DOCUNENT RESUME ,5 - ERICDOCUNENT RESUME ED 189 593 CS 005 579. AUTHOR. Nark, David: And Others. TITLE. The Effects of Anxiety on Reading Comprehension. PUB DATE Oct 76. NOTE. 17p.:

,5DOCUNENT RESUME

ED 189 593 CS 005 579

AUTHOR Nark, David: And OthersTITLE The Effects of Anxiety on Reading Comprehension.PUB DATE Oct 76NOTE 17p.: Paper presented at the Annual Meeting of the

North Cemtral Reading Association (21st, Notre Dame,IN, October 20-21, 1978). The short abstract will notreproduce clearly due to small type size. Occasionallight and broken type..

EDPS PRICEDESCRIPTORS

MP01/PC01 Plus Postage.College Students; Higher :education: *ReadingComprehension: *Reading Research: *Reading Tests:*Test Anxiety: Testing Problems

ABSTRACTThe effects of test anxiety on reading comprehension

were studied in two experiments. In the first 'experiment, 75 collegestudents completed a test anxiety scale and the McGraw-Hill BasicSkills System reading comprehension subtest. The low anxious studentsin this experiment showed higher reading comprehension than the highanxious students. In the second experiment, 102 college students wereadministered the test anxiety scale and the Nelson-Denay IleadingTest, a "speeded" test that requires completion in a short time. Noneof the students in experiment two finished the test, and there wereno significant differences in reading comprehension across high,medium, and low anxiety groups. To examine the "speededness"/anxietyrelationship more carefully, the Nelson-Denny test was administeredto a group of 19 college students who were allowed to complete thetest without a time limit. Under these circumstances, there was somerelationship found between anxiety scores and reading scores:however, the differences did not reach the conventional level ofstatistical significance. (RL)

***********************************************************************Reproductions supplied by EDRS are the best that can be made

from the original document.***********************************************************************

(74%

111

Co

C:31.4J

The Effects of Anxietyon Reading Comprehension

David WarkJ. Michael BennettJulie Wolinski

University of Minnesota

u S OEPARTOENT OF HEAL TNEDUCATION & WELFARENATIONAL INSTITUTE Of

EDUCATION

Doi 1/4,,,,AL NI /40 Eik EN REPRO-OLAF ExAt ILI' AS RECEIVED FROMTHE PEINSON OR ORGANIZATION ORIGIN*

POINIS OF vIEW OR OPINION,.S'AIED DO NOT NICESSARIL REPREcf NI Of f.I:IAL NATIONAL INITIT UTE OFt Out A 1 [ON POSI1ION OR POL. I( Y

-PERMISSION TO REPRODUCE THISMATERIAL HAS BEEN GRANTED BY

ABSTRACT

In this preijeci the authors explore the effectsof anxiety on reading comprehension in series f twoexperiments. Experiment One investigates anxiet) in aHow to Study Class principally by comparing scores onSarason's Text Anxiety Scale with Part III, Comprehen-sion, of the McGraw Hill Basic Skills System readingtest. The second experiment compared performance on theNelson Denny Reading Test for High School and Collegeand scores on the Sarason Anxiety Scale. An analysis ofperformance by item type revealed that students inExperiment One suffered more on items requiring them topiece together information from more than one place.Students in Experiment Two did not show this same anxietyeffect. However, since students in the second experi-ment were lass anxious and worked with a totally differentset of items, it is not clear which variables may accountfor the differences between the two experiments.

Test anxiety, as a scientific concept,is approximately twenty years old (Sara-son, 1972). The most common example oftest anxiety is the student who consist-ently earns lower scores on classroomexams than one would expect, knowing thatstudent's tested ability. Another fre-quent clinical example is the student whoreports forgetting well-studied materialuntil after turning in the exam. Thetruly test anxious student reports a rushof recall, after it's too late. A third,and somewhat less frequent example is thenormally fluent reader who, in the middle

David Wark

TO THE EDUCATIONAL nESOURCESINFORMATION CENTER (ERIC1."

Mirk, Bennett, Wolinski

of a paragraph, reports failing tocomprehend familiar material. All thesephenomena--lower than expected performance,

a sudden loss and then return of memoryunder test taking situations, andAunaccountable gaps in reading comprehen-sion--are examples of test anxiety.

