Math Performance and Its Relationship to Math

Anxiety and Metacognition

Angela M. Legg and Lawrence Locker, Jr.
Georgia Southern University
The current study assessed whether metacognitive skill moderates the
effects of math anxiety on performance, reaction time, and confidence on
a math task. Metacognition moderated math anxiety and predicted that
performance would decrease as anxiety increased, except at high
metacognition levels. Further, metacognition predicted confidence in
accuracy such that individuals higher in metacognitive processing were
more confident in their ability to correctly answer the problems.
Psychological and educational implications are discussed.

Math anxiety is defined as a general fear or tension associated with

anxiety-provoking situations that involve interaction with math. In a
world in which Eastern cultures reliably outperform Western cultures on
math performance tasks, the outcomes associated with math anxiety have
far reaching implications (Ginsburg, Choi, Lopez, Netley, & Chi, 1997;
Siegler & Mu, 2008; Stevenson & Stigler, 1992). Math anxiety can lead
to negative outcomes such as avoidance of college math courses and
majors or avoidance of careers that involve frequent math use (Ashcraft,
2002; Chipman, Krantz, & Silver, 1992, Hembree, 1990). For these
reasons, additional research on the implications of math anxiety and the
cognitive mechanisms associated with math anxiety is essential.
The mechanisms contributing to successful mathematical thinking are
complex and diverse, ranging from components operating within the
memory system to those contributing to problem solving and use of
cognitive strategies. One of the key cognitive mechanisms in math
problem solving, and a significant area of research within the math
cognition domain, is the utilization of the working memory system
(Ashcraft & Kirk, 2001; LeFevre, DeStafano, Coleman, & Shanahan,
2005). A critical issue, therefore, is how and why, math anxiety
modulates efficient cognitive processing, as it has been shown that
anxiety can tax working memory to such an extent that even individuals
with high math aptitude will perform poorly (Beilock & Carr, 2005).
Although anxiety can impact processing in a number of ways, one
proposal relevant to the current research is the view that anxiety may
create a dual-task situation that strains working memory resources
Author info: Correspondence should be sent to: Angela Legg, Georgia Southern
University, Department of Psychology, PO Box 8041, Statesboro, GA 30460
angelalegg@gmail.com
North American Journal of Psychology, 2009, Vol. 11, No. 3, 471-486.
NAJP

472

NORTH AMERICAN JOURNAL OF PSYCHOLOGY

(Ashcraft, 2002; Ashcraft & Krause, 2007; Eysenck & Calvo, 1992).
Individuals might, for example, devote attention to rumination on
anxious thoughts that divert resources from the task of problem solving
(e.g., devoting resources to worrying about performance rather than
application of problem solving strategies).
Given that math anxiety can hinder performance even for individuals
with high aptitude, it is important to investigate under what
circumstances certain mechanisms are involved in the intrusion of
anxiety in performance. A number of potential issues have been
explored, such as individual differences in working memory capacity as
well as the nature of the math problems themselves (Ashcraft & Krause,
2007; Beilock & Carr, 2005). However, a relatively unexamined area is
how ones own thoughts about math ability or cognitive processing may
or may not lead to deleterious effects of anxiety. Ashcraft (2002), for
example, called for research examining consequences of math anxiety in
relation to how individuals perceive their own math competence and
performance when solving math problems. One means of exploring this
aspect of math cognition is through an investigation of individual
differences in metacognition.
Loosely defined, metacognition is often referred to as thinking about
thinking. The abstract nature of metacognition leads to difficulty in
developing one all-encompassing, yet meaningful definition that lends
itself to empirical investigation (e.g., Schoenfeld, 1992). Explanations of
the construct also vary across disciplines, with educational research often
viewing metacognition differently than psychological and cognitive
domains. However, Schraw and Moshman (1995) proposed a fairly
exhaustive description of the components involved in metacognition that
is widely accepted in both the educational and psychological fields.
These researchers defined metacognition as consisting of two domains;
metacognitive knowledge and regulation of cognition. Each domain is
comprised of three subdomains.
The first domain, metacognitive knowledge, encompasses all of the
knowledge and insight possessed regarding what is already known about
cognitions, according to Schraw and Moshman (1995). This domain
basically refers to how aware someone is about their own cognitions or
thoughts. There are three subdomains of metacognitive knowledge or
awareness: declarative, procedural, and conditional awareness.
Declarative knowledge is the knowledge about what factors influence
learning and affect performance. Knowing that a good nights rest and
healthy breakfast can impact test performance is an example of
declarative knowledge. Procedural knowledge involves knowing how to
perform tasks, using skills automatically, and using strategies efficiently.
Driving, for example, benefits from enhanced procedural learning as the

