, 112 Paterson Street, New Brunswick, NJ 08901, United States

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

Article history: The current paper presents a meta-analytic review of the Motivated Strategies for Learning Questionnaire
Received 21 July 2010 (MSLQ), which consists of fteen subscales designed from classic socialcognitive learning theories and
Received in revised form 25 January 2011 which is widely used to predict academic performance. Results based on 2158 correlations from 67
Accepted 12 March 2011
independent samples and 19,900 college students indicate that the subscales of the MSLQ vary in their utility
for predicting grades, with grade-related validities ranging from = .40 for the subscale measuring students'
Keywords:
Motivated Strategies for Learning
effort regulation to = .05 for the subscale measuring students' help-seeking behaviors. Factor-analyses of
Questionnaire (MSLQ) the meta-analytic intercorrelations broadly support the theoretical structure of the MSLQ. Alteration or
Meta-analysis elimination of items with undesirable psychometric characteristics could potentially both augment empirical
Academic performance support for the theoretical structure of the MSLQ and strengthen its subscales' predictive utility for academic
Socialcognitive learning theories performance.
Motivation 2011 Elsevier Inc. All rights reserved.
Learning strategies

1. Introduction focused specically on the manner in which students engage with


academic tasks and material and has sought to shed further light onto
Researchers have predominantly focused on stable student traits, both the determinants of academic performance and the processes
abilities, and behaviors to predict academic performance in college. of learning is represented by the self-regulated learning literature
For example, a substantial literature (e.g., Halpin, Halpin, & Schaer, (e.g., Zimmerman, 1990; Zimmerman & Schunk, 1989; Rotgans &
1981; Hezlett et al., 2001; Kuncel, Hezlett, & Ones, 2001) has provided Schmidt, 2009).
evidence of the relatively strong prediction of college grades by scores Students who engage in self-regulated learning are dened as
on standardized admissions tests, which theoretically assess a metacognitively, motivationally, and behaviorally active participants
student's stable cognitive ability, which reects the ability to process, in their learning (Zimmerman, 1990, p. 4). That is, students who
understand, and remember new information (Humphreys, 1979). engage in self-regulated learning monitor their own progress toward
Others (e.g., Lievens, Coetsier, De Fruyt, & De Maeseneer, 2002; self-set goals and are therefore able to reect on the effectiveness of
O'Connor & Paunonen, 2007; Poropat, 2009; Robbins et al., 2004) have their learning approaches (i.e., have high levels of meta-cognition),
examined the relationship between academic performance and tend to view the learning task as intrinsically interesting and
theoretically stable traits, such as personality (e.g., conscientiousness) worthwhile while having high levels of self-efcacy (i.e., have learning
and motivational factors (e.g., need for achievement), that are thought appropriate cognitions and motivations), and engage in and persist
to directly affect important academic behaviors such as study skills with learning behaviors that maximize the degree to which learning
and class attendance, which are themselves strongly predictive of occurs (i.e., engage in learning appropriate behaviors). All three
college grades (e.g., Cred & Kuncel, 2008; Crede, Roch, & Kieszczynka, components of self-regulated learning (meta-cognitions, motivations,
2010). While traits, abilities, and skills explain a substantial propor- and behaviors) are assumed to be important determinants of learning
tion of the variation in academic performance, research in these areas and hence academic performance, although the effect of meta-cogni-
have been less successful at providing insights into the specic man- tion and motivations on academic performance are typically presen-
ner in which students acquire new knowledge. One approach that has ted as being mediated through learning behaviors (e.g., Duncan &
McKeachie, 2005), such that meta-cognition and appropriate motiva-
tions result in the use of appropriate learning strategies that, in turn,
have a positive effect on academic performance.
An important theoretical addition to the theory of self-regulated
Abbreviations: MSLQ, Motivated Strategies for Learning Questionnaire; GPA, Grade learning came from researchers and theorists (e.g., Duncan &
Point Average; SD, Standard Deviation.
Corresponding author. Tel.: + 1 518 442 2590; fax: + 1 518 442 4867.
McKeachie, 2005; Dweck & Leggett, 1988; Pintrich, 2000; Schunk,
E-mail addresses: mcrede@albany.edu (M. Cred), lphillips@ifh.rutgers.edu 2005a, 2005b; Zimmerman, 1989) who advocated a socialcognitive
(L.A. Phillips). view of learning, which suggests that self-regulated learning exhibits

1041-6080/$ see front matter 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.lindif.2011.03.002
338 M. Cred, L.A. Phillips / Learning and Individual Differences 21 (2011) 337346

not only between-person variation but also within-person variation. by comparing the utility of the MSLQ subscales for predicting course-
That is, self-regulation is seen as situation specic, and self-regulated specic performance to their utility for predicting general academic
learning can therefore be used to explain why one student performs performance (GPA). The third purpose is to highlight potentially
better on an academic task than another student (i.e., explain problematic psychometric properties of some of the MSLQ items that
between-person variation in academic performance), but can also may be limiting the subscales' predictive utility and ability to contri-
be used to explain why a single student performs better on one bute to learning theories. The MSLQ has been used in non-college
academic task than on other academic tasks (i.e., explain within- settings (e.g., Pintrich & DeGroot, 1990) but its most frequent use has
person variation). While meta-cognitive and self-regulatory abilities been in a college setting and we therefore focus specically on the
are likely to be relatively constant across both time and academic relationship between scores on MSLQ subscales and academic perfor-
tasks due to their trait-like nature, socialcognitive theories of learn- mance in college.
ing suggest that both the motivations and cognitions relating to a task
(e.g., judgments regarding the intrinsic value of a task), and the
learning behaviors used by a student to perform that task can vary 2. Overview of the MSLQ
substantially across tasks. For example, a student's motivations,
cognitions, and learning behaviors for academic tasks may vary across The MSLQ is an 81-item self-report measure comprised of fteen
different classes (e.g., advanced seminar in chosen major versus a subscales that assess both student motivation to engage with course
required general education class), and even across different tasks material and their learning strategies. The full MSLQ takes 2030 min
within the same class (e.g., studying for a multiple choice exam versus to complete, although individual subscales can also be administered
writing a term paper). This within-person variation in learning pro- (Pintrich, Smith, Garcia, & McKeachie, 1991). All items are Likert scale
cesses suggests need for a contextualized (task specic) examination items and students use a seven-point response option format
of self-regulated learning processes. (1 = Not at all true of me to 7 = Very true of me) to respond to
Empirical investigations of self-regulated learning processes and each item. Eight of the 81-items are reverse scored.
mechanisms require the existence of valid and reliable measures of the There are a total of thirty-one item and six subscales that attempt
various constructs that are central to self-regulated learning. This to assess students' motivational orientations toward a college course.
need led Paul Pintrich and colleagues to develop the Motivated These six scales were designed to capture three theoretical compo-
Strategies for Learning Questionnaire (MSLQ, Pintrich, Smith, Garcia & nents of motivation, termed value beliefs, expectancy; and affect,
McKeachie, 1991) a single measure designed to assess task-specic (Duncan & McKeachie, 2005). Value beliefs are assessed with three
cognitions and motivations as well as the learning strategies used by motivation subscales: a four-item intrinsic goal orientation scale that
students to engage with that task. Tens of thousands of students have assesses whether a student's participation in the academic task is an
since been evaluated on the MSLQ; a search of the PsychINFO database end in itself and whether the student is focused on the mastery of
reveals over 150 articles that have used the MSLQ and the MSLQ material, a four-item extrinsic goal orientation scale that assesses the
manual has itself been cited over 700 times. The MSLQ has been used to degree to which a student participates in the task in order to obtain
study college students in a wide variety of countries (e.g., Suksamram, good grades, rewards or approval, and a six-item task value scale that
2003; Thongnoum, 2002), settings (e.g., community colleges, Welsh, assesses the degree to which the student has a belief that the class
2007; online classes, Gaythwaite, 2006), and populations (e.g., material is interesting and worth learning. Expectancy is assessed
teaching students, Bhattacharyya, 2004; nursing students, Tutor, with two subscales: self-efcacy (condence that a task can be
2006). However, the use of the MSLQ in these studies is almost always performed) and control beliefs (akin to an internal locus of control);
as a static predictor of academic performance; the context-specicity both capture the degree to which students believe that they have
of the constructs included in the MSLQ as implied by socialcognitive control over their level of achievement in a class. Finally, the affective
views of learning have not yet been rigorously tested except for a component of motivation is measured by the test anxiety subscale,
single recent paper (Rotgans & Schmidt, 2009) that showed signicant which attempts to capture the degree to which students experience
variation in self-regulated learning processes across classes. Review- anxiety and fear when taking tests.
ing the results of studies that have used the MSLQ would, however, The learning strategies subscales are comprised of fty items
provide valuable information on the validity of the more general divided into nine subscales that were designed to capture three
theory of self-regulated learning and provide information as to theoretical types of learning strategies (Duncan & McKeachie, 2005).
whether the MSLQ is likely to be a valuable tool in future examinations First, cognitive strategies are assessed with four subscales that
of the context-specicity of learning processes. Furthermore, testing measure rehearsal (repeating information to oneself), elaboration
the validity of the assumption that scores on the MSLQ normatively (summarizing information), organization (organizing new informa-
predicts academic performance (at a static level) is important for tion using tables and outlines), and critical thinking (evaluating new
educational practitioners and college counselors who may use the ideas and applying them to novel situations). Metacognitive strategies
MSLQ for the early identication of at-risk students. are assessed by a single subscale that attempts to capture students'
To this aim, the current paper has three primary purposes. The rst ability to monitor their own mental processes and adjust them when
is to meta-analyze the many studies that have investigated the needed an important component of self-regulatory frameworks of
predictive utility of the MSLQ subscales for academic performance. learning (e.g., Pintrich, 2000). Lastly, resource management is
Findings regarding the validity of the MSLQ scales vary widely, with assessed with four subscales that attempt to capture students' ability
some authors nding impressive relationships between academic to manage the resources available to them. These include time and
achievement and scores on the MSLQ (e.g., Bell, 2006; Bembenutty, study environment regulation (study skills and appropriate use of
2007; Hsu, 1997; Langley, 2007) and others nding much weaker study time), effort regulation (persisting in studying even when
relationships (e.g., Barker, 1997; Bartels & Magun-Jackson, 2009; material is boring), peer learning (working with others and using
Huang, 2008; Lewis, 2006). The meta-analysis will yield a more study groups), and help seeking (asking for assistance from peers or
comprehensive and extensive picture of the subscales' validity for instructors). An overview of the structure of the MSLQ, along with
normative, static use and for testing more complex theoretical example items is provided in Table 1. Readers interested in a more
relationships regarding the changeability and context-specicity of detailed account of the development of the MSLQ or the items for
students' motivations and learning strategies. The second purpose of each of these subscales are referred to Pintrich et al. (1991), Pintrich,
the current paper is to nd preliminary evidence for the notion that Smith, Garcia, and McKeachie (1993), and Duncan and McKeachie
the MSLQ subscales tap context-specic student traits and behaviors (2005).
M. Cred, L.A. Phillips / Learning and Individual Differences 21 (2011) 337346 339