There are several studies of testanxiety and its impact on college achieve-ment. Gaudry and Spielberger (1971)

reported the relationship between anxiety,grades and college dropouts at DukeUniversity. They found that for low andmedium aptitude students, as measured bythe College Hoards, there was a negativecorrelation with test anxiety and grades.Only the high aptitude students were ableto perform well in spite of high anxiety.And highly anxiods students were morelikely to drop out of college for aca-demic failure than were low anxietyancients, even when ability was held con-

stant. A most illumidating finding, bySarason (1972), is that under non-threat-ening or non-testlike conditions, highanxious students do about as well as lowanxious students.

There have been some studies of testanxiety and reading.. A recent ERICreview by Wildemuth (1977) includesseveral studies showing a negative rela-tionship between test anxiety and variousreading measures in children. Holmes(1972) reports a negative correlation withvocabulary for fourth graders. Johnsonand Hummel (1971) report similar negativecorrelation with vocabulary and paragraphcomprehension on the Stanford Achievementfor fifth graders. Kestenbalm and Weiner(1970) report a negative correlation withreading performance for seventh and eighth

2

Wark, Bennett, Wolinski

grade readers. Lombardi (1974) reports,for seventh graders, another negativecorrelation with reading, at all levelsof measured intellionce. Proger (1973)reports for sixth graders that highanxious students profit from abundantadvanced organizers. Low anxious stu-dents, on the other hand, seem not toneed these kinds of aids. Interestingly,there were no reports in the ERIC reviewof any studies of the relationshipbetween test anxiety and reading incollege students. We now report two suchstudies.

Experiment 1

Test Anxiety in a How to Study Class

Subjects in the first study were 75students in How to Study classes taughtat the University of Minnesota. As partof the introductory testing, they filledout the 16-item Test Anxiety Scale'(Sarason, 1958). The instrument is devel-oped from the notion that anxiety is aresult of negative evaluation. A childbeing so evaluatels1 by its parents be-comes hostile, guilty and eventuallyanxious. To the extent that a testsituation involves the threat of negativeevaluation, some of those same emotionsmay be generated. The test anxiouscollege student pays more attention tothe internal feelings of anxiety than tothe test item. Consequently, performanceis impaired. Sample items from the testcre as follows:

1. I wish examinations did notbother me so much.

2. Thoughts of doing poorly interferewith my performance on tests.

3

Wark, Bennett, Wolineki

During tests, / find myself thinkingof the consequences of failing.

Scores ranha from zero to 16. The distri-bution of scores for the How to Studyclass are presented in Figure 1. The topand bottom 33% of that distribution,cutting at 6.5 and 11.5, were used todefine high and low anxiety within theclass.

The reading test used in thisresearch was the McGraw-Hill Basic SkillsSystem Reading Test, Form A. Part III,the.ComprehenOon section, was used asthe measure of readin! comprehension. Thetest was given under standard conditions,which include a very generous 40 minutestime limit for the 30 comprehensionquestions. All students finished allitems. The scores were examined to findany relationship between levels of test.anxiety and reading comprehension.

The reading and anxiety scores were'evaluated using a two-way analysis ofvariance, presented_in Table I. Themeasured variables-were anxiety and compre-hension. Both main effects were signifi-cant, PIC.001. Low anxious studentsshowed higher comprehension (mean = 23.5)than high anxious students (mean = 18.2).

Another way to interpret the dif-ference between high and low anxiety isto convert the comprehension means topercentiles.for freshmen at four-yearcolleges. The low anxious students fellat the 79 percentile; the high anxiousstudents at the 18th. (In most college-level reading improvement programs in thecountry, the highly anxious students would