Legg & Locker, Jr.

MATH PERFORMANCE & MATH ANXIETY

473

skills and strategies used to drive effectively often become automatic and
more efficient as one gains more experience with them. Finally,
conditional awareness includes knowledge of exactly when and why to
use specific strategies and when and why to choose alternates.
Conditional awareness might come into play when individuals complete
a math task under conditions where time is short. In this situation,
individuals will often have to compromise typical solving strategies and
replace them with shortcuts to solve problems in order to save time.
The second domain Schraw and Moshman (1995) proposed is
regulation of cognition. This domain is implicated in the control of
thought processes. There are three subdomains associated with the
regulation of cognition: planning, monitoring, and evaluation. Planning
involves selecting the appropriate strategies to solve problems and to
allocate resources in a manner that allows for efficient and effective
problem solving. For example, planning procedures might involve
making predictions, such as expecting certain test questions on a test and
thus focusing study efforts on those specific topics, while spending less
time on unanticipated test material. Whereas planning behaviors tend to
occur early before a behavior begins, monitoring, the second subdomain
of cognition regulation, is the present awareness of understanding and
performance. This subdomain regulates checking behaviors during tasks
and self-testing. Monitoring, for example, is employed when a student
attempts to paraphrase a paragraph he or she just read, without looking at
the page, in order to check for comprehension and retention. Finally,
evaluation, the third subdomain, often occurs after a task is completed
and involves appraising performance or the regulatory components of
cognition. This subdomain can also involve re-evaluating goals or
conclusions. For example, rewriting and editing a draft of a manuscript
heavily involves the evaluation subdomain as the person must be able to
take on the readers perspective in order to reevaluate the efficacy of the
writing.
From the above description, it is evident that the active application of
metacognitive processes such as monitoring and planning could be quite
beneficial in the context of math problem solving. However, if processes
such as self-monitoring or evaluation of strategies were negative or
directed to anxiety-focused thoughts or doubts, then metacognitive
processing could conceivably lead to negative outcomes. For example,
overt awareness of the correct strategies to apply problem-by-problem on
a timed test such as the GRE would lead to better performance. However,
ruminating on whether a prior problem was solved correctly or
excessively questioning whether a problem solving strategy is correct
could conceivably lead to hindered performance.

474

NORTH AMERICAN JOURNAL OF PSYCHOLOGY

Although the literature reviewing metacognition as it relates

specifically to math anxiety is fairly sparse, research does provide some
indication of metacognitive processes or perceptions that do relate to
math performance. Lucangeli, Coi, and Bosco (1997) conducted a study
examining metacognition of math difficulty in elementary school
children in relation to the nature of problems presented. Consistent with
Ashcraft and Krauses (2007) explanations of characteristics associated
with creating difficult math problems (number size, number of steps
required to solve, etc.), the fifth graders viewed problems containing
large numbers as more difficult than problems with smaller numbers. In
this study, individuals were classified in terms of the extent to which they
were higher or lower in metacognitive skill as well as the extent to which
they were better or poorer problem solvers. Not surprisingly, those
individuals classified as poor problem solvers demonstrated lower
metacognitive awareness and made more errors when problem solving,
suggesting a relationship between metacognition and effective problem
solving.
There is also evidence that suggests a link between anxiety and
metacognition, albeit within the verbal domain. Everson, Smodlaka, and
Tobias (1994), for example, assessed metacognitive word knowledge and
reading comprehension in college-age students. However, they found that
individuals with low anxiety were better able to use metacognition in a
positive way than their highly anxious counterparts. When anxiety was
high, metacognition was associated with poorer performance. Although
this experiment was conducted within the verbal domain, it may offer
some indication of how metacognition might relate to anxiety and
performance in other domains.
As the studies discussed indicate, there appears to be a link between
math problem solving and metacognition as well as some relationship
between anxiety and metacognition, at least within the domain of verbal
processing (Everson et al, 1994; Lucangeli et al., 1997). However, no
study to our knowledge has directly examined both metacognition and
anxiety within the domain of math performance. The current study
attempted to bridge this gap in the literature though an investigation of
how individual differences specific to math anxiety, as well as the use of
metacognitive skill, relate to math performance. That is, individuals were
assessed in terms of both their pre-existing levels of math anxiety as well
as their use of metacognition during processing. Math performance was
defined in terms of both the accuracy and speed of problem solving.
Participants judgments of the accuracy of their performance were also
assessed to determine whether differences in anxiety and metacognition
relate not only to actual performance, but also to confidence in
performance. If, as some research has indicated (Everson et al., 1994),

Legg & Locker, Jr.

MATH PERFORMANCE & MATH ANXIETY

475

high metacognition is related to more deleterious effects of anxiety, it

would be expected that individuals who are high in metacognitive use
and also experience high math anxiety levels would tend to display
poorer accuracy. Alternatively, individuals with low math anxiety and
high metacognition would be expected to display better accuracy.
We also predicted that the variables would be related to reaction time.
Some research suggests that anxiety leads to avoidance behaviors where
individuals try to solve problems as quickly as possible in an evasive
attempt to end the anxiety-inducing task (Ashcraft, 2002). For this
reason, we expected that individuals with high math anxiety and high
metacognitive use would display shorter reaction times, whereas those
with low math anxiety and high metacognitive use would display the
longest reaction times. Low anxiety and high metacognition would
presumably lead to greater utilization of checking behaviors, strategies,
monitoring behaviors, and accurate evaluation behaviors, thus resulting
in the longer time reaction times.
Confidence in perceived performance was also assessed. We
predicted that participants with high anxiety and high metacognition
would display lower perceptions of their ability to accurately perform
relative to individuals with low anxiety.
METHOD
Participants
Fifty-six Georgia Southern University undergraduates participated in
this study for course credit in an Introduction to Psychology course. The
mean participant age was 19.77 (SD = 2.45 yrs.). The participants
consisted of 41 (73.20%) women and 15 (26.8%) men. Most of the
participants were sophomores (48.20%); 26.80% were first-year students,
19.60% were juniors, and 5.40% were seniors. The majority of
participants (92.20%) had completed at least three high school math
courses and 80.30% of the participants had completed at least one college
level math course.
Measures
The Revised Math Anxiety Rating Scale (RMARS; Plake & Parker,
1982) was used to assess participants levels of math anxiety. This is a
24-item scale that assesses two factors (anxiety for learning math and
anxiety due to evaluation of math performance) and has been shown to
have good validity and reliability. Participants rate their level of anxiety
on a 0-4 Likert-type scale (0 = no anxiety at all and 4 = extreme anxiety).
Participants are asked to provide their level of anxiety during such tasks
as reading the word statistics, working on an abstract mathematical

476

NORTH AMERICAN JOURNAL OF PSYCHOLOGY

problem, and taking a final examination in a math class. The mean

RMARS score for the participants in this study was 1.51 (SD = .68).
Metacognition was measured using the State Metacognitive Inventory
(SMI; ONeil & Abedi, 1996). The SMI is a measure of metacognition in
the context of a task and was used as a measure of metacognitive
application during math performance. The SMI is a 20-item scale
assessing planning, checking, monitoring, and evaluating behaviors.
Specifically, the SMI asks participants to respond to statements such as,
I checked my work while I was doing it, I determined how to solve
the task problems, and I almost always knew how much of the task I
had left to complete. A 1-4 Likert-Type scale is used with this measure
(1 = did not engage in this behavior at all and 4 = engaged in the
behavior very much so.). Participants were instructed to reflect upon
their thinking processes during the math task and indicate how much they
engaged in various metacognitive processes. The mean SMI score for the
participants in this study was 2.68 (SD = .48).
Procedure
Following completion of the informed consent, participants
completed the RMARS. They were then provided instructions
concerning the math task. A modular arithmetic task was used because it
has been found in previous literature (Beilock & Carr, 2005) to be robust
to the effects of mathematical training and because the task is easy
enough to solve without a calculator or pen and paper. Modular
arithmetic involves judging whether the answer to a problem is a whole
number or a fraction. Participants were presented problems such as 60
10 (mod 5). To accurately solve the problem, participants must subtract
10 from 60 and then divide 50 by 5. The resulting number (10) is a whole
number so the original statement is true. An example of a
computationally difficult problem utilized in the task is 1527 349 (mod
4). Participants were asked to indicate whether the solution to each
problem was a whole number or not by indicating Y (yes, it is a whole
number) or N (no, it is a fraction) on the keyboard. Prior to the
experimental trials, participants completed seven practice trials. After
completing the practice trials participants were asked whether they had
any questions or required any clarification regarding the task.
In order to increase performance incentive, participants were
informed prior to the task that the 10 participants with the highest scores
on the math task would receive restaurant gift cards. The participants
were told that both accuracy and reaction time would be factored into
how well they performed on this task. That is, participants with the
highest accuracy within the shortest amount of time would be classified
as the top performers. Participants were then asked to complete the

Legg & Locker, Jr.

MATH PERFORMANCE & MATH ANXIETY

477

experimental math trials consisting of 20 modular arithmetic problems.

Math problems were presented one at a time in randomized order on a
computer screen using E-Prime (Schneider, Eschman, & Zuccolotto,
2002). After completing each problem, participants were prompted to
judge their accuracy on the preceding problem (the preceding problem
was removed from the screen when participants judged their accuracy).
They did so by selecting a number on a seven-point Likert-type scale
indicating their confidence that they provided the correct answer for the
previous problem (1 = Not confident at all, 7 = Extremely confident).
Participants were not provided feedback in terms of their actual
performance. All participants completed the task individually.
After completing this math task, participants completed the State
Metacognitive Inventory followed by a demographics survey that
included questions about age, gender, ethnicity, and number of high
school and college math courses taken. Participants were then debriefed
and told that they had an equal chance of winning one of the 10
restaurant gift cards and that their performance on the math task would
have no bearing on their chances of winning.
RESULTS
Preliminary Analyses
To assess whether demographic characteristics related to the
dependent measures as covariates, six MANOVAs were conducted with
accuracy, reaction time, and confidence entered as dependent variables.
The independent variables analyzed were age (dichotomized into 19 and
below, and 20 and above), ethnicity, gender, year in school, number of
math courses taken in high school, and number of math courses taken in
college. Six separate analyses were conducted in order to optimize the
chances that any significant covariates would be identified. None of the
independent variables significantly related to participants performance
on the modular arithmetic task in terms of accuracy, reaction time, or
confidence ratings. As none of the MANOVAs produced significant
results, no covariates were entered into the primary analyses.
The average proportion correct for the modular arithmetic task was
.80 or 16 out of 20 problems correct (SD = .12). Participants had a mean
reaction time for each problem on the math task of 13.11 seconds (SD =
4.83). The mean confidence rating was 5.32 (SD = .87).
Primary Analyses
Three multiple-regression equations were used to analyze the data.
The first analysis assessed the relationship between the predictors and
accuracy, the second assessed the relationship between the predictors and
reaction time, and the third assessed the relationship between the

478

NORTH AMERICAN JOURNAL OF PSYCHOLOGY

predictors and confidence ratings. Centered SMI and RMARS scores and
the interaction term were entered as predictors sequentially such that SMI
and RMARS were entered in the first block and the interaction term
entered in the second, according to the methodology set forth by Aiken
and West (1991). Centered predictors (RMARS and SMI scores) and the
interaction term were calculated according to the procedure created by
Baron and Kenny (1986). ModGraph was used to graph the prediction
equations (Jose, 2008).
The analysis for accuracy revealed that the overall regression model
was significant, R = .21, F (3, 55) = 4.66, p <.01. Math anxiety
significantly predicted performance, B = -.06, = -.35, t (55) = -2.75, p
<.01. Individuals with higher anxiety performed worse than those with
low anxiety. State metacognition also predicted performance, B = .08,
= .31, t (55) = 2.41, p <.05 in that higher metacognition was related to
higher accuracy. A moderating relationship between metacognition and
anxiety was also significant, B = .12, = .33, t (55) = 2.53 p < .05, in that
at high anxiety levels, individuals performed increasingly worse as their
SMI scores decreased. Accuracy, however, did not differ at low anxiety
levels regardless of state metacognitions. Figure 1 illustrates the pattern
of results.

Accuracy (Proportion Correct)

0.95
0.90
0.85
SMI Level

0.80

+1 SD
Mean

0.75
-1 SD

0.70
0.65
0.60
-1 SD

Mean

+1 SD

Math Anxiety Level

FIGURE 1 Pattern of Results for Accuracy (as measured by proportion

correct).
(-1 SD indicates a low level of math anxiety or state metacognitions and +1 SD
indicates a high level of math anxiety or state metacognitions)

Legg & Locker, Jr.

MATH PERFORMANCE & MATH ANXIETY

479

The regression model for reaction time revealed no significant

relationships. Participants at varying levels of anxiety and metacognitive
skill showed no differential time performance based on these individual
differences.
The overall regression equation for confidence was significant, R =
.18, F (3, 55) = 3.91, p < .05. The results revealed a relationship between
metacognition and confidence, B = .78, = .43, t (55) = 3.28, p < .01
such that individuals with high levels of state metacognition reported
greater confidence in their ability to correctly solve the math problems.
No other significant relationships emerged. Figure 2 illustrates the
pattern of results.

Confidence Ratings

7.00

6.00
SMI Level
+1 SD

5.00

Mean
-1 SD

4.00

3.00
-1 SD

Mean

+1 SD

Math Anxiety Level

FIGURE 2 Pattern of Results for Confidence on the Math Task

(-1 SD indicates a low level of math anxiety or state metacognitions and +1 SD
indicates a high level of math anxiety or state metacognitions)

DISCUSSION
The results of this study demonstrate that metacognition has a
moderating relationship with math anxiety that relates to accuracy in
math performance. In addition, increased metacognition is associated
with greater confidence in performance. In regard to the relationship
between metacognition and anxiety, the results would suggest that
individuals with higher anxiety benefit from higher levels of
metacognition, as their math performance was similar to those

480

NORTH AMERICAN JOURNAL OF PSYCHOLOGY

individuals with low math anxiety. As noted, the literature investigating

the relationship between metacognition and anxiety is rather sparse.
However, some research has suggested that metacognition may have a
negative impact on those individuals with higher anxiety (Everson et al.,
1994). The opposite pattern was observed here. One possibility that
should be considered is that the relationship of metacognition and anxiety
may be largely state-dependent relative to such factors as the
consequences of the outcome, the nature of the material presented as well
as the general context.
For example, in this study, the modular arithmetic task was used. This
task is used primarily to avoid practice effects relative to math
backgrounds as well as be computationally within the abilities of the
participants without a calculator. Further, by using a math task within the
capabilities of the participants, we reduced the potential confound of
assessing math competence rather than metacognitive skill and math
anxiety levels. Thus, the task was not designed to necessarily exceed the
capabilities of the participants. Furthermore, given that it was an
experimental situation, the consequences of failure to perform were
relatively minimal (e.g., as opposed to the GRE or SAT). It is therefore
likely that participants perceived the math task as not being beyond their
capabilities and the cost of not performing at peak was not as threatening.
Still, it should be noted that performance did decrease for some
individuals with higher math anxiety. Thus, the task was sensitive to
effects of math anxiety. However, the nature of the task was such that
individuals who were high in use of metacognition, even if high in math
anxiety, were able to effectively utilize the beneficial aspects of
metacognition. Such aspects may have included checking behaviors,
strategic use of problem solving, or effective deployment of strategies at
appropriate times per the conditional awareness subdomain of Schraw
and Moshmans (1995) metacognitive conception. If this is the case,
these strategies potentially mitigated anxiety-related influences, possibly
by allocating mental attention to metacognitive processes, rather than
anxiety-related thoughts. However, it is possible that had the context
been more analogous to a high-stress testing situation (e.g. the SAT or a
final exam), the highly anxious individuals might have utilized
metacognition in a negative fashion by ruminating on the situation and
potential outcomes, or worrying about poor performance rather than
planning or problem solving.
The notion that highly metacognitive individuals in the present study
were utilizing these processes in a positive fashion is supported by the
finding concerning the relationship between metacognition and
judgments of performance. Overall, higher metacognition was associated
with perceptions of better performance. Notably, a secondary analysis

Legg & Locker, Jr.

MATH PERFORMANCE & MATH ANXIETY

481

indicated that confidence in accuracy was positively correlated with

actual performance on the task (R = .43, p = .001). It would appear that
these participants were not only devoting mental resources to the task at
hand in an efficient manner, but were generally more confident in the
performance. Thus, it is possible that both the application of beneficial
problem solving strategies, and associated confidence as a result of doing
so, may have diminished or diverted attention away from negative
anxiety-related cognitions.
An educational implication of this finding would be to advocate
metacognitive training. Kruger and Dunning (1999) found evidence that
addressing metacognitive processes such as strategy use and checking
behaviors increased college students ability to perform well on varying
tasks. Furthermore, much of the educational literature suggests that
metacognitive training is beneficial to individuals in elementary, middle
and high school (Cardell-Elawar, 1995; Kramarski & Mevarech, 2003;
Teong, 2003).
Notably, metacognitive training has been shown to be a very effective
method in which to overcome mathematics problem-solving difficulties.
Metacognitive training is usually based on the principals set forth by
Polya (1945) and involves directing student and participant attention to
metacognitive thinking such as strategy use, problem solving, and time
and accuracy monitoring. Metacognitive training also involves
encouraging individuals to monitor their confidence in their abilities or
lack of confidence. Kramarski and Mevarech (2003), for example,
examined students performance interpreting a linear graph unit. Some
students received metacognitive training whereas others received
traditional teaching, either in groups or individually. Individuals who
received the metacognitive training performed significantly better than
those who received the traditional teaching method, regardless of
whether they received the metacognitive training in groups or
individually.
Furthermore, Kruger and Dunning (1999) showed that even if
students are examined in terms of differences in high and low
achievement, metacognitive training does have positive benefits,
although greater benefits seem to occur for low-achieving groups. These
researchers found that high achievers benefit most from apprehending the
superiority of their own answers by viewing other individuals responses
to problems. However, individuals at low-achievement levels benefit
from instruction regarding the skills necessary to correctly evaluate
themselves as well as how to positively use metacognitive strategies.
Similarly, Cardell-Elawar (1995) examined elementary and middle
school age children who were considered low-achievers in mathematics.
In this study, individuals were randomly assigned to either receive

482

NORTH AMERICAN JOURNAL OF PSYCHOLOGY

traditional teaching or metacognitive training. The metacognitive training

directed students to answer certain questions throughout the problemsolving process that related to metacognitive functioning such as, Do I
understand the words in this problem? and With what operations
needed to solve this problem do I typically have difficulty completing?
Students receiving the metacognitive training significantly improved
their performance compared to students in the control condition.
Interestingly, the students in the metacognitive training group also
exhibited improved attitudes toward mathematics. This finding supports
the notion that one benefit of metacognition may be related to promoting
feelings of self-efficacy.
An aim of future research, therefore, should be an examination of
math anxiety and metacognition across a broader range of contexts,
particularly in high-stress situations such as the SAT or GRE.
Importantly, the efficacy of metacognitive training should be examined
to determine whether such training could offset effects of anxiety even
within such high-stress situations. We have argued that such benefits
may occur by shifting processing resources from anxiety-related thoughts
to the present problem, as well as perhaps leading to greater feelings of
self-efficacy or confidence.
In terms of limitations of the current study, it should be noted that the
design was quasi-experimental as metacognition and math anxiety were
not manipulated factors. Future research should attempt to explore means
by which to experimentally manipulate both metacognition and anxiety
in order to make more robust claims regarding possible causal
relationships among these variables. One method by which to manipulate
math anxiety may be to use the methodology set forth by Beilock and
Carr (2005) in which participants are videotaped, told that they have a
partner who is relying on them to improve their performance, and that
professors will be evaluating the videotapes in the future. Metacognition
could experimentally be manipulated by providing some participants
with metacognitive training prior to the task. Another way in which to
obtain stronger control over metacognition would be to present some
participants with metacognitive analysis questions throughout a problem
solving task. For instance, if a participant is solving a math problem, at
random times throughout the process, different questions might appear
on the screen asking the participants to respond to various items such as,
How much time is remaining to complete this task? or Are there any
other ways to solve this problem that might be more efficient? Future
replication or extension of the current research should also be aimed at
the identification of the specific metacognitions that may lead to better
performance or buffer math anxiety effects. Experimental investigation
of the various metacognitive skills influencing performance and math

Legg & Locker, Jr.

MATH PERFORMANCE & MATH ANXIETY

483

anxiety (checking, planning, monitoring, and evaluating) will provide a

more complete picture of the beneficial nature of metacognition, as well
as potentially provide more insight into processing mechanisms affected
by math anxiety.
Also, the moderating pattern found in the current study
(metacognitive skill moderating anxiety in relation to accuracy) may
need to be interpreted with some caution due to some participants
performing at ceiling. For example, as the task was intended to be within
the range of capabilities for all participants, additional benefits of
metacognition for low-anxiety individuals may not have been detected
due to close to ceiling performance. But, as discussed above, benefits of
metacognition may be largely a function of the nature of the problems or
the context. Again, the findings presented here may only apply to math
tasks in which difficulty levels do not exceed the capabilities of the
individuals. Differing patterns may emerge in different contexts (e.g.,
high-stakes situations such as the SAT or GRE), although the proper use
of metacognition may be of benefit in those situations as well. Finally,
research should examine the extent to which the patterns obtained in the
current study are also observed in other populations. For example,
research might examine at what age metacognition begins moderating
anxiety. Additionally, adult populations from both math-intensive fields
such as computer science as well as non-math-intensive fields such as
English education might be examined to determine if there are
differences among various professional groups in how metacognition
relates to math performance.
Research on math anxiety and metacognition is still in its infancy as
the relevant variables, mechanisms, and outcomes are identified. The
implications of math anxiety are well known, far-reaching and possibly
contribute to the learning gap between Western and Eastern cultures. As
basic research explores the components related to this issue, applicable
interventions can be identified that mitigate the negative impact of mathrelated stress or anxiety. Metacognition may provide a viable means by
which to approach the problems of math anxiety and math avoidance in
our society.
REFERENCES
Aiken, L. S. & West, S. G. (1991). Multiple Regression: Testing and Interpreting
Interactions. London: Sage Publications.
Ashcraft, M. H. (2002). Math anxiety: Personal, educational, and cognitive
consequences. Current Directions in Psychological Science, 11, 181-185.
Ashcraft, M. H. & Kirk, E. P. (2001). The relationships among working memory,
math anxiety, and performance. Journal of Experimental Psychology:
General, 130, 224-237.

484

NORTH AMERICAN JOURNAL OF PSYCHOLOGY

Ashcraft, M. H. & Krause, J. A. (2007). Working memory, math performance,

and math anxiety. Psychonomic Bulletin and Review, 14, 243-248.
Baron, R. M. & Kenny, D. A. (1986). The moderator-mediator variable
distinction in social psychological research: Conceptual, strategic, and
statistical considerations. Journal of Personality and Social Psychology, 51,
1173-1182.
Beilock, S. L. & Carr, T. H. (2005). When high-powered people fail: Working
memory and choking under pressure in math. Psychological Science, 16,
101-105.
Cardell-Elawar, M. (1995). Effects of metacognitive instruction on low achievers
in mathematics problems. Teaching and Teacher Education, 11, 81-95.
Chipman, S. F., Krantz, D. H., & Silver, R. (1992). Mathematics anxiety and
science careers among college able women. Psychological Science, 3, 292295.
Everson, H. T., Smodlaka, I., & Tobias, S. (1994). Exploring the relationship of
test anxiety and metacognition on reading test performance: A cognitive
analysis. Anxiety, Stress, and Coping, 7, 85-96.
Eysenck, M. W. & Calvo, M. G. (1992). Anxiety and performance: The
processing efficiency theory. Cognition and Emotion, 6, 409-434.
Ginsburg, H. C., Choi, Y. E., Lopez, L. S., Netley, R., & Chi, C. Y. (1997).
Happy birthday to you: Early mathematical thinking of Asian, South
American and U.S. children. In T. Nunes & P. Bryant (Ed.), Learning and
teaching mathematics: An international perspective (pp. 163207). Hove,
England: Psychology Press.
Hembree, R. (1990). The nature, effects, and relief of mathematics anxiety.
Journal for Research in Mathematics Education, 21, 33-46.
Jose, P.E. (2008). ModGraph-I: A programme to compute cell means for the
graphical display of moderational analyses: The internet version, Version
2.0. Victoria University of Wellington, Wellington, New Zealand. Retrieved
January 27, 2008 from http://www.victoria.ac.nz/psyc/staff/paul-josefiles/modgraph/modgraph. php
Kramarski, B., & Mevarech, Z. R. (2003). Enhancing mathematical reasoning in
the classroom: The effects of cooperative learning and metacognitive
training. American Educational Research Journal, 40, 281-310.
Kruger, J. & Dunning, D. (1999). Unskilled and unaware of it: How difficulties
in recognizing ones own incompetence lead to inflated self-assessments.
Journal of Personality and Social Psychology, 77(6), 1121-1134.
LeFevre, J. A., DeStafano, D., Coleman, B., & Shanahan, T. (2005).
Mathematical cognition and working memory. In J. I. D. Campbell (Ed.),
Handbook of Mathematical Cognition (pp. 361-378). New York:
Psychology Press.
Lucangeli, D., Coi, G., & Bosco, P. (1997). Metacognitive awareness in good and
poor math problem solvers. Learning Disabilities Research and Practice,
12, 209-212.
ONeil, H. F. Jr. & Abedi, J. (1996). Reliability and validity of a state
metacognitive inventory: Potential for alternative assessment. The Journal of
Educational Research, 89, 234-245.

Legg & Locker, Jr.

MATH PERFORMANCE & MATH ANXIETY

485

Plake, B. S. & Parker, C. S. (1982). The development and validation of a revised

version of the mathematics anxiety rating scale. Educational and
Psychological Measurement, 42, 551-557.
Polya, G. (1945). How to solve it? Garden City, NY: Doubleday.
Schneider, W. Eschman, A., & Zuccolotto, A. (2002). E-Prime User's Guide.
Pittsburgh: Psychology Software Tools, Inc.
Schoenfeld, A. H. (1992). Learning to think mathematically: Problem solving,
metacognition, and sense-making in mathematics. In D. Grouws (Ed.),
Handbook for Research on Mathematics Teaching and Learning (pp. 334370). New York: MacMillan.
Schraw, G. & Moshman, D. (1995). Metacognitive theories. Educational
Psychology Review, 7, 351-371.
Siegler, R. S., & Mu, Y. (2008). Chinese children excel on novel mathematics
problems even before elementary school. Psychological Science, 19, 759763.
Teong, S. K. (2003). The effect of metacognitive training on mathematical wordproblem solving. Journal of Computer Assisted Learning, 19, 46-55.

Notes: We thank two anonymous reviewers for their valuable feedback and
comments regarding this manuscript. We presented portions of this research at
the Fourth Annual Meeting of the Georgia Psychological Society in Macon, GA
on April 11, 2009.

Copyright of North American Journal of Psychology is the property of North American Journal of Psychology
and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright
holder's express written permission. However, users may print, download, or email articles for individual use.