OL15

COMMUNICATIONAPPREHENSION OFELEMENTARY ANDSECONDARY STUDENTSAND TEACHERSOver the past decade, a substantialamount of research literature has de-veloped concerning oral communicationapprehension (CA). Most of the em-pirical work in this area has focused onthe correlates, effects, and treatments for

CA.l However, causal explanations forthe development of CA have receivedmuch less attention.

The most commonly employed theo-retical explanation of CA developmenthas been framed within a reinforcement

paradigm.:! As :\IcCroskey has suggested,"If a child is reinforced for being silentand is not reinforced for comumuni-

eating, the probable result is a quietchild. In addition, if the child not onlyis not reinforced for communicatingbut often experiences some aversive ex-perience (parent shouting, big brotherhitting) when attempting to com-municate, the quiet child result is even

James C. McCroskeyJanis F. Andersen

Virginia P. RichmondLawrence R. Wheeless

more probable."3 Evidence from casestudy analyses,4 research on CA treat-ment,a and research on social anxiety8has clearly supported such a reinforce-ment explanation.

No other potentially causal mechanismin the development of CA has receivedmore than limited empirical validation,although potential hereditary influenceon CA development has received someattention. After reviewing the literatureon infant sociability, social apprehen-sion, and physical attractiveness, Dalyconcluded that genetic factors may in-fluence the development of CA.7 Inextending this possible relationship,Andersen and Singleton found body

type (which is primarily geneticallydetermined) to predict CA in women.sWhile hereditary factors must be con-sidered a possible contributing elementin CA development, even ardent

proponents of social biology stress theJames C. McCroskey is Professor and Chair-p~rs~n: Janis .F. AndeT:en is ~ssistant Professor, 3 McCroskey. p. 80.Virginia P. Richmond IS Associate Professor. and 4 Gerald M. Phillips. "Reticence: PathologyLawrence R. Wheeless is Professor and Asso- of the Normal Speaker," Speech Monographs,date Ch3:irp~rson of the Depa~u~ent of. Spe~h 35 (March 1968). 39-49.CommunIcation, West VlrgtnIa UnIversity. IiJames C, McCroskey. "The ImplementationMorgantown, WV 26506. of a Large.Scale Program of Systematic De-

l For a summary of this research. see James sensitization for Communication Apprehension."~. McCroskey. "Oral Communication Apprehen- Speech Teacher, 21 (Nov. 1972). 255.64.SlOn: A Summary of Recent Theory and Re- 8 See. for example. John A. Daly, "The De-search," Human Communication Research, 4 velopment of Social-Communicative Anxiety,"(Fall 1977). 78.96. Paper presented at the annual convention of the

:!See. for example. William K. Ickes. "A ICA. Berlin. West Germany. 1977; and DalyClassical Conditioning Model for 'Reticence:" and Friedrich.Western Speech. 35 (Winter 1971). 48-55; John 7 Daly.A. Daly and Gustav W. Friedrich. "The De- 8 Petcr A. Andersen and George W. Singleton.velopment of Communication Apprehension: A '"The Relationship between Body Type andRetrospective Analysis of Some Contributory Communication Avoidance," Paper presented atCorrelates,"' Paper presented at the annual SCA the annual convention of the Eastern Communi-con\'cntion, ~Iil1ne:J.polis, 1978: and McCroskey. cation Assn., Boston. 1978.

COM:\IC\'ICATION EDt:CATION. Volume 30. April 1981

environment in altering genetically de-termined predispositions. We are left,then, with some indication that

hereditary elements may contribute toCA development but with most availabledata that points toward reinforcementpatterns as the primary causal agent.

Evidence on reinforcement as a

causal force in communication apprehen-sion suggests two potentially influentialenvironments, the home and the school.

Research by Giffin and Heider providesevidence that early experiences in thehome environment explain adult anx-iety.!) Other developmental literaturesupports this position.Io Daly andFriedrich note research findings whichpoint to differences in the home en-vironment, such as amount of familytalk and style of parent-child inter-action, as predictive of children'scommunication behaviors. Furthermore,they found that college students' recol-lections of early experiences in the homeenvironment were predictive of com-munication apprehension.ll Randolphand i\JcCroskey advanced a theory offamily size (based upon presumed dif-ferential reinforcement patterns) topredict communication apprehension.Early results were promising, butreplication with a larger sample failedto demonstrate a relationship betweenfamily size and communication appre-hension.I2 McCroskey and Richmond

!) Kim Giffin and M. Heider, "The Relation-ship between Speech Anxiety and the Sup-pression of Communication in Childhood,"Psychiatric Quarterly Supplement, part 2 (1967),311-22.

10 See, for e.."Cample, D. Thomas Porter,"Communication Apprehension Causation: To-ward an Empirical Answer," Paper presented atthe annual meeting of the ICA, Chicago, 1978.

11 Dalv and Friedrich.12 Fred L. Randolph and James C. Mc-

Croskey, "Oral Communicaticn ApprehensionAs a Function of Familv Size: A PreliminaryInvestigation," Paper presented at the annualconvention of the Eastern CommunicationAssn., New York, 1977; idem., "The Cause(s) ofOral Communication Apprehension: Failure ofa Theory," Paper presented :n the annual con-

COMMUNICATION APPREHENSION-I23

found that children from families in

rural environments were more likely tobe highly communication apprehen-sive.13

The school envionment has also beendiscussed as a causal .force in the de-

velopment of communication apprehen-sion. Phillips' analysis of case studiesof reticent individuals points to the firstfew years of school as importantto the development of communicationreticence.14 Also employing case studies,Davey found a large number of highreticents in the early grade schoolyears.15 Daly and Friedrich found thatcollege students' recollections of gradeschool environments were predictive ofcurrent apprehension.16 Furthermore,Porter found recollections of earlyschool experiences to be more relevantto communication apprehension de-velopment than recollections of pre-school home experiences. IT He citesdevelopmental theories which supportthe important influence of the schoolenvironment on child personality andsocial development.

NORMATIVE LEVELS OFCOMMUNICATIONAPPREHENSION

The primary purpose of the present re-search was to provide a basis for sub-sequent research probing the schoolenvironment as a potential cause ofincreased communication apprehension

vention of the Eastern Communication Assn.,Boston, 1978.

13 James C. McCroskey and Virginia P:Richmond, "Community Size As a Predictor ofDevelopment of Communication Apprehension:Replication and E."Ctension," CommunicationEducation, 27 (Sept. 1978), 212-19.

14 Phillips.15 William G. Davey, "Communication Per-

formance and Reticence: A Diagnostic CaseStudy in the Elementary Classroom." Paperpresented to the annual convention of the\Vestern Speech Communication Assn., Seattle.1975.

16 Daly and Friedrich.17 Porter.

124-COMMUNICA TION EDUCATION

in some children. Preliminary interviewswith elementary school teachers clearlyindicated that such teachers could

recognize widely differing levels ofcommunication apprehension amongyoung children, even at the point whenthey first enter the school environment.This information suggests at least twopossibilities: (1) Communication ap-prehension levels are well establishedbefore entering the school environment.Thus, previous research which hasidentified correlations between com-munication apprehension and recalledschool experiences may infer falsecausation-the communication apprehen-sion may generate either the experienceor the memory of the experience, forexample. (2) Communication apprehen-sion levels are only weakly establishedbefore entering the school environmentand, thus, are subject to substantialmodification (either higher or lower) asa result of school experiences.

No extended longitudinal researchon communication apprehension amongchildren has been reported. Were suchdata available, at least one of the abovepossibilities could be ruled out. Eitherthe communication apprehension levelremains relatively constant throughchildhood and into adulthood or it

doesn't. Unfortunately, we do not knowthe answer to this question. Only sub-stantiallongitudinal reseachcan providea definitive answer. Anecdotal evidence,

however, is suggestive of a possibleanswer. Numerous, but unsystematic,discussions with college and adult stu-dents, who have been exposed to theliterature on communication apprehen-sion, have resulted in many commentsto the effect that "I used to be a highcommunication apprehensive, but I amnot anymore." Interestingly, oppositestatements have been totally absent,perhaps because no one ever becomes

more apprehensive, or because people

who are now high apprehensives, butwere not previously, prefer to keep thatinformation to themselves.

The anecdotal evidence, then, suggeststhe possibility that communication ap-prehension levels can change, at least inone direction-down. Hard data, how-ever, suggest that normative communi-cation apprehension levels in largesamples of college freshmen, otheradults, and even senior citizens areessentially the same.1S Unfortunately, nocomparable normative data for largesamples of younger children are yetavailable. One of the purposes of thepresent research, therefore, was to gen-erate such normative data for school-agechildren, K-12.

Because of the limited research on

communication apprehension amongyoung children, formulation of a priorihypotheses was difficult. The only pre-vious research which was suggestive wasdirected toward speech fright ratherthan communication apprehension. Thisresearch employed observer ratings,introspective tests, and physiologicalmeasures (GSR). The results indicateda substantial increase in speech fright be-tween third graders and sixth graders.19On this tenuous base, the followinghypothesis was advanced:

H1: Mean communication apprehension scoresof children in grades K-3 are lower thanmean communication apprehension scoresof children in grades 4-6. 7-9, and 10-12.

18 See, for example, James C. McCroskey,"Measures of Communication-Bound Anxiety,"Speech Monographs, 37 (1970). 269-77; DennisL. Moore, "The Effects of Systematic Desensitiza-tion on Communication Apprehension in anAged Population," Thesis Illinois State Univ.1972; and Raymond L. Falcione, James C.McCroskey, and John A. Daly. "Job SatisfactionAs a Function of Employees' CommunicationApprehension, Self-Esteem. and Perceptions ofTheir Immediate Supervisor," in Communica-tion Yearbook I, ed. Brent D. Ruben (NewBrunswick, N.J.: Transaction. 1977), 363-76.

19 For a summarv of this research, seeLawrence R. Wheeless, "Communication Ap-prehension in the Elementary School," SpeechTeacher, 20 (Nov. 1971), 297-99.

INSTRUMENT DEVELOPMENT

The primary instrument for measuringCA in previous research is the PersonalReport of Communication Apprehen-sion (PRCA). This instrument has beendemonstrated to be both reliable andvalid.2O However, the PRCA was de-veloped for, and has been used almostexclusively for, measuring the CA ofhigh school students and adults. Thelanguage level of the PRCA limits itsuse with young children. Thus, thePRCA could not be used with confidence

for all of our student samples.

Recently, ,a measure designed to beadministered to preliterate childrenhas been reported. This instrument, theMeasure of Elementary CommunicationApprehension (MECA), is framed inlanguage appropriate for younger chil-dren and uses various forms of smilingand frowning faces for response op-tions.21 This instrument was considered

appropriate for our younger studentsamples, but, because of the responseformat, the measure was not considered

appropriate for junior and senior highschool student samples.

As a result, a preliminary study wasdesigned to develop a measure of CAthat could be administered to all ele-

mentary and secondary school students,regardless of age level. Twenty itemswere written at a level believed to beunderstandable for children in the

preliterate stage of development as wellas older students. The instrument was

administered orally to children belowseventh grade level. For the responseformat employed, see Table 1. The

20 ~IcCroskey, "Measures of Communication.Bound Anxietv"; and idem. "Validity of .thePRCA As an' Index of Oral CommunicationApprehension:' Communication Monographs,45 (Aug. 1978). 192-203.

::1 Karen R. Garrison and John P. Garrison.":\Icasurement of Communication .~pprehensionamong Children:' Paper presented at theannual cOlI\'ention of the ICA. Berlin. 'VestGermanv, 1977.

COMMUNICATION APPREHENSION-l25

instrument was administered to 2,228students in five school districts. The

sample included the following numbersof subjects at the various grade levels:K-3, 248; 4-6, 462; 7-9, 762; and 10-12,756. The instrument was also ad.

ministered to 875 college students.

The data for each of the five subjectgroups were submitted to factor analysisto determine whether the instrument

was unidimensional, as presumed initial-ly. The results for each subject groupclearly indicated the presence of two in-terpretable factors, although the factorswhen subjected to oblique rotation werecorrelated (from .39 to .57, dependingon sample). The first factor could belabeled the "fear or anxiety" dimension,the second the "shyness or verbosity"dimension. Fourteen of the items had

their primary loading on the first factor,six were loaded primarily on the secondfactor. Although the magnitude of theloadings varied somewhat among the fivesamples, the general pattern was highlyconsistent.

Since this preliminary study wasfocused on the development of an in-strument (as it turned out, two in-struments), reliability and validity wereprimary concerns. The obtained re-liabilities for the two dimensions (split-half, internal consistency) for eachgrade-level grouping are reported inTable 3. Although the reliabilities weregenerally within a satisfactory range atall grade levels, it was decided to try toimprove the instruments by addingitems. Consequently, an eighteen-itemversion of the "fear" scale and a sixteen-

item version of the "shyness" scale wereprepared and administered to 705 col-lege students. Factor analysis withoblique rotation indicated the presenceof the two dimensions, with a correla-tion of .55 between the dimensions.Four of the items on the "fear" dimen-

sion and two on the "shyness" dimension

126-COMMUNICATION EDUCATION

were dropped for subsequent use sincetheir loadings split between the factors.The resulting two scales, named the

Personal Report of CommunicationFear (PRCF) and the Shyness Scale (SS),are reported in Tables I and 2. The

TABLE I

PERSONALREPORT OF COMMUNICATIONFEA1l (pRCF)

DIRECTIOSS: The following 14 statements concern feelings about communicating with otherpeople. Please indicate the degree to which each statement applies to you by circling yourresponse. Mark "YES" if you strongly agree, "yes" if you agree, "?" if you are unsure, "no" ifyou disagree, or "NO" if you strongly disagree. There are no right or wrong answers. Workquickly; record your first impression.

SCORING: YES = I, yes = 2. ?= 3,no = 4, NO = 5.To obtain the score for the PRCF, complete the following steps:1. Add the scores for the following items: 2, 3, 4. 6, 8, 9, and 12.2. Add the scores for the following items: 1,5,7, 10,11, 13, and 14.3. Add 42 to the total of step 1.4. Subtract the total of step 2 from the total of step 3.The score should be between 14 and 70.

TABLE 2

SHYNESSSCALE (SS)

DIRECTIOSS: The following 14 statements refer to talking with other people. If the statementdescribes you very well, circle "YES." If it somewhat describes you, circle "yes.» If you are notsure whether it describes you or not, or if you do not understand the statement, circle "?" Ifthe statement is a poor description of you, circle "no," If the statement is a very poor descriptionof you, circle "NO." There are no right or wrong answers. Work quickly; record your firstimpression.

SCORING: YES = I, yes = 2, ? =3, no = 4, NO =5.To obtain the score for the SS. complete the following steps:1. Add the scores for the following items: 2, 3. 5, 8, 10, 13, and 14.2. Add the scores on the following items: 1,4,6,7,9,11, and 12.3. Add 42 to the total of step 2.4. Subtract the total of step I from the total of step 3.The score should be between 14 and 70

YES yes ? no NO 1. Talking with someone new scares me.YES yes ? no NO 2. I ook forward to talking in class.YES yes ? no NO 3. I like standing up and talking to a group of people.YES yes ? no NO 4. I like to talk when the whole class listens.YES yes ? no NO 5. Standing up to talk in front of other people scares me.YES yes ? no NO 6. I like talking to teachers.YES yes ? no NO 7. I am scared to talk to people.YES yes ? no NO 8. I like it when it is my turn to talk in class.YES yes ? no NO 9. I like to talk to new people.YES yes ? no NO 10. When someone asks me a question, it scares me.YES yes ? no NO 11. There are a lot of people I am scared to talk to.YES yes ? no NO 12. I like to talk to people I haven't met before.YES yes ? no NO 13. I like it when I don't have to talk.YES yes ? no NO 14. Talking to teachers scares me.

YES yes ? no NO I. I am a shy person.YES yes ? no NO 2. Other people think I talk a lot.YES yes ? no NO 3". I am a very talkative person.YES yes - no NO 4. Other people think I am shy.YES yes ? no NO 5. I talk a lot.YES yes ? no NO c.. I tnd to be very quiet in class.YES yes no NO i. I don't talk much.YES yes ? no NO R. I talk more than most people.YES yes ? no NO 9. I am a quiet person.YES yes no NO 10. I talk more in a small group (3.6 people)

than other people do.YES yes ? no NO II. Most people talk more than I do.YES yes ? no 1';0 12. Other people think I am very quiet.YES yes ? no NO 13. I talk more in class than most people do.YES yes ? no NO 14. Most people are more shy than I am.

reliabilities obtained for the follow-upcollege sample, reported in Table 3, werein excess of .90 for both scales. Con-sequently,. these scales were used in thelater study.