FIGURE I

TEST ANXIETY SCALE SCORES FOR TWO STUDENTGROUPS

11 X

10 X HOW9 X TO8 X STUDY

7 X X X X N=75

6 X X X X X

5 X X X X X X X

4 X X X X X X X X X X X

3 XXXXXXXX X X XX X

2 X X X X X X X X X X X X X X

1X X X X X X X X X X X X X X X X X

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

TEST ANXIETY SCALE

17

16

15 X X X

14 X X X

13 X X X

12 X X X

11 X X X X

10 XXXXX X

9 XXXXX X

81X X X X X X X

7 X X X X X X X

6 X X X X X X X X

5 X X X X X X X X

4 X X X X X X X X X



ii X X X X X X X X X X X

EFFICIENTREADINGN=102

X

X

X

N X

X X X

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

TEST ANXIETY SCALE

5


aa

TABLE I

ANALYSIS OF VARIANCE OF TEST ANXIETY ANDREADING COMPREHENSION

ANXIETY

HO MED

LO

0

r=

MED

HI

HI

1

19 16.2 14.0

24.3 19.75 17.66

28.14 25.22 21.77

23.5 20

79

18.2

34

= .45 N=75

ANOVA

J.6.57

20.45

24.80

SOURCE. df S S M S

ANXIETY 2 359.00 179.5 50.70 4

COMPREHENSION 2 862.135 431.08 121.77 *

INTERACTION 4 .113.370 28.34 8.005*

ERROR 66 233.926 3.54

*P (.01

6

7

Wark, Benneit, Wolinski

have been categorized as a remedial treat-ment group.) The interaction betweenanxiety and comprehension was significantat P< .001. Clearly, a student's levelof anxiety is related to performance onthe McGraw-Hill Reading Test.

Experiment II

Test Anxiety in anEfficient Reading Class

In an effort to expand our sphereof test anxiety research, we tried anexperiment with another group of Univer-sity/of Minnesota students. For thistreatment, we used the Nelson-Denny Read-ing Test for High School and College,Form C. The Nelson-Denny is a very well-known and frequently used test for collegestudents and adults. The Nelson-Denny isquite different from the McGrati-Hill andmany other tests in that it is "speeded"--that is, the time provided for complet-ing the test is quite short. The primaryreason for the short testing time is toallow administration of the complete testduring one normal class period. Bilt italso seems to create an interesting situa-tion insofar as test anxiety is concerned:Would speededness increase anxietyeffects?

First, we administered the SarasonAnxiety Scale to 102 students registeredin three sections of the University ofMinnesota Efficient Reading course. Theirtest anxiety scores are reported inFigure 1. This group was clearly lessanxious than was the How to Study classgroup reported in Figure 1. The anxietyscores were grouped into High, Medium, and

,/

.;d'.Wark, Bennett, Wolinaki

Low Anxiety categories, cutting at4.5 and 10.5.

The Nelson-Denny Reading Test wasadministered under standard conditions.The comprehension section of,the testallows the student fifteen minutes toread eight separate passages and answera total of 36 questions. None of thestudents finished the te:":. The readingscores were tested by an analysis ofvariance which found no significant dif-ferences between the three anxiety groups.Because of unproportional differences ingroup sizes, the data could not be testedfor°an interaction between comprehensionand anxiety. Data are reported in the,top row of Table.2,s.

In order to examine the speededness/anxiety relationship more carefully, weadministered the Nelson-Denny to a secondgroup of 19 students. They were olderthan average, registered in eveningschool, and generally more anxious thaneven the How to Study students. In thisinstance, they were given the test-anxiety questionnaire and then the Nelson-Denny under "power" conditions--that is,

with no time limits. Under such circum-stances, there was some relationshipbetween anXiety scores and reading scores.The product-moment correlation betweenanxiety and performance was r = -.45.While the differences between the threegroups (High, Medium and Low Anxiety) didnot reach the conventional level ofstatistical significance, the probabilitywas sufficiently small (141(.09) to suggestthat with a larger group under higheranxiety and power conditions, even theNelson-Denny Reading Test is sensitive to

8

9

Wark, Bennett,;,Wolinski

TABLE 2

HI ANX. MEDIUM ANX.ERAELSPEED CONDITION

T.A.S. 0 - 4 5 - 9 104.

2: 62 19

41.23 40.70 37.26

P v.43

POWER CONDITION

T.A.S. 0 - 5 6 - 9 10+

6

56.00 53.60 46.50

P= .09r= 745

READING COMPREHENSION SCORES FOR THREE

LEVELS OF TEST ANXIETY UNDER TWO TESTING

CONDITIONS.