Table 1
Structure and example items for the MSLQ.

Section Sub-section Scales # Items Example item

Motivational scales Value components Intrinsic goal orientation 4 In a class like this, I prefer course material that really
challenges me so that I can learn new things.
Extrinsic goal orientation 4 Getting a good grade in this class is the most satisfying thing
for me right now.
Task value 6 It is important for me to learn the course material in this class.
Expectancy component Control beliefs 4 If I study in the appropriate ways, then I will be able to learn
the material in this course.
Self-efcacy for learning and 8 I believe I will receive an excellent grade in this class.
performance
Affective component Test anxiety 5 When I take tests I think of the consequences of failing.
Learning strategies Cognitive and meta-cognitive Rehearsal 4 When I study for this class, I practice saying the material to
scales strategies myself over and over.
Elaboration 6 When I study for this class, I pull together information from
different sources, such as lectures, readings, and discussions.
Organization 4 When I study the readings for this course, I outline the
material to help me organize my thoughts.
Critical thinking 5 I often nd myself questioning things I hear or read in this
course to decide if I nd them convincing.
Meta-cognitive self-regulation 12 When reading for this course, I make up questions to help
focus my reading.
Resource management Time and study environment 8 I usually study in a place when I can concentrate on my course
strategies work.
Effort regulation 4 I work hard to do well in this class even if I don't like what we
are doing.
Peer learning 3 When studying for this course, I often try to explain the
material to a classmate or a friend.
Help seeking 4 I ask the instructor to clarify concept I don't understand well.

3. Methodology 3.2. Inclusion and exclusion criteria

3.1. Search strategy Studies were included if they were described in English and
reported zero-order correlations between MSLQ subscales and either
We identied potential data sources via keyword searches of the GPA or grades in an individual class for college students, or reported
PsychINFO, Eric, and Dissertation Abstracts databases as well as data in a manner that allowed such correlations to be calculated (e.g.,
internet searches and examinations of the reference lists of identied MSLQ scores between high and low performing students). Studies
articles. Dissertations judged to be likely to include relevant data were were excluded if they reported non-Pearson correlations or if they
purchased from Proquest. We also attempted to contact authors of selectively reported only statistically signicant correlations (i.e., did
articles that did not present information in a manner that allowed not report any of their non-signicant results), as the inclusion of such
inclusion in this review (criteria described below) in order to obtain studies would have yielded biased meta-analytic estimates of the
the relevant data. Each data source was coded by one of the two relationships among variables. Studies were also excluded if results
authors using standardized coding sheets that cued coders to attend were based on dramatically shortened or reworded versions of the
to specic study characteristics that were identied as important for MSLQ or if the results reported in the studies suggested aws in the
meta-analytic purposes. This information included: correlations manner in which the MSLQ was scored (e.g., reporting negative alpha
between scores on MSLQ subscales and academic performance, reliabilities for some MSLQ subscales, a likely artifact of items that
correlations among MSLQ subscales, reliability information for scores had not been reverse scored). We also screened our nal database
on MSLQ subscales, the design of the study (predictive versus carefully to ensure that data that might have been published in two
concurrent), the measure of academic performance (e.g., grades in a separate venues (e.g., once as a dissertation and then as a journal
single class versus GPA), the type of class in which the information article) was not included twice in our analysis.
was gathered (e.g., Chemistry, English, and Mathematics), the source For studies that utilized predictive designs with more than one
of academic performance ratings (self-ratings or instructor records), assessment of MSLQ subscales (e.g., MSLQ subscales administered at
and a variety of other sample characteristics (e.g., gender, ethnicity, time 1 and time 2 with academic performance gathered at time 2), we
age, and language of administration). This information was subse- coded both the predictive and concurrent correlations (for the
quently entered in a database to facilitate the appropriate sorting of moderator analysis; see Section 3.6) and took the average of the
information by the relationship captured by the correlation, criterion two correlations for the overall analysis.
category, and type of research design.
The coding of studies for meta-analysis has previously been 3.3. Criterion categories
shown to exhibit high inter-rater reliability especially when coding
decisions are easy to make (Whetzel & McDaniel, 1988; Zakzanis, Only two broad criteria were considered in this meta-analysis. The
1998), but a random sample of these articles was nevertheless coded MSLQ was specically designed to assess motivation and learning
by both authors to allow for an examination of coding accuracy. strategies in a particular class and we therefore examined the
Agreement in the coding of relevant study characteristics was almost relationship between scores on MSLQ subscales and performance in
unanimous; the few disagreements resulted from articles providing an individual class as our primary analysis. The MSLQ is founded on
discrepant data between tables and text (e.g., sample size in cor- the assumption (Duncan & McKeachie, 2005) that motivational
relation table not corresponding with sample size given in descrip- variables and learning strategies are changeable across tasks (e.g.,
tion of methodology). All such disagreements were resolved via Ilies & Judge, 2005), an assumption shown to be valid by Rotgans and
discussion. Schmidt (2009) who found signicant within-person variation in
340 M. Cred, L.A. Phillips / Learning and Individual Differences 21 (2011) 337346

MSLQ scores across three subjects (English, Math, and Science). At the Table 2
same time these constructs also exhibit some stability across situa- Reliability artifact distributions for MSLQ subscales, college GPA, and grades.