TABLE 3RELlABILlTIES OF PRCF AND 55 INSTRUMENTS

To assess validity of the new instru-ment (ultimately two instruments), theK-3 and 4-6 samples in the preliminarystudy were administered the MECA, andall of the samples were administered thePRCA-Short Form.:!:! Presuming thevalidity of the PRCA and MECA scales,although the case for the latter is yetto be argued empirically, the obtainedcorrelations between the two new scales

and the PRCA/:\IECA provided us withconcurrent validity coefficients for thenew scales. The raw validity coefficientsand disattenuated coefficients are re-

ported in Table 4, which indicates thatthe PRCF consistently obtained higher,

:!:!For a copy of this form. see McCroskey,"Validity of the PRCA."

COMMUNICATION APPREHEN5ION-127

usually very much higher, validity co-efficients than the SS. When using thePRCA as the criterion, the PRCF gen-erated very high validity coefficients.Given the cibvious language-level validityproblems with the 'PRCA for preliteratechildren, the coefficient at the K-3 level

was particularly encouraging.The validity coefficients for the SS

suggest this scale is not measuring CAas well as the PRCF. Rather, it likely istapping communication behavior relatedto CA, but not solely caused by CA.While this type of conceptual distinctionbetween CA and shyness has beenalluded to previously,23 this is the firstempirical indication in the literature.Consequently, it was decided that it wasimportant to include both measures inthe main study of school children, eventhough no hypothesis about shynessoriginally had been proposed.

PROCEDURES AND RESULTS:

STUDENTS AS SUBJECTS

Subjects in this study included 5,795elementary and secondary school stu-dents enrolled in sixty-seven school dis-tricts in three states. The breakdown by

grade level was as follows: K-3, 1,252;4-6, 1,741; 7-9, 1,752; and 10-12, 1,050.All stUdents were administered the

:!3McCroskey. "Oral Communication Appre-hension."

TABLE 4

VALlDITYCOEFFICIENTSFOR PRCF AND55

Raw

.32

.16

.65 .

.28

Validity CriterionPRCA

DisattenuatedMECA

Disattenuated Raw

.50

.24

.93

.34

.37

.33

.67

.28

.74

.41

.80

.53

.71

.40

.63

.53

.98

.48

.98

.49

.95

.60

.84

.45

AdministrationPreliminary

Main StudyStudiesGrade Level PRCF SS PRCF SS

K-3 .60 .69 .70 .824-6 .62 .84 .79 .867-9 .71 .86 .84 .9010-12 .84 .89 .90 .92College .84 .90College .91 .92

(folIow-u.I1

Grade Level Scale

K-3 PRCF55

4.6 PRCF55

7-9 PRCF55

10-12 PRCF55

College PRCF55

l28-COMMUNICA TION EDUCATION

PRCF and 55 by their regular classroominstructor. For those students undergrade 7, the scales were administeredorally. Obtained reliabilities for thescales are reported in Table 3. The -datawere submitted to single-classification(four levels) analyses of variance. Ob-tained means for both scales are reportedin Table 5.

TABLE 5

MEAN PRCF AND SS SCORES BY GRADE-LEVELGROUPING

Scale

a score range = 14-70. High score = high fear.b score range = 14-70. High score = highlytalkative.

The results indicated a significanteffect for grade level on both the PRCFscores (F = 87.43, P <.0001) and the 5Sscores (F = 69.21, P <.0001). As notedin Table 5, the mean PRCF scores were

yirtually identical for all grade levelsexcept K-3. As hypothesized, the scoresfor the K-3 group were significantlylower. Note that the hypothetical meanof the PRCF scale is 42; thus, the ob-tained differences are not likely to be afunction of lower reliability or validityfor the younger subjects. Random re-sponses would cancel themselves out toproduce higher rather than lower scoresfor the younger groUp.~4 Although no

~-IThis only applies to randoI/! error. Afterreading an earlier draft of this paper. John Dalyof the University of Texas raised the possibilitythat there may be a systematic error in .thescores. As he correctly noted, in the typicallower-grade classroom there is a great deal ofinieraction, but this tends to be reduced inhigher grades as classrooms become more struc-tured with rules regarding interaction. Dalyasks if it is possible that the students completedthe instrument with their primary focus on theclassroom in which thev were students. If so,the scores mav simply' reflect their classroomexperiences rather than general dispositions. 'Vecannot completely discount this possibility.However, the items on the measures, par-

hypothesis was framed for the 55 scores,the pattern of means on this scale in-dicates that the younger children aremore verbal (less shy) than their olderpeers. Again the -devial!t group is thechildren in grades K-3.

Although the sample sizes are sub.stantially lower, and thus the means lessstable, it is interesting to consider thePRCF scores at each grade level duringthe early years. The means are asfollows: K, 25.8; 1, 33.0; 2, 32.9; 3, 33.0;4, 34.7; and 5, 37.1. Since our subjectscompleted the measures during the firstsemester of their reported grade, eachgrade level presumably reflects impactof the previous year. Notable changesappear to occur during kindergartenand grades 3 and 4. Means from thatpoint on (to and including college) re-main virtually identical.

DISCUSSION

Although we had little basis for ourprimary hypothesis, the results of theresearch provide considerable supportfor that hypothesis. Children in lowerelementary school (K-3) report lowerlevels of CA than do children in upperelementary school (4-6), junior highschool (7-9), or high school (10-12). Thebiggest change appears to occur inkindergarten during the child's firstexposure to the school environment. An-other substantial increase appears tooccur during grades 3 and 4. Thus, be-fore puberty, CA norms are achievedthat remain relatively stable through allsubsequent age groups.

While these normative data clearlyindicate that some factor or factors re-sult in increased CA among childrenwhile they are attending elementaryschool, they do not establish that anyelement in the school is the causal agent.

ticularly the Shyness Scale, appear to be sogeneral as to preclude this as the sole explana-tion of the obtained results.

Grade Level PRCFa SSb

K-3 32.5 43.94-6 36.5 40.67-9 36.5 39.610-12 36.4 39.4

Indeed, biological and/or social ma-turational elements unrelated to the

school may account for all of thisvariation. Nevertheless, until established

otherwise, we should continue to suspectthe school environment as a potentialcausal agent for increased levels of CAin children.

Given that we continue to suspect theschool environment, and given that wemay wish to alter that environment soas to remove any negative influence, ourfocus on the school environment shouldbe narrowed to the factors that have the

highest probability of impact, eithernegative or positive. Although thenumber of variable elements in the

school environment are nearly limitless,we believe we can narrow our concern

to three: (1) the physical facilities, (2)the peer environment, and (3) theteachers.

Although school facilities have anobvious relationship to the type of teach-ing which can occur and to the com-munication environment of the child,

we discounted this factor as a possiblecause of the differences in norms. Ourdata were obtained from schools with

virtually every type of facilities imagin-able-old and new buildings, large andsmall buildings, open and traditionalclassrooms, urban and rural schools, andso on. Supplementary analyses, whichwere conducted to provide feedback tothe teachers who assisted us in this

project, gave no indication of any pat-tern of differential results from schoolto school or area to area. In short, eventhough physical facilities may have amajor impact on classroom communica-tion, they are an inadequate explanationof the differential norms we observed,

The impact of peers cannot be dis-counted. Many children encounter theirfirst real contact with peers when theyenter kindergarten, and this contact mayserve to increase a child's inhibitions

COMMUNICATION APPREHENSION-129

and, hence, the CA level. Future re-search should compare changes in CAlevel of children who have had little

peer contact before kindergarten withthose children who have. Perhaps chil-dren with early, extensive peer contactare immunized against developing CA,or they may simply develop CA earlier.In any event, such research can give ussome insight as to whether children's CAis being impacted purely by peer contactor whether some other school-relatedelement is influential.

The third element in the school en-vironment, teachers, also is difficult todiscount as a causative factor. Casestudies and other anecdotal evidence

regularly highlight the impact of a giventeacher's behavior on children. At this

point, unfortunately, we are forced tospeculate as to the types of teacherbehaviors that might lead to highercommunication apprehension amongschool children. No direct observationalevidence exists of teachers' behavior with

children who develop higher or lowerCA. Similarly, some types of teachersmay be more or less likely to increasechildren's CA levels than other types;but, we may only speculate at this pointas to what those types may be.

Our extensive experience workingwith in-service elementary and secondaryteachers caused us to suspect one elementthat may be a contributing factor to thekinds of normative CA level increases

observed in the above study. Althoughprevious research has indicated thatnormative levels of CA among elemen-tary and secondary teachers approximatethose of college student and adultgroups, we have observed that an un-usually large number of elementaryteachers have high CA; at the same time,a very low proportion of secondaryteachers have high CA. If, indeed, thereare significant numbers of high CAteachers in the lower elemen tary grades,

l30-COMMUNICA TION EDUCA TIO:\

this may provide at least a partial ex-planation for the increases in children'sCA levels in these grades.

Previous research has established thatpeople with high CA communicate lessfrequently and in different ways thanpeople with low CA. Although none ofthis research has involved direct ob-servation of teacher communication be-

havior, teachers' behavior is as likely tobe impacted by CA as is behavior ofpersons in other occupations. For ex-ample, high CA teachers may talk lessin the classroom, thus providing modelsof quietness that could be reinforced ina child's mind. Similarly, the high CAteacher may be more likely to reinforcewithdrawn behavior than would another

teacher. Thus, if a significantly largerproportion of teachers in lower elemen-tary grades are high CA's than in othergrades, a possible explanation for thedifferential norms we observed would

center on those high CA teachers andtheir behavior in the classroom. On the

basis of these speculations, we posed thefollowing hypotheses:

H,: There is a higher proportion of teacherswith high CA in the lower elementarygrades (K.4) than at other grade levels.

H.: There is a higher proportion of teacherswith high CA in the lower elementarygrades (K-4) than there are teachers withlow CA in those grades.

Hypothesis I is the expectation based onour experience with in-service teachers.If it were not confirmed, our speculationconcerning the impact of high CAteachers on children obviously would beout of order. If there were the same

number of high CA teachers at eachlevel, something other than CA level ofteacher would have to account for thechildren's increased CA during the earlyelementary years. Hypothesis 2 also teststhe heart of our speculation. If therewere an equal number of high and lowCA teachers in the elementary schools,

their impact on the norms for a largenumber of students would be expectedto cancel each other out. Thus, for our

concern about the impact of high CAteachers on the development of CA inchildren to be worthy of additionalstudy, both of these hypotheses must beconfirmed. Any other result wouldindicate our concern to be misplaced.

PROCEDURE AND RESULTS:

TEACHERS AS SUBJECTS

The sample studied included 573 in-service elementary and secondary schoolteachers from fifty-seven school districtsin five states. All of the subjects wereenrolled in graduate instructional com-munication classes taught in ten separatelocations. Subjects completed the re-quired instruments for the study on thefirst day of class before any discussionof the subject matter.

The subjects completed the PRCA-Short Form23 and responded' to twoother questions. The first asked them toindicate the grade level at which theytaught, and the second asked them toindicate the grade level at which theywould prefer to teach. Approximatelytwenty-one percent of the teacherstaught at multiple levels (music teachers,coaches, special education, etc.) andcould not be classified into categoriesappropriate for this study. Consequently,they were classified as "others." In ad-dition. because only S of the 5i3 teacherstaught at the kindergarten level, theywere added to the "others" classification.

To draw conclusions concerning kinder-

garten teachers from such a smallsample was inappropriate, althoughthree of these teachers reported high CAwhile none reported low CA.

Subjects who scored more than onestandard deviation above the mean for

23 McCroskey, "Validity of the PRCA."

the sample were classified as high in CA,those scoring more than one standarddeviation below the mean were classifiedas low in CA, and the remainder wereclassified as moderates. The estimated

reliability (split-half) of the PRCA was.90. As indicated in Table 6, creating the"others" classification did not distort

the remainder of the categories, since thedistribution of subjects into CA levelsfor this group was what should be ex-pected from a normal distribution.