9


the effects of test anxiety. The compar-ison between the speed and power groupsis presented in the bottom line ofTable 2.

More research into this matter iswarranted, however, because the Nelson-Denny is minimally sensitive to anxietywhen administered under standard condi-.tions. Perhaps the speededness of thetest tends to put everyone under the samethreat and thus masks the effects of testanxiety. Alternatively, the effects mayhave been attenuated in Experiment 11because the students were less anxiousthan those in Experiment I.

We are now left with the intriguingquestion: By what mechanism does anxietyaffect reading comprehension? Some sug-gestions for an answer may be found in theresearch literature on anxiety. Onetheoretical position that may be relevantis the cue utilization hypothesis(Muller, 1976). The general notion isthat anxious or aroused subjects usefewer of the available cues in codingstimuli for memory. The type of codingnegatively affects recall. Some of therelevant experimental work involvedincidental learning (McLaughlin, 1965).For example, if a student's,olcplicit taskis to underline all the nodns on severalpages of text, his score on an unexpectedrecall test of those nouns is considereda measure of incidental learning. 'Highanxious subjects tend to show less inci-dental learning, suggesting they attendto the features of only one task at atime, or that their memory may be impairedby anxiety.

Muller (1976) reports other studies

10

1

rkv 'c'


of the cognitive functioning of highanxiety subjects. In one experiment, )1epresented a series of words which were tobe read, meMorized and then repeated back.The word list was designed so that all theitems had a type of organization built in.One of the lists was.designed to have highverbal association between words scattere0throughout the list (black-white, king-queen, high-low). The other list wasdesigned to contain words that wererelated by category (six metals, sixanimals, six colors). The prediction wasthat highly anxious students would payattention to the words, and not theassociations or categories. If so, whentested on recall, they would ahow weakevidence of clustering, reproducing wordsthat were learned close together ratherthan relating worda scattered in the list.That's what Muller found, weaker cluster-ing by anxious students. In anotherexperiment, he also found the same clust-ering effect for words that had no built-in'associations. The less anxious stu-dents tended to create, their own clustersand recall certain words together, regard-les- of where they were placed in the \original list.

How might we translate these findingson recall and clustering under anxiety tocomprehension testing under anxiety? Thecue utilization hypothesis is that highanxious students may get part of a messagp,but miss some of the important detailsand associations. Consequently, they maymiss some of the relationships 1;etweenideas. If a comprehension question calledfor a particular fact, found in a particu-lar place in the reading text, thereshould be relatively little difference

1 2


between high and low anxious students.Students with different levels of aaxietyshould not be discriminated by that typeof item. On the other hand, there may besome questions that require inference andjudgment. They vrould be ffinswered bymaterial that would be found in severalplaces in the paragraph. The reader wouldhave to keep the relevant facts in mind,cluster them, see their associations, andthen answer a question. These types ofquestions, which are answered based onmateriAl from various places in the para-graph, might discriminate between high andlow anxious students.

These derivations were empiricallytested. Questions from the McGraw-Hilltest were divided .into two types, Type Iin which the answer was clearly located inonly one place, (Items 51, 53, 55, 57, 61,62, 63, 64, 65, 74, 77) and Type 2 inwhich the material needed for the answerappeared in more than one place (Items 54,56, 58, 59, 66, 67, 69, 71, 72, 73, 76,78, 79, 80).

e4The students in the study skills

class were Categorized into the top andbottom 277g. Then the scores for these twogroups of students were computed separ-ately for Type 1 and Type 2 items.

For Type 1, where we predicted nodifference for anxiety, the difference wasM=3.27, in favor of the low anxietystudents. In Type 2, where we expected agreater difference, we found it. Thedifference was M=4.66, again in favor oflow anxiety as we would have expected.The overall difference between the dif-ferences, D=1.39, has probability of .058;

12


very, very close to conventional levelsof statistical significance.

