Anda di halaman 1dari 5

CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 181

Math Anxiety: Personal, Educational, mathematical equivalent of literacy


(Paulos, 1988).
and Cognitive Consequences
Mark H. Ashcraft1
Department of Psychology, Cleveland State University, Cleveland, Ohio MATH ANXIETY DEFINED
AND MEASURED

Math anxiety is commonly de-


fined as a feeling of tension, appre-
laughter, and so forth. Many ask, hension, or fear that interferes with
Abstract
defensively, if their performance math performance. The first sys-
Highly math-anxious indi-
says anything about their overall tematic instrument for assessing
viduals are characterized by a
intelligence. These occasionally ex- math anxiety was the Mathematics
strong tendency to avoid math,
treme emotional reactions are not Anxiety Rating Scale (MARS), pub-
which ultimately undercuts
triggered by deliberately provoca- lished by Richardson and Suinn
their math competence and
tive procedures—there are no per- (1972). In this test, participants rate
forecloses important career
sonally sensitive questions or in- themselves on the level of anxiety
paths. But timed, on-line tests
tentional manipulations of stress. they would feel in various every-
reveal math-anxiety effects on
Instead, we merely ask college day situations, such as trying to re-
whole-number arithmetic
adults to solve elementary-school figure a restaurant bill when they
problems (e.g., 46  27),
arithmetic problems, such as 46  think they have been overcharged
whereas achievement tests
18  ? and 34  19  ? or taking a math test. My co-work-
show no competence differ-
The reactions are obvious symp- ers and I use a shortened version of
ences. Math anxiety disrupts
toms of anxiety, in this case math the test, which yields scores that
cognitive processing by com-
anxiety induced by ordinary arith- correlate well with scores obtained
promising ongoing activity in
metic problems presented in timed using the original test and also has
working memory. Although
tasks. On the one hand, it is almost very acceptable test-retest reliabil-
the causes of math anxiety are
unbelievable that tests on such fun- ity (i.e., an individual who takes
undetermined, some teaching
damental topics can be so upset- the test on different occasions gen-
styles are implicated as risk
ting; knowing that 15  8  7 erally receives similar scores). We
factors. We need research on
ought to be as basic as knowing have also found that for a quick de-
the origins of math anxiety and
how to spell “cat.” On the other termination, one can merely ask,
on its “signature” in brain ac-
hand, U.S. culture abounds with “On a scale from 1 to 10, how math
tivity, to examine both its emo-
attitudes that foster math anxiety: anxious are you?” Across at least a
tional and its cognitive
Math is thought to be inherently half-dozen samples, responses to
components.
difficult (as Barbie dolls used to this one question have correlated
say, “Math class is hard”), aptitude anywhere from .49 to .85 with
Keywords
is considered far more important scores on the shortened MARS.
anxiety; mental arithmetic;
than effort (Geary, 1994, chap. 7), There is a rather extensive litera-
math competence; working
and being good at math is consid- ture on the personal and educa-
memory; problem solving
ered relatively unimportant, or tional consequences of math anxi-
even optional. ety, summarized thoroughly in
In this article, I discuss what has Hembree (1990). Perhaps the most
My graduate assistant recently been learned about math anxiety pervasive—and unfortunate—ten-
told me about a participant he had across the past 30 years or so, and dency is avoidance. Highly math-
tested in the lab. She exhibited in- suggest some pressing issues to be anxious individuals avoid math.
creasing discomfort and nervous- pursued in this area. An important They take fewer elective math
ness as the testing session pro- backdrop for this discussion is the courses, both in high school and in
gressed, eventually becoming so fact that modern society is increas- college, than people with low math
distraught that she burst into tears. ingly data and technology ori- anxiety. And when they take math,
My assistant remarked that many ented, but the formal educational they receive lower grades. Highly
of our participants show some un- system seems increasingly unsuc- math-anxious people also espouse
ease or apprehension during test- cessful at educating students to an negative attitudes toward math,
ing—trembling hands, nervous adequate level of “numeracy,” the and hold negative self-perceptions

Copyright © 2002 American Psychological Society


182 VOLUME 11, NUMBER 5, OCTOBER 2002

about their math abilities. The cor- with other forms of anxiety range untimed, pencil-and-paper ver-
relations between math anxiety from .30 to .50. In a particularly sions of the math problems studied
and variables such as motivation clear display of the specificity of in the lab. For example, we (Faust,
and self-confidence in math are math anxiety, Faust (1992) found Ashcraft, & Fleck, 1996) found no
strongly negative, ranging between physiological evidence of increas- anxiety effects on whole-number
 .47 and  .82. It is therefore no ing reactivity (e.g., changes in heart arithmetic problems when partici-
surprise that people with math rate) when a highly math-anxious pants were tested using a pencil-
anxiety tend to avoid college ma- group performed math tasks of in- and-paper format. But when par-
jors and career paths that depend creasing difficulty. When the same ticipants were tested on-line (i.e.,
heavily on math or quantitative participants performed an increas- when they were timed as they
skills, with obvious and unfortu- ingly difficult verbal task, there solved the problems mentally un-
nate consequences. was hardly any increase in their re- der time pressure in the lab), there
Interestingly, math anxiety is activity (e.g., Ashcraft, 1995, Fig. 6), were substantial anxiety effects on
only weakly related to overall in- and participants with low math the same problems.
telligence. Moreover, the small cor- anxiety showed virtually no in- We have also taken a second ap-
relation of .17 between math anx- crease in either task. proach (see Ashcraft, Kirk, &
iety and intelligence is probably Hopko, 1998). In brief, we adminis-
inflated because IQ tests include tered a standard math achievement
quantitative items, on which indi- test to individuals with low, me-
viduals with math anxiety perform MATH ANXIETY AND dium, or high math anxiety, and
more poorly than those without MATH COMPETENCE replicated the overall result re-
math anxiety. The small correlation ported by Hembree (1990; i.e.,
(.06) between math anxiety and An obvious but unfortunate math achievement scores decrease
verbal aptitude supports this inter- consequence of the avoidance ten- as math anxiety increases). But we
pretation. However, math anxiety dency is that compared with peo- then scored the achievement test to
is related to several other impor- ple who do not have math anxiety, take advantage of the line-by-line
tant characteristics. As conventional highly math-anxious individuals increases in difficulty. With this
wisdom suggests, it is somewhat end up with lower math compe- scoring method, we found that
higher among women than men. tence and achievement. They are there were no math-anxiety effects
The gender difference is rather small, exposed to less math in school and whatsoever on the first half of the
may be particularly apparent in apparently learn less of what they test, which measured performance
highly selected groups (e.g., col- are exposed to; as a result, they on whole-number arithmetic prob-
lege students), and may be partly show lower achievement as mea- lems. Anxiety effects were appar-
attributable to a greater willingness sured by standardized tests (e.g., ent only on the second half of the
on the part of women to disclose Fennema, 1989). The empirical re- test, which introduced mixed frac-
personal attitudes. Nonetheless, lationship is of moderate strength tions (e.g., 10 1/4  7 2/3), per-
when we recruited participants for (a correlation of  .31 for college centages, equations with un-
research on math anxiety, we students), but sufficient to pose a knowns, and factoring. For these
found fewer men than women at dilemma for empirical work. That problems, there was a strong nega-
high anxiety levels, but just the re- is, when highly math-anxious indi- tive relationship between accuracy
verse at low levels (Ashcraft & viduals perform poorly on a test, and math anxiety. Thus, individu-
Faust, 1994). their poor performance could in als with high levels of math anxiety
Individuals who are high in fact be due to low competence and do not have a global deficit in math
math anxiety also tend to score achievement rather than height- competence, and they can perform
high on other anxiety tests. The ened math anxiety. If the relation- as well as their peers on whole-
strongest interrelationship is with ship between anxiety and compe- number arithmetic problems. In-
test anxiety, a .52 correlation. De- tence holds for all levels of math vestigations of higher-level arith-
spite the overlap among kinds of difficulty, then variations in com- metic and math, though, do need
anxiety, however, the evidence is petence will contaminate any at- to take the competence-anxiety re-
convincing that math anxiety is a tempt to examine math perfor- lationship into account.
separate phenomenon. For in- mance at different levels of math There is still reason to be some-
stance, intercorrelations between anxiety. what suspicious of this relationship
alternative assessments of math Fortunately, there are ways out between anxiety and competence,
anxiety range from .50 to .70, but of this dilemma. One is to test addi- however. Effective treatments for
intercorrelations of math anxiety tional samples of participants on math anxiety (see Hembree, 1990,

Published by Blackwell Publishing Inc.


CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 183

Table 8) have resulted in a signifi- only by sacrificing considerable ac- pened, we could infer that the pri-
cant improvement in students’ curacy. This behavior resembles mary task indeed depended on
math achievement scores, bringing the global avoidance tendency working memory, and that the
them nearly to the level shown by characteristic of highly math-anx- combination of tasks began to ex-
students with low math anxiety. ious individuals, but at an immedi- ceed the limited capacity of work-
Because the treatments did not in- ate, local level: By speeding through ing memory.
volve teaching or practicing math, problems, highly anxious individu- When the addition problem in-
the improvement could not be due als minimized their time and in- volved carrying, errors increased
to a genuine increase in math com- volvement in the lab task, much as substantially more for participants
petence. We suspect instead that they probably did in math class. Such with high math anxiety than for
these students’ original (i.e., pre- avoidance came at a price, how- those with low anxiety (Ashcraft &
treatment) math competence scores ever—a sharp increase in errors. Kirk, 2001, Experiment 2). More-
were artificially low, depressed by Second, the results showed that over, as we predicted, this was es-
their math anxiety. When the anxi- addition problems with carrying pecially the case when the second-
ety was relieved, a truer picture of were especially difficult for highly ary task became more difficult, that
their competence emerged. math-anxious individuals. In par- is, with a six-letter memory load.
ticular, the time disadvantage for On carry problems (e.g., 6  9,
carry versus no-carry problems was 27  15), highly anxious individu-
three times larger for participants als made 40% errors in the heavy-
COGNITIVE CONSEQUENCES with high anxiety (753 ms) than for load condition, compared with only
OF MATH ANXIETY those with low anxiety (253 ms), 20% errors for individuals with
even aside from the difference in low anxiety in the high-load condi-
Our original studies were ap- accuracy between the two groups. tion and 12% errors for both groups
parently the first to investigate Our interpretation was that carrying, in the light-load condition. In the
whether math anxiety has a mea- or any procedural aspect of arith- control conditions, with each task
surable, on-line effect on cognitive metic, might place a heavy demand performed separately, the compa-
processing, that is, whether it actu- on working memory, the system rable error rates were only 16% and
ally influences mental processing for conscious, effortful mental pro- 8%. These results could not be at-
during problem solving. In our cessing. In other words, we pro- tributed to overall differences in
early studies (Ashcraft & Faust, posed that the effects of math anxi- working memory. That is, we ex-
1994; Faust et al., 1996), we found ety are tied to those cognitive amined the participants’ working
that math anxiety has only minimal operations that rely on the re- memory spans (the amount of in-
effects on performance with single- sources of working memory. formation they were able to re-
digit addition and multiplication In an investigation of this possi- member for a brief amount of time)
problems. One anxiety effect we bility, Kirk and I (Ashcraft & Kirk, and found no differences between
did find, however, was in a deci- 2001) tested one- and two-column the groups when spans were as-
sion-making process sensitive to addition problems, half requiring a sessed with a verbal task. But span
“number sense” (Dehaene, 1997)— carry. We embedded this test scores did vary with math anxiety
when making true/false judg- within a dual-task procedure, ask- when they were assessed with an
ments, highly math-anxious indi- ing our participants to do mental arithmetic-based task.
viduals made more errors as the math, the primary task, while si- These results are consistent with
problems became increasingly im- multaneously remembering ran- Eysenck and Calvo’s (1992) model
plausible (e.g., 9  7  39), dom letters, a secondary task that of general anxiety effects, called
whereas low-anxiety participants taxes working memory. Two or six processing efficiency theory. In this
made fewer errors on such prob- letters were presented before each theory, general anxiety is hypothe-
lems. addition problem, and after partici- sized to disrupt ongoing working
Arithmetic problems with larger pants gave the answer to the prob- memory processes because anxious
numbers (e.g., two-column addi- lem, they were asked to recall the individuals devote attention to
tion or multiplication problems), letters in order. We reasoned that their intrusive thoughts and wor-
however, showed two substantial as the secondary task became more ries, rather than the task at hand. In
math-anxiety effects. First, partici- difficult (i.e., when more letters the case of math anxiety, such
pants at high levels of anxiety rou- had to be held in working mem- thoughts probably involve preoc-
tinely responded rapidly to these ory), performance on the primary cupation with one’s dislike or fear
problems, sometimes as rapidly as task might begin to degrade, in ei- of math, one’s low self-confidence,
participants with low anxiety, but ther speed or accuracy. If that hap- and the like. Math anxiety lowers

Copyright © 2002 American Psychological Society


184 VOLUME 11, NUMBER 5, OCTOBER 2002

math performance because paying swers . . . . He held them responsi- Sources of mathematical think-
attention to these intrusive thoughts ble for their lack of understanding,” ing: Behavioral and brain-imaging
acts like a secondary task, distract- p. 102). Turner et al. speculated evidence. Science, 284, 970–974.
ing attention from the math task. It that students with such teachers Eysenck, M.W. (1997). Anxiety and
follows that cognitive perfor- may feel “vulnerable to public dis- cognition: A unified theory. Hove,
England: Psychology Press.
mance is disrupted to the degree plays of incompetence” (p. 101), a Steen, L.A. (Ed.). (1997). Why numbers
that the math task depends on hypothesis consistent with our par- count: Quantitative literacy for to-
working memory. ticipants’ anecdotal reports that morrow’s America. New York: Col-
In our view, routine arithmetic public embarrassment in math lege Entrance Examination Board.
processes like retrieval of simple class contributed to their math anx- Tobias, S. (1987). Succeed with math:
Every student’s guide to conquering
facts require little in the way of iety. Thus, it is entirely plausible, math anxiety. New York: College
working memory processing, and but as yet undocumented, that Entrance Examination Board.
therefore show only minimal ef- such classroom methods are risk
fects of math anxiety. But problems factors for math anxiety.
involving carrying, borrowing, and Other gaps in the evidence in-
Note
keeping track in a sequence of op- volve the cognitive consequences
erations (e.g., long division) do rely of math anxiety, including those that 1. Address correspondence to Mark
on working memory, and so interfere with an accurate assess- H. Ashcraft, Department of Psychol-
should show considerable math- ment of math achievement and com- ogy, Cleveland State University, 2121
Euclid Ave., Cleveland, OH 44115;
anxiety effects. Higher-level math petence. My co-workers and I have e-mail: m.ashcraft@csuohio.edu.
(e.g., algebra) probably relies even shown that the transient, on-line
more heavily on working memory, math-anxiety reaction compromises
so may show a far greater impact the activities of working memory,
of math anxiety; note how difficult and hence should disrupt perfor- References
it will be, when investigating high- mance on any math task that relies Ashcraft, M.H. (1995). Cognitive psychology and
level math topics, to distinguish on working memory. The mecha- simple arithmetic: A review and summary of
new directions. Mathematical Cognition, 1, 3–34.
clearly between the effects of high nisms for this interference are not Ashcraft, M.H., & Faust, M.W. (1994). Mathemat-
math anxiety and low math com- yet clear, however. It may be that ics anxiety and mental arithmetic perfor-
mance: An exploratory investigation. Cognition
petence. intrusive thoughts and worry per and Emotion, 8, 97–125.
se are not the problem, but instead Ashcraft, M.H., & Kirk, E.P. (2001). The relation-
that math-anxious individuals fail ships among working memory, math anxiety,
and performance. Journal of Experimental Psy-
to inhibit their attention to those chology: General, 130, 224–237.
GAPS IN THE EVIDENCE distractions (Hopko, Ashcraft, Ashcraft, M.H., Kirk, E.P., & Hopko, D. (1998). On
the cognitive consequences of mathematics
Gute, Ruggiero, & Lewis, 1998). anxiety. In C. Donlan (Ed.), The development of
Math anxiety is a bona fide anxi- Finally, as research on mathe- mathematical skills (pp. 175–196). Hove, En-
gland: Psychology Press.
ety reaction, a phobia (Faust, 1992), matical cognition turns increas-
Dehaene, S. (1997). The number sense: How the mind
with both immediate cognitive and ingly toward the methods of cogni- creates mathematics. New York: Oxford Univer-
long-term educational implica- tive neuroscience, it will be sity Press.
Eysenck, M.W., & Calvo, M.G. (1992). Anxiety and
tions. Unfortunately, there has interesting to see what “signature” performance: The processing efficiency theory.
been no thorough empirical work math anxiety has in brain activity. Cognition and Emotion, 6, 409–434.
on the origins or causes of math The neural activity that character- Faust, M.W. (1992). Analysis of physiological reactiv-
ity in mathematics anxiety. Unpublished doc-
anxiety, although there are some izes math anxiety should bear toral dissertation, Bowling Green State
strong hints. For instance, Turner strong similarities to the activity University, Bowling Green, Ohio.
Faust, M.W., Ashcraft, M.H., & Fleck, D.E. (1996).
et al. (2002) documented the pat- associated with other negative af- Mathematics anxiety effects in simple and
terns of student avoidance (e.g., fective or phobic states. And our complex addition. Mathematical Cognition, 2,
25–62.
not being involved or seeking help) work suggests that the effects of
Fennema, E. (1989). The study of affect and mathe-
that result from teachers who con- math anxiety should also be evi- matics: A proposed generic model for re-
vey a high demand for correctness dent in neural pathways and re- search. In D.B. McLeod & V.M. Adams (Eds.),
Affect and mathematical problem solving: A new
but provide little cognitive or moti- gions known to reflect working perspective (pp. 205–219). New York: Springer-
vational support during lessons memory activity. Verlag.
Geary, D.C. (1994). Children’s mathematical develop-
(e.g., the teacher “typically did not ment: Research and practical applications. Washing-
respond to mistakes and misunder- Recommended Reading ton, DC: American Psychological Association.
standings with explanations,” p. Hembree, R. (1990). The nature, effects, and relief
101; “he often showed annoyance Dehaene, S., Spelke, E., Pinel, P., of mathematics anxiety. Journal for Research in
Stanescu, R., & Tsivkin, S. (1999). Mathematics Education, 21, 33–46.
when students gave wrong an- Hopko, D.R., Ashcraft, M.H., Gute, J., Ruggiero,

Published by Blackwell Publishing Inc.


CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 185

K.J., & Lewis, C. (1998). Mathematics anxiety acy and its consequences. New York: Hill and Turner, J.C., Midgley, C., Meyer, D.K., Gheen, M.,
and working memory: Support for the exist- Wang. Anderman, E.M., Kang, Y., & Patrick, H.
ence of a deficient inhibition mechanism. Jour- (2002). The classroom environment and stu-
nal of Anxiety Disorders, 12, 343–355. Richardson, F.C., & Suinn, R.M. (1972). The Math- dents’ reports of avoidance strategies in math-
ematics Anxiety Rating Scale. Journal of Coun- ematics: A multimethod study. Journal of
Paulos, J.A. (1988). Innumeracy: Mathematical illiter- seling Psychology, 19, 551–554. Educational Psychology, 94, 88–106.

Copyright © 2002 American Psychological Society

Anda mungkin juga menyukai