tions (e.g., classes) for the same individual (e.g., Bong, 2001, 2004; M rxx SD rxx krel
Gottfried, 1985; Vermeeten, Lodewijks, & Vermunt, 1997; Warr &
MSLQ subscales
Downing, 2000), and so a number of authors have either used the Intrinsic Goal Orientation .69 .05 21
MSLQ to measure general tendencies to use certain learning strategies Extrinsic goal orientation .66 .10 16
(e.g., Wolters, 2003) or have presented correlations between MSLQ Task value .87 .04 17
Control of learning beliefs .65 .11 14
scores and performance across multiple classes (i.e., GPA). We
Self-efcacy .91 .02 21
therefore also present analyses for the relationship between scores Test anxiety .77 .06 12
on MSLQ subscales and college GPA. Comparing predictive utility of Rehearsal .68 .05 14
the MSLQ subscales for GPA versus course-specic grades allows for Elaboration .76 .05 19
initial investigation into the theoretical claim that motivation and Organization .70 .07 14
Critical thinking .77 .04 16
learning strategies may be context-specic.
Metacognitive self-regulation .77 .06 29
Time and study environment .72 .07 15
3.4. Final databases Effort regulation .61 .10 14
Peer learning .68 .10 12
Help seeking .59 .12 13
In total, the database consisted of 2158 correlations from 67
Criterion
independent samples (from a total of 59 articles) representing 19,900 GPA .80 .07 4
independent college students. The database for the relationships Grades .74 .08 5
between MSLQ subscales and GPA consisted of 98 correlations from Note: MSLQ = Motivated Strategies for Learning Questionnaire; M rxx = mean of
24 independent samples representing 9696 independent college reliability distribution; SD rxx = standard deviation of reliability distribution; and krel =
students. The database for the relationships between MSLQ subscales number of independent reliability coefcients on which distributions are based.
and grades in individual classes consisted of 431 correlations from 53
independent samples representing 15,321 independent college
students. The database for intercorrelations among MSLQ subscales The Hunter and Schmidt method provides three primary pieces of
consisted of 1629 correlations from 35 independent samples repre- information: robs, the mean sample-size weighted observed correla-
senting 11,507 independent college students. tion; , the sample-size weighted estimate of the true score
correlation after correcting for study artifacts; and SD, the standard
3.5. Analytic strategy deviation of true score correlations. SD is, in turn, used as evidence
for the presence of moderators of the relationship, inasmuch as it
We used the Hunter and Schmidt (2004) interactive random- represents the amount of variability across studies that remains after
effects meta-analytic method to aggregate the ndings of the accounting for study artifacts. Large SD values are interpreted as
examined studies. Meta-analytic methods that combine study-level evidence for the likely presence of moderators. SD values are also
effects are preferable to a simple pooling of data since this avoids the used for the construction of credibility intervals around the estimate
potential for introducing Simpson's paradox (i.e., reversing the of the true score correlation. Credibility intervals represent the range
direction of effects observed at a subgroup level when aggregating of plausible values for the true score correlation across situations.
data across the subgroups) into research ndings. The Hunter and Wide credibility intervals suggest the presence of moderators. The
Schmidt method is preferable to alternative meta-analytic methods Schmidt and Le (2004) software (version 1.1) used for this analysis
(e.g., Rosenthal & Rubin, 1982, method, or Hedges, 1982, method), for also provides an estimate of the proportion of the observed variance
two primary reasons. First, it provides an estimate of the relationship that can be accounted for by examined study artifacts. Hunter and
that would be observed between variables in the absence of study Schmidt (2004) argue that moderators are unlikely to exist if the
artifacts such as unreliability in measurement of the dependent and proportion of variance that is accounted for by examined artifacts
independent variables, range restriction, or articial dichotomization. exceeds 75% since most meta-analyses cannot account for all likely
Second, it provides information as to the presence of possible study artifacts. This meta-analysis, for example, was not able to
unexamined moderators of the relationship by providing an estimate account for likely range restriction artifacts. Finally, we also reported
of the variability in observed effect sizes after accounting for the operational validities that represent an estimate of the relationship
variability that is due to sampling error and other study artifacts that would be observed between the scores on a scale and a
(e.g., differential reliability of measurement). theoretical criterion measured with perfect reliability. That is, we
The interactive meta-analytic method is appropriate when only corrected correlations for unreliability in the criterion but not for
some studies report artifact information. Specically, it involves using unreliability in the predictor since educational practitioners would
the reported artifact information to construct artifact distributions have to use the MSLQ as it currently stands and would not be able to
that are then used to correct the distribution of observed effect sizes assess the MSLQ constructs with perfect reliability.
for measurement artifacts. For this meta-analysis we constructed Lastly, the meta-analytic intercorrelation matrix of MSLQ subscales
artifact distributions for unreliability in the measurement of each of was used to examine the structure of the MSLQ in more detail.
the MSLQ subscales and for unreliability in the measurement of the Specically, the full meta-analytic inter-correlation matrix was factor
academic performance criterion. Given that very few studies in the analyzed (principal axis factoring and oblique rotation) using
educational literature have ever reported reliability information for exploratory factor analysis (EFA) based on a harmonic mean sample
GPA and none (to our knowledge) have reported reliability informa- size of 4230.
tion for individual grades, we imported the four published reliability
estimates known to us for GPA (Barritt, 1966; Bendig, 1953; Reilly & 3.6. Moderators
Warech, 1993; Stricker, Rock, Burton, Muraki, & Jirele, 1994), while
estimates of grade reliability were taken from ve large undergrad- Only two possible moderators, of the relationship between MSLQ
uate classes recently taught by the rst author. Summary information subscales and academic performance were examined; both integral to
for the various reliability distributions is provided in Table 2. Mean a better understanding of self-regulated learning processes. First, we
reliabilities for scores on the MLSQ subscales range from a low of examined the difference between studies that used predictive designs
= .59 (help seeking) to a high of = .91 (self-efcacy). (i.e., MSLQ subscales gathered at time 1 and academic performance at
M. Cred, L.A. Phillips / Learning and Individual Differences 21 (2011) 337346 341

time 2) versus concurrent designs. Given that students are likely to Australia Thailand, Taiwan, India, and Malaysia (87.3% of the studies
base their responses to MSLQ subscales in part on their known were conducted in North America) and a wide variety of class types:
academic performance, we expected correlations to be higher for economics, science, mathematics, medicine and English courses. A
concurrent designs than for predictive designs. That is, a student who sample-size weighted review of available study characteristics
has done poorly in a class is more likely to attribute his/her poor indicated that 37.1% of participants were male, 67.6% were Caucasian,
performance to factors such as poor strategies or low motivation 17.8% were Asian, 10.5% were African-American, and 6.1% were
rather than low capability (Schunk, 2005a) particularly when that Hispanic. The sample-size weighted mean age participants was
student has good self-regulation skills. The resultant higher concor- 23.1 years.
dance between achieved grades and self-rated motivation and
strategies will increase correlations in concurrent designs above 4.2. Criterion correlations
those observed in predictive designs. Second, we examined the
difference in correlations for studies that used GPA as the measure of Meta-analytic results for the relationships between MSLQ sub-
academic performance and studies that used the grade in a specic scales and both GPA and grades in individual classes are presented in
class as the measure of academic performance. Since the MSLQ is Table 3. As hypothesized, the relationships between scores on MSLQ
phrased to assess motivation and strategies for a specic class and subscales and grades in individual classes were higher than those
since a student's motivation and strategies are likely to be somewhat observed between scores on MSLQ subscales and GPA for all fteen
different for different classes, we expected correlations to be higher subscales (the relationship between peer learning and GPA was not in
for studies that used the grade in an individual class as a measure of the expected direction). The relationships between MSLQ scores and
academic performance. academic performance were generally weak to moderate (Cohen,
1988). The highest observed validities for grades in individual classes
3.7. Critique of meta-analysis were found for effort regulation ( = .40, k = 24, N = 5180), self-
efcacy ( = .37, k = 39, N = 8123), and time and study environment
Criticisms of meta-analytic methods have typically focused on the ( = .31, k = 24, N = 4892). Similarly, the highest observed validities
subjective nature of the process whereby studies are selected for for GPA were found for effort regulation ( = .23, k = 5, N = 2721),
inclusion in the meta-analysis, the possibility that meta-analytic time and study environment ( = .23, k = 5, N = 2721), and meta-
ndings quickly become outdated as new studies are published (e.g., cognitive self-regulation ( = .22, k = 13, N = 4390). Especially
Gay, Mills, & Airasian, 2009), and the possibility of drawing inappro- notable were the ndings that specic learning approaches (e.g.,
priate inferences due to Simpson's Paradox and the Ecological Fallacy rehearsal, elaboration, organization, and peer learning) were largely
(e.g., Cooper & Pattall, 2009). Although these issues do pose potential unrelated to academic performance while the theoretically more
threats to the validity of meta-analytic ndings, the methods employed stable (i.e., less contextual) abilities and tendencies such as meta-
in this study have attempted to minimize these threats. Subjectivity in cognitive self-regulation and effort regulation were most strongly
study selection was largely avoided by including all available sources of related to academic performance.
data, even going so far as to purchase numerous dissertations in order to The size of SD estimates (and hence the width of the credibility
minimize the le-drawer effect, and only screening out studies with intervals) was somewhat variable but large enough to suggest the
obvious data-analytic errors (e.g., reporting of negative alpha re- possible presence of moderators for at least some MSLQ subscales
liabilities). The Ecological Fallacy would occur if we (or authors of the (e.g., test anxiety and effort regulation). The type of class may
summarized articles) had inappropriately used group level character- inuence the degree to which constructs assessed by the MSLQ
istics to draw inferences about individuals. However, all of the examined inuence academic performance in that class. For example, classes in
data sources were based on students' individual grades and students' which grades are based less on tests taken under exam conditions and
individual scores on the MSLQ and the Ecological Fallacy therefore does more on take-home assignments and essays may not exhibit a strong
not represent a likely threat. Simpson's Paradox is perhaps more likely relationship between test anxiety and academic performance.
since it is plausible that gradeMSLQ relationships are substantially
moderated by unexamined group variables, but the Hunter and Schmidt 4.3. Prospective and concurrent designs
meta-analytic method precludes Simpson's paradox from being
introduced since it involves the computation of sample-size weighted Results for the moderation analysis examining the differences in
average correlations rather than the pooling of data across samples relationships between scores on MSLQ subscales and grades in indi-
followed by the computation of a single correlation. It is however vidual classes for studies using prospective versus concurrent designs
possible that Simpson's paradox is present at the level of the individual are presented in Table 4. Contrary to expectations, the correlations for
studies included in our review. That is, unexamined group moderators concurrent study designs were not uniformly higher than those for
of MSLQgrade relationships may exist such that an examination of data prospective study designs; concurrent designs exhibited higher
at the overall sample level suggests relationships in opposite direction to correlations for only six of the fteen subscales. Interestingly, all six
those that would be found if the data was examined at the appropriate of these subscales were motivational subscales with the largest
sub-group level. A number of authors (e.g., Bembenutty, 2007) had observed difference observed for test anxiety ( = .42 for concurrent
examined the effect of group moderators such as gender and ethnicity designs versus = .14 for prospective designs), self-efcacy ( = .58
on the validity of MSLQ scores (i.e., MSLQgrade relationships) but versus = .31) and task value ( = .42 versus = .19).
found no evidence for strong effects. These examinations, together with
the relatively high level of congruence between our ndings and the 4.4. Intercorrelations
theoretically expected effects suggest that our ndings are relatively
unlikely to be characterized by Simpson's paradox. Meta-analytic estimates of the relationships among the fteen
subscales are presented in Table 5. Most disattenuated correlations
4. Results (i.e., ) are of a medium to large size (Cohen, 1988), and the pattern of
intercorrelations was generally supportive of the assumption that
4.1. Sample characteristics cognitions and motivations inuence learning strategies. That is,
cognitions and motivations exhibited moderate to large correlations
Data was drawn from a wide variety of college populations. with the various learning strategies. It is important to note that some
Samples were drawn from 7 countries: the United States, Canada, of the meta-analytic intercorrelations suggested non-trivial construct
342 M. Cred, L.A. Phillips / Learning and Individual Differences 21 (2011) 337346

Table 3
Meta-analytic results for the MSLQ for college GPA and current-semester grades.

Subscale Criterion N k robs SDobs SD Lower 90% Upper 90% % Var rop

Intrinsic goal orientation GPA 4702 10 .11 .08 .15 .11 .00 .29 23 .12
Grade 10,325 40 .14 .09 .20 .12 .05 .35 35 .17
Extrinsic goal orientation GPA 4420 8 .04 .12 .06 .16 .15 .27 12 .05
Grade 9433 33 .07 .11 .10 .16 .10 .31 22 .08
Task value GPA 2508 6 .12 .07 .14 .08 .04 .24 36 .13
Grade 7658 34 .17 .11 .21 .14 .03 .39 26 .20
Control of learning beliefs GPA 2354 5 .12 .05 .17 .07 .09 .26 49 .14
Grade 6117 30 .14 .02 .20 .03 .16 .24 92 .16
Self-efcacy GPA 3798 9 .18 .06 .21 .07 .12 .01 39 .20
Grade 8123 39 .30 .15 .37 .18 .15 .60 16 .35
Test anxiety GPA 1923 6 .09 .15 .11 .19 .36 .13 12 .10
Grade 6993 30 .14 .14 .18 .19 .42 .06 17 .16
Rehearsal GPA 2461 6 .05 .06 .07 .08 .03 .18 40 .06
Grade 5958 24 .08 .09 .12 .13 .05 .29 31 .10
Elaboration GPA 3735 8 .10 .02 .13 .03 .10 .17 81 .12
Grade 6774 28 .13 .12 .17 .17 .04 .39 21 .15
Organization GPA 2721 5 .07 .03 .09 .04 .04 .14 71 .08
Grade 5291 24 .08 .08 .11 .10 .03 .24 44 .09
Critical thinking GPA 1528 4 .06 .00 .08 .00 .08 .08 100 .07
Grade 5245 26 .09 .14 .12 .19 .12 .35 20 .10
Meta-cognitive Self-regulation GPA 4390 13 .17 .16 .22 .20 .03 .47 11 .19
Grade 7155 32 .18 .13 .23 .17 .02 .45 21 .20
Time and study environment GPA 2721 5 .17 .08 .23 .11 .09 .36 22 .19
Grade 4892 24 .22 .10 .31 .13 .14 .48 33 .26
Effort regulation GPA 2721 5 .16 .11 .23 .15 .04 .42 15 .18
Grade 5180 24 .27 .13 .40 .19 .16 .64 23 .31
Peer learning GPA 1528 4 .06 .00 .08 .00 .08 .08 100 .07
Grade 4347 20 .05 .10 .08 .13 .10 .25 34 .06
Help seeking GPA 1528 4 .01 .00 .02 .00 .02 .02 100 .02
Grade 4726 23 .04 .06 .06 .09 .05 .17 61 .04

Note. MSLQ = Motivated Strategies for Learning Questionnaire; N = number of participants; k = number of studies; robs = sample size weighted mean observed correlation; SDobs =
observed standard deviation; = true score correlation; SD = standard deviation of true score correlation; Lower 90% = lower bound of 90% credibility interval based on operational
validity; Upper 90% = upper bound of 90% credibility interval based on operational validity; % Var = the percentage of variance in the observed correlation accounted for by study
artifacts; rop = operational validity.

overlap. Specically, disattenuated correlations close to one were the item in one way while average students are likely to respond to the
observed for the relationships between time and study environment item in a different way. The item, I ask the instructor to clarify concepts I
and effort regulation ( = .95, k = 16, N = 4833), and peer learning don't understand well, would probably elicit a disagreement from high
and help seeking ( = .92, k = 14, N = 4583), while a number of other performing students who do not require help, as well as from low
disattenuated correlations were also high enough to suggest some performing students who might not be motivated enough to ask for help
non-trivial level of construct redundancy (e.g., the relationship or lack the ability to recognize that they are struggling; average students
between elaboration and organization: = .82, k = 23, N = 6533). who encounter occasional academic difculties but recognize the need to
The full matrix of disattenuated correlations was used to conduct ask for help may agree with the statement. Ideal-point items can be
an EFA of the structure of MSLQ scales. Parallel analysis (Hayton, desirable when attempting to measure intermediate levels of a construct
Allen, & Scarpello, 2004) for the EFA suggested a four-factor solution, with high delity but require different scoring methods (Drasgow,
and an examination of factor loadings (see Table 6) suggests that the Chernyshenko, & Stark, 2010), and do not lend themselves to the type of
rst and second factors are composed of the learning strategies, the analyses typically conducted with MSLQ scores (e.g., factor analyses and
third factor is composed of motivational factors, and the fourth factor correlations with grade outcomes).
largely reects test anxiety. This structure is largely supportive of the Of the eighty-one examined items both raters independently
theoretical distinction between motivations and cognitions (factors 3 identied ten items characterized by conditional content and eight
and 4) on the one hand and learning strategies (factors 1 and 2) on the items with ideal-point characteristics. These potentially problematic
other hand. items disproportionately represent the peer learning and help seeking
constructs. All three of the peer-learning items were classied as
4.5. Review of items either having conditional item content (once) or having likely ideal-
point characteristics (twice). Similarly, three out of the four items
Given the relatively low observed validities for many of the individual making up the help-seeking scale were classied as having both
MSLQ subscales, we also conducted a review of the individual items that conditional item content and being likely ideal-point items. These
comprise the MSLQ. Specically, we asked two independent raters with possible psychometric issues may help to explain some of the low
PhD-level training in psychometrics to evaluate the eighty-one individ- validities reported in this meta-analytic review.
ual items listed in Duncan and McKeachie (2005) for two undesirable
item characteristics: rst, conditional-content items, and second, ideal- 5. Discussion
point items. Items with conditional content assume that a particular
event has taken place and present a second statement with which the Our ndings have important implications for not only research
respondent must agree or disagree (e.g., Whenever X occurs, I do Y). relying on the MSLQ but also on the broader self-regulated learning
Such items are psychometrically problematic because disagreement with literature and the even more general literature regarding the deter-
such items could imply that X has never occurred for the respondent or minants of and inuences on college academic performance. Our
that the respondent does not do Y when X occurs. Ideal-point items are meta-analytic review suggests that the MSLQ is a reasonably reliable
items where both high-achieving and low-achieving students respond to measure of constructs, some of which exhibit meaningful
M. Cred, L.A. Phillips / Learning and Individual Differences 21 (2011) 337346 343

Table 4
Meta-analytic comparison of MSLQ validities from prospective and concurrent research designs.

Subscale Study design N k robs SDobs SD Lower 90% Upper 90% % Var explained

Intrinsic goal orientation Prospective 8722 31 .13 .06 .19 .09 .07 .30 46
Concurrent 1086 5 .28 .13 .39 .18 .17 .62 21
Extrinsic goal orientation Prospective 8219 27 .06 .12 .08 .16 .13 .29 20
Concurrent 885 4 .13 .05 .18 .07 .09 .27 64
Task value Prospective 6289 28 .15 .09 .19 .11 .05 .34 35
Concurrent 885 4 .33 .11 .42 .13 .25 .58 26
Control of learning beliefs Prospective 4748 24 .12 .05 .17 .07 .08 .26 68
Concurrent 1115 5 .17 .00 .24 .00 .24 .24 100
Self-efcacy Prospective 6166 29 .26 .15 .31 .18 .08 .54 17
Concurrent 1172 6 .47 .00 .58 .00 .58 .58 100
Test anxiety Prospective 5793 24 .11 .13 .14 .17 .37 .08 19
Concurrent 716 3 .32 .01 .42 .00 .42 .42 100
Rehearsal Prospective 4123 19 .10 .09 .14 .13 .03 .30 35
Concurrent 1462 3 .06 .06 .08 .08 .02 .19 38
Elaboration Prospective 5137 22 .13 .12 .17 .15 .03 .37 24
Concurrent 1516 4 .10 .15 .14 .19 .11 .38 11
Organization Prospective 3654 18 .07 .08 .10 .11 .04 .23 46
Concurrent 1516 4 .04 .02 .06 .03 .02 .09 88
Critical thinking Prospective 3551 19 .10 .14 .13 .18 .10 .36 22
Concurrent 1115 4 .06 .12 .08 .16 .13 .29 19
Meta-cognitive self-regulation Prospective 5223 23 .18 .12 .24 .16 .04 .44 23
Concurrent 2041 8 .09 .16 .12 .20 .14 .38 14
Time and study environment Prospective 2919 15 .23 .06 .31 .07 .22 .41 62
Concurrent 1526 6 .22 .18 .30 .24 .00 .61 11
Effort regulation Prospective 3112 16 .28 .07 .41 .10 .29 .54 54
Concurrent 1719 6 .27 .22 .40 .32 .00 .80 7
Peer learning Prospective 2760 15 .06 .11 .09 .15 .10 .28 33
Concurrent 1004 2 .00 .00 .00 .00 .00 .00 100
Help seeking Prospective 2822 15 .05 .04 .08 .06 0 .16 77
Concurrent 1227 4 .04 .00 .05 .00 .05 .05 100

Note. MSLQ = Motivated Strategies for Learning Questionnaire; N = number of participants; k = number of studies; robs = sample size weighted mean observed correlation; SDobs =
observed standard deviation; = true score correlation; SD = standard deviation of true score correlation; lower 90% = Lower bound of 90% credibility interval based on operational
validity; Upper 90% = upper bound of 90% credibility interval based on operational validity;% Variance = the percentage of variance in the observed correlation accounted for by
measurement error.

relationships with college academic performance. Moderate to strong learning (e.g., Zimmerman, 1990). That is, students who are able to
relationships (Cohen, 1988) were observed between class grades and engage in metacognitive self-monitoring and effort regulation, who
self-efcacy, effort regulation, and time and study environment view the academic task as having intrinsic interest and value, have
subscales, although most other relationships between grades and high levels of self-efcacy, and who use appropriate learning stra-
MSLQ constructs were weaker. Encouragingly, the strongest of these tegies are shown to have higher average grades than students without
relationships (e.g., = .41 for effort regulation) are similar to those these attributes and behaviors. Further, the pattern of relationships
observed in meta-analytic reviews of some of the best known among the constructs assessed by the MSLQ supports the notion
traditional predictors of academic performance: scores on admission that the effect of motivations on academic performance is mediated
tests, prior academic performance, and study skills and study habits by learning strategies in the manner suggested by the authors of
(Cred & Kuncel, 2008; Hezlett et al., 2001). the MLSQ (e.g. Duncan & McKeachie, 2005). That is, motivational
More importantly, our results are broadly supportive of some of variables assessed by the MSLQ are related to learning strategies,
the basic assumptions that underpin theories of self-regulated many of which are, in turn, related to academic performance.

Table 5
Meta-analytic results of intercorrelations among MSLQ subscales.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1 Intrinsic goal orientation .38 .78 .57 .65 .09 .35 .63 .43 .62 .64 .49 .53 .28 .22
2 Extrinsic goal orientation .25 .41 .39 .36 .27 .40 .28 .32 .23 .36 .27 .33 .25 .22
3 Task value .60 .31 .54 .54 .01 .33 .56 .40 .38 .52 .47 .56 .18 .21
4 Control of learning beliefs .38 .25 .40 .68 .03 .16 .38 .20 .19 .34 .25 .32 .02 .00
5 Self-efcacy .51 .28 .48 .52 .35 .21 .46 .30 .33 .47 .43 .54 .08 .08
6 Test anxiety .06 .19 .01 .02 .29 .19 .02 .07 .03 .09 .16 .22 .15 .09
7 Rehearsal .24 .26 .25 .11 .17 .14 .62 .80 .39 .66 .60 .51 .38 .37
8 Elaboration .46 .19 .45 .27 .38 .01 .44 .80 .72 .82 .66 .63 .40 .44
9 Organization .30 .22 .31 .13 .24 .05 .55 .58 .47 .78 .68 .55 .44 .39
10 Critical thinking .45 .16 .31 .13 .28 .02 .28 .55 .34 .69 .33 .29 .44 .35
11 Metacognitive self-regulation .47 .26 .43 .24 .40 .07 .48 .63 .57 .53 .74 .76 .44 .45
12 Time and study environment .35 .18 .37 .17 .35 .12 .42 .49 .48 .25 .55 .95 .20 .31
13 Effort regulation .34 .20 .41 .20 .40 .14 .33 .43 .35 .20 .52 .63 .11 .28
14 Peer learning .19 .17 .14 .02 .06 .10 .26 .29 .30 .32 .31 .14 .07 .92
15 Help seeking .14 .13 .15 .00 .06 .06 .23 .29 .25 .24 .30 .20 .17 .58

Note. MSLQ = Motivated Strategies for Learning Questionnaire. Lower diagonal entries are sample-size weighted mean observed correlations, and upper diagonal entries (in italics)
are true score correlations. The number of studies represented by each correlation ranges from 11 to 24 and the sample sizes for each cell range from 2752 to 7814. More detailed
meta-analytic results are available from the rst-author upon request.
344 M. Cred, L.A. Phillips / Learning and Individual Differences 21 (2011) 337346

Table 6 on multiple choice examinations may require memorization but not


Factor loadings for exploratory factor analysis of MSLQ subscales. require much critical thinking and students who engage in critical
MSLQ scales Factor loadings thinking may therefore not necessarily score more highly on tests. This
would, in turn, result in trivial correlations between academic
Factor 1 Factor 2 Factor 3 Factor 4
performance in the class and critical thinking behaviors. Future research
1 Intrinsic goal orientation 0.07 0.15 0.65 0.04
on the predictive utility of scores on the MSLQ subscales with different
2 Extrinsic goal orientation 0.08 0.03 0.35 0.27
3 Task value 0.15 0.04 0.63 0.04 types of performance tasks is warranted.
4 Control of learning beliefs 0.08 0.10 0.68 0.04 A review of the items that comprise the MSLQ also suggests that low
5 Self-efcacy 0.05 0.00 0.72 0.27 correlations between the scales and academic performance may be due
6 Test anxiety 0.04 0.01 0.03 0.72 to poorly constructed items. Specically, we identied a number of items
7 Rehearsal 0.63 0.06 0.06 0.25
8 Elaboration 0.52 0.21 0.22 0.00
with what we have termed conditional item content. That is, the item
9 Organization 0.67 0.12 0.04 0.13 assumes that an event has occurred, presents the respondent with a
10 Critical thinking 0.18 0.35 0.25 0.00 response to the events, and asks the respondent to indicate whether
11 Metacognitive self-regulation 0.61 0.20 0.17 0.06 he/she has engaged in this response. For example, one item asks
12 Time and study environment 0.79 0.08 0.01 0.14
respondents to indicate a level of endorsement for the statement: When
13 Effort regulation 0.65 0.12 0.12 0.17
14 Peer learning 0.11 0.88 0.01 0.01 I can't understand the material in this course, I ask another student in this
15 Help seeking 0.06 0.62 0.06 0.03 class for help. This item attempts to assess help-seeking behavior, but it
Note. Loadings are from pattern matrix based on EFA with principal axis factoring and
is not clear if a disagreement with the item indicates that the respondent
oblimin rotation. would not seek help or has never not understood the material.
It is also important to note that the MSLQ appears to be charac-
terized by a non-trivial amount of redundancy at the measurement
Our factor analytic ndings are also broadly supportive of the theory level. Scores on at least two pairs of subscales (time and study envi-
of self-regulated learning (Zimmerman, 1990; Pintrich, 1999) inasmuch ronment and effort regulation; peer learning and help seeking) are so
as they show a clear distinction between motivational and affective strongly related with each other ( = .92 and = .95, respectively) as
processes on one hand (factor 3 and factor 4, respectively) and learning to suggest that the same construct is being assessed. A rewriting of
approaches (factor 1 and factor 2) on the other hand. Our ndings are items to better distinguish between these pairs of constructs or a
also supportive of socialcognitive learning theories inasmuch as they combination of these subscales may well be warranted.
support the contextual nature of self-regulated learning processes (e.g., Despite the identied psychometric problems and possible construct
Schunk, 2005b). That is, that motivations and learning strategies are redundancy of the MSLQ, the main ndings of our review suggest that the
class-specic and likely to vary across classes for the same individual; MSLQ is likely to be a valuable tool in both research and practice settings.
scores on the MSLQ were consistently more strongly related to The MSLQ appears to capture many of the most important constructs that
performance in a specic class than to students' average performance are central to self-regulated learning and should therefore be valuable for
across multiple classes. We hope that these results will encourage future investigations of self-regulated learning in general and the
researchers to further pursue investigations into the determinants of contextual nature of self-regulated learning processes in particular
within-person variations in academic performance. We are aware of (e.g., Rotgans & Schmidt, 2009). At a practical level, the MSLQ is likely to
only one study in the MSLQ literature (Rotgans & Schmidt, 2009) that be useful for counselors attempting to understand the learning
has formally examined the issue of within-person variability in MSLQ approaches of individual students and thereby target interventions to
scores and the ability of this variation to account for within-person potentially problematic approaches (e.g., an overreliance on rehearsal
variation in grades. This relative dearth of research on the determinants strategies). Some researchers (e.g., Wolters, 2003) have even used the
of within-person variation in learning and academic achievement is, of MSLQ to identify general approaches to learning and academic tasks (as
course, not limited to the MSLQ literature but is a more general feature opposed to the specic case of a single class), suggesting that the MSLQ
of research on academic performance in college settings. could be used by researchers and instructors to make students more
One set of ndings of our meta-analytic review that is less supportive aware of their general approaches to learning and use it to engage
of the general validity of the MSLQ and (to a lesser degree) the theory of students in a discussion of the efcacy of different learning strategies
self-regulated learning is that many of the specic learning strategies
(i.e., rehearsal, elaboration, organization, critical thinking, peer learning, 6. Limitations and future research
and help seeking) appeared to be largely unrelated to academic
performance. Two possible reasons for these low relationships present The MSLQ is explicitly based upon the notion that students are
themselves. First, we suspect that some of the learning strategies, differently motivated and use different learning strategies for different
particularly peer learning and help-seeking may exhibit curvilinear courses and different academic tasks (Duncan & McKeachie, 2005), but
relationships with grades that are not captured by correlation co- we are not aware of any explicit examination of the validity of this
efcients. That is, neither very high performing nor very low performing assumption beyond the recent paper by Rotgans and Schmidt (2009).
students are likely to engage in peer learning or help seeking high Capturing such within-person variation in motivations, behaviors, and
performing students because they have no need for help and low learning strategies should substantially increase our ability to explain
performing students because they do not know where to nd help or are variance in academic performance above and beyond the variation
insufciently motivated to do so. Rehearsal is another strategy that may explained by more stable student characteristics and behaviors, and also
not be related to academic performance in a linear manner. Rehearsal, as help researchers to better understand the act of learning (Schunk,
measured by the MSLQ, reects rote memorization (2 items) and 2005a). We therefore encourage future research to focus on the degree
repetition of material (2 items) two behaviors that are unlikely to be of variability in motivations and learning strategies (i.e., MSLQ scores)
practiced by either high achieving students, who likely have no need for across academic tasks and courses. Similarly, we encourage further
it, or low achieving students, who may be too unmotivated to engage in investigations into whether specic course characteristics may moder-
these behaviors or do not realize that they need to engage in these ate the relationships of MLSQ constructs with academic performance.
behaviors. The second possible reason for the low relationships between Rehearsal and elaboration strategies may, for example, be more
some learning strategies and academic performance is that the manner important in classes that rely on multiple-choice tests for student
in which grades are determined in college often does not require assessment while critical thinking and help-seeking behaviors may be
students to use effective learning strategies. For example, performance more important for courses that are difcult or theoretical in content.
M. Cred, L.A. Phillips / Learning and Individual Differences 21 (2011) 337346 345

The observed variation in effect sizes are certainly large enough to Crede, M., Roch, S. G., & Kieszczynka, U. M. (2010). Class attendance in college: A meta-
analytic review of the relationship of class attendance with grades and student
suggest the presence of some unexamined moderator variables. characteristics. Review of Educational Research, 80, 272295.
We also encourage further research on how and when the motiva- Drasgow, F., Chernyshenko, O. S., & Stark, S. (2010). 75 years after Likert: Thurstone was right.
tions and learning strategies assessed by the MSLQ are developed. Some Industrial and Organizational Psychology: Perspectives on Science and Practice (pp. 3).
Duncan, T. G., & McKeachie, W. C. (2005). The making of the Motivated Strategies for
encouraging work has already been done that illustrates that learning Learning Questionnaire. Educational Psychologist, 40, 117128.
strategies and motivations can change in response to direct interven- Dweck, C. S., & Leggett, E. L. (1988). A socialcognitive approach to motivation and
tions (e.g., work on the Learning to Learn courses, Van der Stoep & personality. Psychological Review, 95, 256273.
Gay, L. R., Mills, G., & Airasian, P. W. (2009). Educational research: Competencies for
Pintrich, 2003), but we encourage further examinations of their stability, analysis and interpretation (9th edition). New Jersey: Pearson.
whether they are already xed by the time that students arrive in Gaythwaite, E. S. (2006). Metacognitive self-regulation, self-efcacy for learning and
college, and whether or not they change in response to performance on performance, and critical thinking as predictors of academic success and course
retention among community college students enrolled in online, telecourse, and
individual test or assignments. We also encourage researchers to
traditional public speaking courses. Unpublished doctoral dissertation, University
examine the relationship of constructs assessed by the MSLQ with other of Central Florida.
widely used predictors of academic performance such as scores on Gottfried, A. E. (1985). Academic intrinsic motivation in elementary and junior high
entrance examinations, study behaviors, and stable student attributes school students. Journal of Educational Psychology, 77, 631645.
Halpin, G., Halpin, G., & Schaer, B. B. (1981). Relative effectiveness of the California
such as personality. Some (e.g., Bidjerano & Dai, 2007) have already Achievement Tests in comparison with the ACT Assessment, College Board
begun such examinations, but signicantly more work in this area is Scholastic Aptitude Test, and high school grade point average in predicting college
required to establish the unique contribution of the constructs assessed grade point average. Educational and Psychological Measurement, 41, 821827.
Hayton, J. C., Allen, D. G., & Scarpello, V. (2004). Factor retention decisions in
by the MSLQ over more stable predictors of academic performance. exploratory factor analysis: A tutorial on parallel analysis. Organizational Research
Further, we hope that some of the possible non-linear relationships Methods, 7, 191205.
between learning strategies and academic performance that we outline Hedges, L. V. (1982). Estimation of effect size from a series of independent experiments.
Psychological Bulletin, 92, 490499.
above will be examined. This could easily be done with large archival Hezlett, S.A., Kuncel, N.R., Vey, M.A., Ahart, A., Ones, D.S., Campbell, J.P., et al. (2001,
datasets and would not require the collection of new data. April). The predictive validity of the SAT: A comprehensive meta-analysis. In D.S.
Finally, given that learning occurs for students that are nested Ones & S.A. Hezlett (Chairs), Predicting performance: The interface of I/O
psychology and educational research. Symposia presented at the annual confer-
within classrooms, universities, and countries, future research should ence of the Society for Industrial and Organizational Psychology, San Diego, CA.
examine attempt to examine the size and nature of the classroom, Hsu, J. T.-S. (1997). Value, expectancy, metacognition, resource management, and
university, or country effect on the relationship between MSLQ cons- academic achievement: A structural model of self-regulated learning in a distance
education context. Unpublished doctoral dissertation, University of Southern
tructs and academic performance.
California.
Huang, S. C. (2008). Assessing motivation and learning strategies using the Motivated
Strategies for Learning Questionnaire in a foreign language learning context. Social
7. Conclusion Behavior and Personality, 36, 529534.
Humphreys, L. G. (1979). The construct of general intelligence. Intelligence, 3, 105120.
The MSLQ is based on important theoretical insights into the Hunter, J. E., & Schmidt, F. L. (2004). Methods of meta-analysis: Correcting error and bias
in research ndings (2nd edition). Newbury Park: Sage Publications.
nature of learning and the determinants of academic performance. Ilies, R., & Judge, T. A. (2005). Goal regulation across time: The effects of feedback and
Despite some of the identied psychometric problems, we believe it to affect. The Journal of Applied Psychology, 90, 453467.
be an important tool for educational researchers and practitioners, Kuncel, N. R., Hezlett, S. A., & Ones, D. S. (2001). A comprehensive meta-analysis of the
predictive validity of the Graduate Record Examinations: Implications for graduate
particularly when paired with measures designed to assess stable school selection and performance. Psychological Bulletin, 127, 162181.
student attributes and behaviors. Langley, S. R., II (2007). Identifying self-regulatory factors that inuence the academic
achievement motivation of underprepared college students. Unpublished doctoral
dissertation, University of Minnesota.
References Lewis, J. P. (2006). Effects of self-regulated learning on metacognitive strategies,
academic performance, and transfer of preservice teachers in an educational
Barritt, L. S. (1966). Note: The consistency of rst-semester college grade point technology class. Unpublished doctoral dissertation, University of South Alabama.
average. Journal of Educational Measurement, 3, 261262. Lievens, F., Coetsier, P., De Fruyt, F., & De Maeseneer, J. (2002). Medical students'
Barker, J. R. (1997). An evaluation of the effects of self-regulated learning on USMLE personality characteristics and academic performance: A ve-factor model
Step 1 scores at the University of Mississippi School of Medicine. Unpublished perspective. Medical Education, 36, 10501056.
doctoral dissertation, Vanderbilt University, Tennessee. O'Connor, M. C., & Paunonen, S. V. (2007). Big ve personality predictors of post-secondary
Bartels, J. M., & Magun-Jackson, S. (2009). Approach-avoidance motivation and academic performance. Personality and Individual Differences, 43, 971990.
metacognitive self-regulation: The role of need for achievement and fear of failure. Pintrich, P. R. (1999). The role of motivation in promoting and sustaining self-regulated
Learning and Individual Differences, 19, 459463. learning. Educational Research, 31, 459470.
Bell, P. D. (2006). Can factors related to self-regulated learning and epistemological Pintrich, P. R. (2000). The role of goal orientation in self-regulated learning. In M.
beliefs predict learning achievement in undergraduate asynchronous web-based Boekaerts, P. R. Pintrich, & M. Zeidner (Eds.), Handbook of self-regulation
courses? Perspectives in Health Information Management, 3, 117. (pp. 451502). San Diego, CA: Academic.
Bembenutty, H. (2007). Self-regulation of learning and academic delay of gratication: Pintrich, R. R., & DeGroot, E. V. (1990). Motivational and self-regulated learning
Gender and ethnic differences among college students. Journal of Advanced components of classroom academic performance. Journal of Educational Psychology,
Academics, 18, 586616. 82, 3340.
Bendig, A. W. (1953). The reliability of letter grades. Educational and Psychological Pintrich, P. R., Smith, D. A. F., Garcia, T., & McKeachie, W. J. (1991). A manual for the use of
Measurement, 13, 311321. the Motivated Strategies for Learning Questionnaire (MSLQ). Ann Arbor: Michigan,
Bhattacharyya, S. (2004). Dynamics of motivation and self-regulation among college National Centre for Research to Improve Postsecondary Teaching and Learning.
pre-service teachers: A cross-cultural perspective. Unpublished doctoral disserta- Pintrich, P. R., Smith, D. A. F., Garcia, T., & McKeachie, W. J. (1993). Reliability and
tion, The University of Memphis. predictive validity of the Motivated Strategies for Learning Questionnaire (MSLQ).
Bidjerano, T., & Dai, D. Y. (2007). The relationship between the big-ve model of Educational and Psychological Measurement, 53, 801813.
personality and self-regulated learning strategies. Learning and Individual Differ- Poropat, A. E. (2009). A meta-analysis of the ve factor model of personality and
ences, 17, 6981. academic performance. Psychological Bulletin, 135, 322338.
Bong, M. (2001). Between- and within-domain relations of academic motivation Reilly, R. R., & Warech, M. A. (1993). The validity and fairness of alternatives to cognitive
among middle and high-school students: Self-efcacy, task value, and achievement tests. In L. C. Wing, & B. R. Gifford (Eds.), Policy issues in employment testing
goals. Journal of Educational Psychology, 93, 2334. (pp. 131224). Boston: Kluwer.
Bong, M. (2004). Academic motivation in self-efcacy, task value, achievement goal Robbins, S. B., Lauver, K., Le, H., Davis, D., Langley, R., & Carlstrom, A. (2004). Do
orientations, and attributional beliefs. Journal of Educational Research, 97, 287297. psychosocial and study skill factors predict college outcomes? A meta-analysis.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Psychological Bulletin, 130, 261288.
Lawrence Erlbaum Associates. Rosenthal, R., & Rubin, D. B. (1982). Comparing effect sizes of independent studies.
Cooper, H., & Pattall, E. A. (2009). The relative benets of meta-analysis conducted with Psychological Bulletin, 92, 500504.
individual participant data versus aggregated data. Psychological Methods, 14, 165176. Rotgans, J., & Schmidt, H. (2009). Examination of the context-specic nature of self-
Cred, M., & Kuncel, N. R. (2008). Study habits, skills, and attitudes: The third pillar regulated learning. Educational Studies, 35, 239253.
supporting collegiate academic performance. Perspectives on Psychological Science, Schmidt, F. L., & Le, H. A. (2004). Hunter and Schmidt meta-analysis programs (version 1.1)
3, 425453. [computer software].
346 M. Cred, L.A. Phillips / Learning and Individual Differences 21 (2011) 337346

Schunk, D. H. (2005a). Self-regulated learning: The educational legacy of Paul R. Howey, S.C. (1999). The relationship between motivation and academic success of
Pintrich. Educational Psychologist, 40, 8594. community college freshmen orientation students. Unpublished doctoral disser-
Schunk, D. H. (2005b). Social cognitive theory and self-regulated learning. In B. J. tation, Kansas State University.
Zimmerman, & D. H. Schunk (Eds.), Self-regulated learning and academic *Jolly, M. G. (1992). Assessing learning skills, strategies and motivational orientations in
achievement: Theoretical Perspectives. New York, NY: Lawrence Erlbaum Associates. introductory accounting education. Unpublished doctoral dissertation, University
Stricker, L. J., Rock, D. A., Burton, N. W., Muraki, E., & Jirele, T. J. (1994). Adjusting college of Idaho.
grade point average criteria for variations in grading standards: A comparison of *Justice, E. M., & Dornan, T. M. (2001). Metacognitive differences between traditional-age
methods. The Journal of Applied Psychology, 79, 178183. and nontraditional-age college students. Adult Education Quarterly, 51, 236249.
Suksamram, A. (2003). Factors relating to self-regulated learning of the rst year *Klassen, R. M., Krawchuk, L. L., Lynch, S. L., & Rajani, S. (2008). Procrastination and
nursing students. Unpublished doctoral dissertation, Mahidol University, Thailand. motivation of undergraduates with learning disabilities: A mixed-methods inquiry.
Thongnoum, D. (2002). Self-efcacy, goal orientations, and self-regulated learning in Learning Disabilities Research and Practice, 23, 137147.
Thai students. Unpublished doctoral dissertation, Oklahoma State University. *Klassen, R. M., Krawchuk, L. L., & Rajani, S. (2008). Academic procrastination of
Tutor, P. T. (2006). Factors inuencing nursing students' motivation to succeed. undergraduates: Low self-efcacy to self-regulate predicts higher levels of
Unpublished doctoral dissertation, University of Southern California. procrastination. Contemporary Educational Psychology, 33, 915931.
Van der Stoep, S. W., & Pintrich, P. R. (2003). Learning to learn: The skill and will of college *Kosnin, A. M. (2007). Self-regulated learning and academic achievement in Malaysian
success. Upper Saddle River, NJ: Prentice Hall. undergraduates. International Education Journal, 8, 221228.
Vermeeten, J., Lodewijks, H., & Vermunt, J. (1997). Change and stability in learning *Lee, L. H. (1997). Goal orientation, goal setting, and academic performance in college
strategies during the rst two years at the university. Paper presented at the annual students: An integrated model of achievement motivation in school settings.
meeting of the American Educational Research Association, Chicago, March 2428. Unpublished doctoral dissertation, University of Southern California.
Warr, P., & Downing, J. (2000). Learning strategies, learning anxiety, and knowledge *Lynch, D. -J. (2006). Motivational factors, learning strategies and resource manage-
acquisition. British Journal of Psychology, 91, 311333. ment as predictors of course grades. College Student Journal, 40, 423428.
Welsh, J.B. (2007). Identifying factors that predict student success in a community *Lynch, R., & Dembo, M. (2004). The relationship between self-regulation and online
college online distance learning course. Unpublished doctoral dissertation, learning in a blended learning context. The International Review of Research in Open
University of North Texas. and Distance Learning, 5, 116.
Whetzel, D. L., & McDaniel, M. A. (1988). Reliability of validity generalization databases. *Maslin, L. Y. L. (1997). Self-regulated learning and science achievement in a community
Psychological Reports, 63, 131134. college. Unpublished doctoral dissertation, University of Southern California.
Wolters, C. A. (2003). Understanding procrastination from a self-regulated learning *McClendon, R. C. (1993). Construct validity of the MSLQ and cognition as related to
perspective. Journal of Educational Psychology, 95, 179187. motivation and learning strategy use of preservice teachers. Unpublished doctoral
Zakzanis, K. K. (1998). The reliability of meta-analytic review. Psychological Reports, 83, dissertation, The University of Akron.
215222. *McCulloch, L. M. (2006). The relationship among hope, optimism, gender, and
Zimmerman, B. J. (1989). A socialcognitive view of self-regulated academic learning. academic achievement. Unpublished Thesis, University of British Columbia, Canada.
Journal of Educational Psychology, 81, 329339. *McGee, J. V. (2005). Cognitive, demographic, and motivational factors as indicators of
Zimmerman, B. J. (1990). Self-regulation and academic achievement: An overview. help-seeking in supplemental instruction. Unpublished doctoral dissertation, Texas
Educational Psychologist, 25, 317. A&M University.
Zimmerman, B. J. & Schunk, D. H. (1989). Self-regulated learning and academic *McKenzie, K., & Gow, K. (2004). Exploring the rst-year academic achievement of
achievement: Theory, research, and practice. New York: Springer Verlag. school leavers and mature-age students through structural equation modeling.
Learning and Individual Differences, 14, 107123.
*McKnight, G. L. (1993). The inuence of learning strategies training upon student
strategy use, motivation, and achievement in a university physiology course.
FURTHER READING Unpublished doctoral dissertation, The University of Texas at Austin.
*Muis, K. R., & Franco, G. M. (2009). Epistemic beliefs: Setting the standards for self-
*Akin Palmer, J. (2008). The relationship between motivational proles and admission regulated learning. Contemporary Educational Psychology, 34, 306318.
grade point scores to rst semester attrition in an associate degree nursing (ADN) *Perry, R. P., Hladkyj, S., Pekrun, R. H., & Pelletier, S. T. (2001). Academic control and
program. Unpublished doctoral dissertation, Capella University, Minnesota. action control in the achievement of college students: A longitudinal eld study.
*Baldwin, C.A. (2001). Achievement goals and exam performance: An exploration of the Journal of Educational Psychology, 93, 776789.
role of study strategies and anticipatory test anxiety. Unpublished doctoral Pintrich, P. R. (1989). The dynamic interplay of student motivation and cognition in the college
dissertation. Indiana State University, Indiana. classroom. In C. Ames, & M. L. Maehr (Eds.), Advances in motivation and achievement:
Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Motivation-enhancing environments, Vol. 6. (pp. 117160)Greenwich, CT: JAI.
Englewood Cliffs, NJ: Prentice-Hall. *Ray, M. W. (2003). The role of aptitude and achievement in developmental college
*Black, K. (2008). Self-regulated learning as a predictor of persistence among rst year students' motivational beliefs, perceptions of self-regulated learning, and accuracy
students. Unpublished doctoral dissertation, University of Northern Colorado. of grade predictions. Unpublished doctoral dissertation, University of Missouri
*Brackney, B. E., & Karabenick, S. A. (1995). Psychopathology and academic Kansas City.
performance: The role of motivation and learning strategies. Journal of Counseling *Reed, B. J. (2003). Inuences upon academic achievement: The relationships among
Psychology, 42, 456465. cognitive strategies, metacognitive strategies and motivation. Unpublished
*Campbell, M. M. (2007). Motivational systems theory and the academic performance doctoral dissertation, Wayne State University, Michigan.
of college students. Journal of College Teaching and Learning, 4, 1123. *Selvester, P. M. (2004). Self-regulated learning: Understanding how preservice teachers
*Clarke, R. B. (2006). Undergraduate mathematics students: How do differences in learn to teach. Unpublished doctoral dissertation, University of Southern California.
motivation and learning strategies use affect performance? Unpublished doctoral *Silagyi-Rebovich, E. J. (1996). Motivation and learning strategies among dietetic
dissertation, The University of Memphis. students at four-year colleges and universities. Unpublished doctoral dissertation,
*Coutinho, S. (2008). Self-efcacy, metacognition, and performance. North American University of South Carolina.
Journal of Psychology, 10, 165172. *Singleton-Williams, S. (2009). Motivational and cognitive learning strategies as
*Coutinho, S. A., & Neuman, G. (2008). A model of metacognition, achievement goal predictors of academic success in economically disadvantaged community college
orientation, learning style and self-efcacy. Learning Environments Research, 11, 131151. students. Unpublished doctoral dissertation, Capella University, Minnesota.
*Creason, L. M. (2005). Relationships among community college developmental *Stump, G., Husman, J., Chung, W., & Done, A. (2009). Student beliefs about intelligence:
reading students' self-regulated learning, Internet self-efcacy, reading ability Relationship to learning. Paper presented 39th ASEE/IEEE Frontiers in Education
and achievement in blended/hybrid and traditional classes: A program review. Conference, San Antonio, Texas.
Unpublished doctoral dissertation, University of Missouri Kansas. *Sullivan, M. E. (2003). Self-regulated learning of medical students: Assessment of a
*Daniels, L. M., Haynes, T. L., Stupnisky, R. H., Perry, R. P., Newall, N. E., & Pekrun, R. social cognitive model. Unpublished doctoral dissertation, University of Southern
(2008). Individual differences in achievement goals: A longitudinal study of California.
cognitive, emotional, and achievement outcomes. Contemporary Educational *Talbot, G. L. (1994). The assessment of student study skills and learning strategies to
Psychology, 33, 584608. prepare teachers for academic advising tasks. Canada: Champlain St Lawrence.
*Daniels, L. M., Stupnisky, R. H., Pekrun, R., Hayes, T. L., Perry, R. P., & Newall, N. E. *Van Zile Tamsen, C. (1997). Self-directed learning and adults: The role of personal
(2009). A longitudinal analysis of achievement goals: From affective antecedents to control beliefs. Paper presented at the annual meeting of the Eastern Educational
emotional effects and achievement outcomes. Journal of Educational Psychology, Research Association. South Carolina: Hilton Head.
101, 948963. *Watson, M., McSorley, M., Foxcroft, C., & Watson, A. (2004). Exploring the motivation
*Davenport, M. A. (1999). Modeling motivation and learning strategy use in the orientation and learning strategies of rst-year university learners. Tertiary
classroom: An assessment of the factorial, structural, and predictive validity of the Education and Management, 10, 193207.
Motivated Strategies for Learning Questionnaire. Unpublished doctoral disserta- *Wilkinson, L. F. (1990). Interactions of motivational and learning strategy variables
tion, Auburn University, Alabama. with achievement when controlling for knowledge base in an undergraduate
*Dunigan, B., & Curry, K. J. (2006). Motivation and learning strategies of students in science course. Unpublished doctoral dissertation, The University of Akron, Ohio.
distance education. Journal of the Mississippi Academy of Sciences, 51, 140155. *Zachariah, S. (1995). The effects of multimedia-based instruction on the academic
*Harackiewicz, J. M., Barron, K. E., Elliot, A. J., Tauer, J. M., & Carter, S. M. (2000). Short- achievement of students studying economics. Unpublished doctoral dissertation,
term and long-term consequences of achievement goals: Predicting interest and The University of Akron, Ohio.
performance over time. Journal of Educational Psychology, 92, 316330. *Zusho, A., Pintrich, P. R., & Coppola, B. (2003). Skill and will: The role of motivation and
cognition in the learning of college chemistry. International Journal of Science

denotes sources included in the meta-analyses. Education, 25(1081), 1094.