Tables 6 and 7 report the results ofthe study and show that the proportionof teachers with high CA was muchgreater in grades 1-4 than at other levels,as was the proportion of teachers whowould prefer to teach in those grades.Similarly, the ratio of high to low CAteachers in grades 1-4 was approximately4-1 for people actually teaching in those

COMMUNICATION APPREHENSION-131

grades and 3-1 for people preferring toteach in those grades. The chi-squaresfor the interaction between grade leveland CA level were 34.53 (p < .00I; C =.27) for actual teaching level and 30.38(p <.001; C = .24) for preferred teach-ing level. Two sample-pro portioI' testsfor the first hypothesis indicated thatthere was a significantly higher propor-tion of high apprehensives teaching ingrades K-4 than in either grades 5-9(Z = 4.32, P <.0001) or grades 10-12(Z = 4.02, P <.0001). In addition, theproportion test for the third hypothesisindicated that there was a significantly

higher proportion of teachers with highCA in grades K-4 than there wereteachers with low CA in those grades

(Z = 12.10, P <.0001). Both hypotheseswere supported by these results.

TABLE 6

FREQUENCY AND PERCENTAGEOF TEACHERS 'WITH COMMUNICATION ApPREHENSION

BY GRADE LEVEL TAUGHT

a frequency of teachers at grade levelb percentage of teachers at grade level

TABLE 7

FREQUENCY AND PERCENTAGEOF TEACHERS WITH COMMUNICATION ApPREHENSION

BY GRADE THEY WOULD PREFER To TEACH

a frequency of teachers at grade levelh percentage of teachers at grade level

A.pprehensionLevelGrade Level Low Moderate High Total

1-4 13a 109 52 174(.07)b (.63) (.30)

5-9 30 117 18 165(.18) (.71) (.11)

10-12 23 73 10 106(.22) (.69) (.09)

Others 23 84 21 128(.18) (.66) (.16)

Preferred A.pprehension LevelGrade Level Low Moderate High Total

1-4 23a 147 63 233

(.10)b (.63) (.27)5.9 25 103 16 144

(.17) (.72) (.11)10-12 28 80 11 119

(.24) (.67) (.09)Others 13 53 11 77

(.17) (.69) (.14)

132-COMMUNICATION EDUCATION

DrsCussION

Care must be taken not to over interpretthe positive results of this study. Theseresults suggest that future research whichinvestigates the behavior of highlyapprehensive elementary school teachersmay uncover a contributing causalelement leading to increased CA amongsome elementary school children. Thisstudy does not establish that such arelationship exists, only that it is aplausible hypothesis worthy of futureinves tigation.

On the basis of the results of thisseries of studies we can draw severalconclusions with reasonable confidence.

First, it is possible to measure CA levelsof children, even preliterate children.Second, normative levels for CA changemarkedly during kindergarten andthird and fourth grade years. Finally,some factor or combination of factors

causes increases in CA during the earlyelementary school years that are sus-tained into adulthood. At this point wecannot confidently state what the factor(factors) is, but peer contact and teacherbehavior may be critical causativeagents.

Two lines of research need to be

pursued. First, we need to obtain dataon kindergarten and first grade studentswho have differential degrees of peercontact before entering school. If these

students differ in their level of CA, im-

plications for intervention would beclear. We could either implementsystems which would increase early peercontact, if such contact is found to beassociated with lower CA, or work to

modify the impact of early contact, ifsuch contact is found to be associated

with higher CA. If no meaningful differ-ences are observed, of course, we wouldthen be able to discount peer influence(at least of the type suggested here) inthe school environment as a causalelement in increased CA. .

The second line of research that isindicated involves direct observation ofteachers known to be high and low inCA. We need to determine whether suchteachers behave differentially towardstudents, and, if so, whether these dif-ferential behaviors can be associatedwith increased or decreased CA in stu-dents. Such research would not onlyprobe the cause of the normative changeswe have observed in this series of studies,but also would isolate potential teacherbehaviors that could be modified to

enhance positive growth in young chil-dren. Such studies will be difficult and

time consuming. However, the resultsof the present studies indicate that theyhave high potential for producingvaluable insights into the developmentof CA in children and the preventionof such development.

"[I am] happily too busy doing science to have timeto worry about philosophizing about it.rr - Arno Penzios

(Nobel Prizewinner. 1978)