DISCUSSION

Test anxiety has been an area ofprofessional concern for two &cedes. Theearliest theory was oriented more towardquasi-psychoanalytical concepts. Exper-ience with punishing and evaluating par-ents was thought to affect a student'slevel of anxiety. Testing or classroomprocedures that reactivated that emotionallearning was considered the source oftest performance decriments at the colle6level, when the test was used for evalua-tion. More recent research by Wine (1971)has turned investigator's concern fromprior history to current performance.What the student attends to while testingis considered more critical than priorparental relations in evaluating testperformance. Using that interpretation,Wine has taught students6how to look atand analyze test items. Under her regime,anxious students tend to improve theirtest taking. Meuller's work, on thecue-use hypothesis, ic clearly alignedwith an attention theory.

The high anxious students in Experi-ment I showed significantly lower readingcomprehension. Analysis of their perform-ance by item types suggested theit theysuffered more on those items requiringthat they pull together information fromtwo or more places in the paragraph. Theywere also operating without any speedconstraints. Everyone of them finishedall the items. That suggests that theyhad all the time they wanted to go backand re-evaluate their work. It's worthnoting that under a speeded condition, the

L3

1 ei


students in Experiment 2 did not showAnxiety effects. However, there were twoconfounding differences between theexperiments. In the second experiment,the students were less anxious, and wereworking with a totally different set ofitems. It is not clear which, if any, ofthese variables account for the differ-ences between Experiments 1 and 2.Obviously, more work is called for toexplicate the theory of test anxiety, andfind out more about how the problemaffects reading comprehension.

REFERENCES

Gaudry, E., and Spielberger,Anxiety and Educational AcSidney: John Wiley & SonsPTY, Ldt., 1971.

Holmes, Charles Curtis. Specific Effects

Ph .D . dissertation University ofGeorgia, 1972.

Johnson, Gary L., and Hummel, Thomas J.

The Effects of Three Models of TestAdministrationon the Reading Achieve-ment Scores of Fifth Graders. Elemen-aryadvmpoi,Octo er 9 VOl. $ No. 1,

Kestenbaum, Joel M., and Weiner, Bernard.Achievement Performance Related toAchievement Motivation and Test Anxiety.Journal of Consulting and ClinicalPsychology, June 1970, Vol. 34, No. 3,

343-344.

Charles D.hievement., Australasia

14

1 5


Lombardi, Vincent A. The, Effect of TestAnxiety on the Relationship BetweenIntelligence and Reading Achievement ina Group of Seventh-Grade SuburbanChildren. Ed.D. dissertation, Temple,University, 1974.

McLaughlin, B. Intentional and IncidentalLearning in Human Subjects: The Role ofInstructions to Learn and Motivation.Psychological Bulletin, 1965, 63,359-376.

Muller, John H. Anxiety and Cue Utiliza-tion in Human Learning and Memory.In Zuckerman, Marvin and Spielberger,Charles D. (eds.) Emotion mid Anxiety.New York: Lawrence Erlbaum,'Associates,1976.

Prager, Barton B., and others. Advanceand Concurrent Organizers for DetailedVerbal, Passage Used with ElementarySchool Pupils. Journal of EducationalResearch, July-August 1973, Vol. 66,No. 10, 451-456.

Sarason, Irwin G. Experimental ApproachesTo Test Anxiety: Attention and the Useqof Information. In Spielberger,Charles D., ed., Anxiety, New York:Academic Press, 1972.

Sarason, .Irwin G. Inter-relationshipsAmong Individual Difference Variables,Behavior in Psychotherapy, and VerbalConditioning. The Journal of Abnormalan'd Social Psychology, 1958, 56,71q=3-44.

15

1 f


Wildemuth, Barbara M. Test Anxiety. AnExtensive Bibliography. ERIC/TM Report,1964. ERIC Clearing House on Tests,Measurement, and Evaluation. Princeton,New Jersey: Educational Testing Service,1977.

Wine, J. Test Anxiety and Direction ofAttention. Ruchological Bulletin,1971, 76, 92=IU4.

16

Documents

DOCUNENT RESUME ,5 - ERICDOCUNENT RESUME ED 189 593 CS 005 579. AUTHOR. Nark, David: And Others. TITLE. The Effects of Anxiety on Reading Comprehension. PUB DATE Oct 76. NOTE. 17p.: