The Communication Anxiety Inventory:

Validation of State-
and Context-Communication
Apprehension
Steven Booth-Butterfield and Malloy Gould
Research reported in this paper investigates the reconceptualization
(McCroskey, 1984) of communication apprehension (CA), focusing on the
relationship between context-CA and state-CA. The validity of the recon-
ceptualization can be enhanced by demonstrating that state-CA (level of fear
and anxiety) is systematically related to context-CA (predisposition). Two
research studies using a new operational measures of state-CA and context-
CA, the Communication Anxiety Inventory, offer empirical support for the
relationship between context-CA and state-CA. Results show that context-
CA explains a minimum of 50% of the state-CA variance. Conclusions
suggest that level of fear or anxiety varies closely with context-CA and that
the new scales have strong predictive validity.

KEY CONCEPTS Communication apprehension, communication anxiety

inventory, form state, form trait, context-communication apprehension,
state-CA communication apprehension, hierarchical multiple regression,
canonical correlation.

STEVEN BOOTH-BUTTERFIELD (M.A., Central Missouri State University, 1979) is a

doctoral candidate in the Department of Speech Communication, West Virginia University,
Morgantown, WV 25606. MALLOY GOULD (Ph.D., Ohio State University, 1973) is an Assistant
Professor in the Department of Communication, Central Missouri State University, Warrens-
burg, MO 64093.

T
he latest evolution of communication apprehension (CA) has created a
rich, new source of conceptual and empirical tools (McCroskey, 1981;
1984). The conceptual tools now available have expanded from the
original "broadly based anxiety related to oral communication" (McCroskey,
1970, p. 270) to states, traits, contexts, person-group and situational factors.
Of greatest concern here are the conceptualizations of trait-CA, context-
CA, and state-CA. Currently, trait-CA and context-CA are seen as relatively
enduring, personality orientations or predispositions to communication in
general or to specific communication settings. A high trait-CA person is
assumed to fear most communication situations, while a high context-CA
person is likely to fear only specific contexts, such as speaking or interpersonal
settings. By contrast, state-CA is viewed as a "here-and-now" anxiety

194 Communication Quarterly Vol. 34, No. 2, Spring 1986, Pages 194-205
response of a person in any communication situation. Most simply, trait- or
context-CA is a predisposition, while state-CA is the actualj response of fear
or anxiety.
One immediate concern is to demonstrate and verify empirically, the
theoretical relationships posited under the new model of CA. Research
described in this report focuses upon the empirical relations;hip between the
various context-CAs and state-CA.

Context-CA
The newest conceptual tool, context-CA, is defined as "... a relatively
endurring, personality-type orientation toward communication in a given
context" (McCroskey, 1984, p. 16). The intent of this definition leads us to
conclude that a context-CA is also trait-like. It is assumed that, like trait-CA, a
person's context-CA will be stable across time when within the particular
context. Thus, we expect that a person who experiences high context-CA in
public speaking situations to report a high level of fear or anxiety in most
public speaking settings and especially so when compared to those who
report low context-CA in public speaking situations. ;
Past research in this area would support that contention. For example, the
findings of Behnke and others have consistently shown that higher public
speaking context-CA scores are positively and strongly correlated with heart
rate at confrontation and self-reported state-CA (Behnke |& Beatty, 1981;
Behnke, Carlile, & Lamb, 1974; Behnke & Carlile, 1971; Lamb, 1969).
Research on other context-CA's is limited. The interpersonal or dyadic
context-CA has received some attention (Freimuth, 1976; Richmond, 1978,
for example); findings in this context are similar to the findings reported in the
public speaking context. The group and meeting contexts posited by McCros-
key, to our knowledge, have not yet been investigated with designs, opera-
tionalizations, or models that are consistent with the current conceptualiza-
tion.

State-CA
Some confusion over labeling appears to exist in the C A literature. Most of
it revolves around the terms "state-anxiety," and "communication apprehen-
sion." Both of these terms have used to describe the same phenomena. The
standard definition of communication apprehension is, "... level of fear or
anxiety associated with either real or anticipated communication with another
persons." (McCroskey, 1984, p. 13). We see little difference between this
definition and the term "state-anxiety" as employed by Spielberger (1966).
His definition of state anxiety is, "consciously perceived feelings of apprehen-
sion and tension, accompanied by or associated with activation or arousal of
the autonomic nervous system" (p. 17).
Thus, whether we call it "communication apprehension" or "state-
anxiety," we are discussing the same phenomena. It is the here-and-now
response of a person in some situation. The response has cognitive, affective,
and physiological elements. McCroskey clearly distinguishes this response

Communication Quarterly Spring 1986 195

from the predispositional qualities which mark the concepts of trait-CA,
context-CA, and person-group CA.
We think that the term "state-CA" more clearly combines the here-
and-now response quality expressed in Spielberger's work and the communi-
cation-specific nature inherent in McCroskey's research. This combination is
not unique in the literature. Richmond (1978) and McCroskey and Beatty
(1984) have employed it.
Typically, the State Anxiety Inventory of Spielberger, Gorsuch, and
Lushene (1970) is employed to measure state-CA, but copyright concerns
make its use somewhat problematic at times. It is also appropriate and valid to
measure the state-CA response with a variety of physiological devices. Again,
the work of Behnke would illustrate this approach.

The Context-CA And State-CA Relationship

There is a clear link in the reconceptualization of CA between state-CA
and trait- and context-CA. The trait- and context- function as predisposing
agents which increase the likelihood of a state response and increase the
elevation of that response. Thus, people who report high trait-CA or high
context-CA, should experience higher levels of fear or anxiety. However, as
we survey the earlier literature of CA, we can find few examples where the
"level of fear or anxiety" was actually shown to be systematically related to
varying trait- or context-scores. (Exceptions would include the previously
cited works of Behnke and his associates; Freimuth, 1976; and Richmond,
1978).
We can cite a body of research that tested the relationship between CA
and academic achievement (Powers & Smythe, 1980), instructional prefer-
ence (McCroskey & Anderson, 1976), and a variety of personality characteris-
tics (McCroskey, Daly, and Sorensen, 1976; Lustig, 1974). (Consult McCros-
key, 1977 for a full review.) Within this body, we cannot find examples which
directly show that trait-or context-CA scales predict state-CA. Thus, the
central conceptual relationship has been tested in a highly inferential and
indirect manner. Only recently has a study appeared that directly tests this
proposition (McCroskey & Beatty, 1984; we compare our research with this
study in some detail in our DISCUSSION section below).

Testing The State-CA/Context-CA Relationship

Since 1979 and independently of McCroskey, we have been developing
our own measurement of trait-, context-, and state-CA. The result is a 41
item self-report scale. Our scale, the Communication Anxiety Inventory
(CAI), is composed of two separate inventories which we label as Eorm Trait
and Form State. Form Trait is remarkably similar to the latest version of the
PRCA. A comparison of the face and content validity of the two different
inventories suggests that they may be parallel forms. Form Trait is composed
of 21 self-report items that measure trait-CA and three context-CAs. (See
Figure 1). The three contexts we chose to assess correspond to the dyadic,
group, and meeting factor. Form Trait employs a 4 point response format

196 Communication Quarterly Spring 1986

FIGURE 1 Form Trait
Directions: This inventory is composed of 21 statements that describe various communication
events. You are asked to respond in terms of how you genera///feel about these events. Please
mark your response in the appropriate blank on the answer sheet. Be sure to give the response
that best describes how you generally iee\.

Almost Never Sometimes Often Almost Always

1 2 3 4 !

1. * I think I communicate effectively in one-to-one situations.

2. My heart beats faster than usual when I speak out in a small group meeting.
3. * I enjoy speaking in public.
4. I avoid talking with individuals I don't know very well.
5. I think I make a poor impression when I speak at a small group meeting.
6. I feel disappointed in myself after speaking in public.
7. * I enjoy talking with someone I've just met.
8. * My body feels relaxed when I speak during a small group meeting.
9. I avoid speaking in public if possible.
10. My body feels tense when I talk with someone I don't know very well.
11. * I speak out during small group meetings. '
12. I am terrified at the thought of speaking in public.
13. My heart beats faster than usual when I talk with someone I've just met.
14. * I enjoy talking at a small group meeting.
15. * I make a good impression when I speak in public.
16. * I would like to have a job that requires me to talk often on a one-to-one basis.
17. I feel disappointed in my efforts to communicate at a small group rrieeting.
18. My body feels tense and stiff when I speak in public.
19. When conversing with someone on a one-to-one basis, I prefer to listen rather than to
talk. I
20. I avoid talking during small group meetings. '
21. * I look forward to speaking in public.

Scoring: reverse "*" items then

Dyadic = sum of 1, 4, 7, 10, 13, 16, and 19.,

Small Group = sum of 2, 5, 8, 11, 14, 17, and 20.
Public Speaking = sum of 3, 6, 9, 12, 15, 18, and 21. ]

ranging from "almost always" to "almost never." Thus, scores may range from
21 (low CA) to 84 (high CA); each context score ranges from 7 to 28.
We administered From Trait to several samples of university and high
school students over the course of a four year period. (N = 754; males
composed 47.2% of sample with a mean age of 21.3 and a standard deviation
of 6.4; the sample was predominately white, small town, and midwestern). On
each occasion we analyzed descriptive statistics, reliability estimates, and
factor structure. The results were consistent and psychom^trical adequate.
Reliability estimates for all samples showed: Cronbach's alpha=.898, split
half=.919, and lambda ranges from .856 to .919. Reliability estimates for the
context scores showed alpha ranges from .654 (dyad) to .887 (public) with
split half estimates and lambda's varying similarity. (See table 1 for descriptive
statistics.) Factor analysis (principle factors, multiple squared correlations in
the diagonal, oblique rotation, and eigenvalues greater than one) consistently

Communication Quarterly Spnr\g}986 ! 197

TABLE 1 Form Trait Descriptive Statistics and Reliabilities
Context Trait
IP GP PS CA
mean 12.996 15.033 18.874 46.903
std. dev. 3.204 4.235 5.174 10.495
std. err. .117 .154 .188 .382
alpha .654 .846 .887 .898
split half .665 .855 .894 .919
lower bound lambda .561 .725 .761 .919
upper bound lambda .660 .850 .891 .856
N = 754.

revealed a four factor solution—the three contexts and a physiological

factor.
Form State, by contrast, is a 20 item measure that assesses a person's
anxiety response in any communication situation. Current published research
typically employs Spielberger's x-1 scale. However, copyright concerns
noted by McCroskey (1984) make use of this scale somewhat problematic.
Research presented in this report will show that Form State appears to
function as well as X - 1 , measures state-CA in any communication setting, and
avoids the copyright problems. Form State is based directly upon the model
provided by Lamb (1972). It employs a 4 point response format from "not at
all" to "very much so" (See Figure 2).
We administered Form State to a sample of undergraduate students
(N = 163; females composed 47.8% of the sample; the mean age was 20.34).
The students were told to imagine that they had just presented a speech
before their class and to respond to Form State on the basis of how they would
feel during the imagined speech. Reliablity estimates showed: alpha=.912,
split half=.921, and lambda ranges from .865 to .927. Descriptive statistics
showed a mean of 44.06, a standard deviation of 11.07, standard error of .867.
A factor analysis following the same procedures noted earlier showed that a
one factor solution from the initial factor matrix was the most meaningful. All
loads were above .35. Sixteen of the items loaded above .45.
We administered Form Trait and Form State with the same imaginary
public speaking instructions to a group of high school students (N=64). We
then correlated the context-CA scores with the Form State scores. The
obtained Pearson correlations were: DYAD = .357; GROUP = .574; and
PUBLIC SPEAKING = .711. Each of the correlations were statistically signifi-
cant beyond the .05 level. Given the assumption that the contexts should be
moderately intercorrelated (McCroskey, 1984), the obtained correlations are
not surprising. However, the PUBLIC SPEAKING context-CA is clearly most
strongly related to the state-CA when compared to the other context CA's.
The " g " effect index (Cohen, 1977) shows a medium effect (.31) for the
difference between the PUBLIC SPEAKING and the GROUP correlations. The
" g " index for PUBLIC SPEAKING versus the DYAD correlation is large (.51).
Since the imaginary instructions told the subjects to respond on the basis of a
public speaking situation, the correlations are expected and predictable.

198 Communication Quarterly Spring 1986

FIGURE 2 Form State
Directions: The following items describes how people communicate in| various situations.
Choose the number from the following scale that best describes how you felt during the
communication experience you just completed.

Not At All Somewhat Moderately So Very Much So

1 2 3 4

1. I felt tense and nervous.

2. I felt self-confident while talking.
3. While talking, I was afraid of making an embarrassing or silly sip of the tongue.
4. I worried about what others thought of me.
5. I felt calm when I was talking.
6. I felt ill at ease using gestures when I spoke.
7. I could not think clearly when I spoke.
8. My listener(s) seemed interested in what I had to say.
9. I felt poised and in control while I was talking.
10. My body felt tense and stiff while I was talking.
11. My words became confused and jumbled when I was speaking.
12. I felt relaxed when I was talking.
My fingers and hands trembled when I was speaking.
rl I felt I had nothing worthwhile to say.
14. I had a "deadpan" expression on my face when I spoke.
15. I found myself talking faster or slower than usual.
16. While speaking, it was easy to find the right words to express myself.
17. I felt awkward when I was talking.
18. My heart seemed to beat faster than usual.
19. I maintained eye contact when I wanted to.
20.
Scoring: reverse "*" items then sum all items.

We also investigated the context-CA and state-CA relationship in a

hierarchical multiple regression analysis. The public speaking context scores
were entered into the equation first, followed by the small group context
scores, then the dyadic context scores. The public speaking context scores
were the only significant predictor (F(1,62) = 63.21, p < .001). It accounted
for 50.4% of the variance in the state-CA score.
One parenthetical point should be made. The Commuhication Anxiety
Inventory is not presented as a rival operationalization of trait-, context-, or
state-CA. It exists as quiet testimony that scientific discoveries are often made
by different researchers simultaneously. Whatever validity can be ascribed to
the CAI can, we think, also be applied to the PRCA and the reconceptualiza-
tion of CA. It is apparent that there are differences between the two scales.
Form Trait employs a frequency response format, asking the subject to note
how often a particular anxiety behavior is present. This approach is consistent
with the response categories developed by Spielberger to distinguish states
and traits. The brief discussion of the psychometric developnient of the CAI is
intended to function as support for the two studies which will follow, and not
as an initial marketing strategy to displace the PRCA.
The CAI permits us to directly test the state-CA/ context^CA relationship.
a relationship that we think is critical to the new conceptua ization of CA. If

Communication Quarterly Spring 1986 199

the field cannot consistently demonstrate that the level of fear or anxiety
varies with trait- and context-CA, then we must severely overhaul our
theorizing. The two studies to be presented here do test that relationship. We
limited our research to the context-CAs and thus cannot generalize our
findings to the trait-CA.

Study One
This study investigated the relationship between context-CA and state-
CA in three communication settings. We hypothesized that state-CA should
vary with context-CA and setting. Thus, the dyad context-CA should be
highly related to state-CA in dyadic settings, but not in other settings. By
comparison, the public speaking context-CA should be highly related to
state-CA in public speaking settings, but not in dyadic settings.

Method
A week prior to the study. Form Trait and X-2 (Spielberger, Gorsuch, &
Lushene, 1970) were administered to three intact classes of students enrolled
in undergraduate communication courses. The scales were given by the class
instructor who worked from a standard protocol. Participation was voluntary.
The day of the study, the instructor announced that a departmentally
approved experiment was scheduled. Participation was invited as voluntary.
Students from the three classes then asssembled in a large room and were
randomly assigned to one of four conditions: INTERPERSONAL, SMALL
GROUP, PUBLIC SPEAKING, or CONTROL. At the conclusion of the sched-
uled activity, the students completed Form State and X - 1 , then de-briefed.
A total of 50 students completed all scales and participated in the study
(no student refused participation). Ages ranged from 18 to 26 with a mean,
median, and mode of 20. Women composed 56% of the sample.

Experimental Conditions
In the INTERPERSONAL condition, six randomly assigned, mixed-sex
pairs completed a structured dyad activity. This activity consists of a series of
open ended statements each partner completes aloud. The statements are
arranged along a continuum of "public" information to "personal" informa-
tion with the dyad becoming progressively deeper. Students could refuse to
complete any statement.
In the SMALL GROUP condition, three mixed-sex groups (N = 4, 5, and 4
respectively) were instructed to reach consensus on the ever popular "NASA
Survival Checklist." Students were randomly assigned to the groups.
In the PUBLIC SPEAKING condition 14 students were randomly divided
into two equal sections. Each student was allowed 3 minutes to prepare a 2-3
minute impromptu speech on one of three topics of local interest. Speaking
order was randomly decided and announced one speaker at a time.
In the CONTROL condition, 11 students were told "Your part in the

200 Communication Quarterly Spring 1986

experiment will not begin for another few minutes. Please wajt quietly." After
10 minutes, the students completed x - 1 .

Predictor Variables
Three context-CA variables were operationalized as the raw score
obtained on the appropriate factor of Form trait. The three contexts were
dyad (IP), small group (GP), and public speaking (PS). The higher the score,
the higher the CA.
x-2 operationalized general, socio-evaluative A-trait (GA). X-2 is the
most widely used measure of the GA-trait (Buros, 1978). Consult the manual
in Spielberger, et al. (1970) for complete reliability and validity information.

Criterion Variables
State-CA was operationalized by two scales: Form State and x - 1 . x-1 is a
measure of state anxiety in any setting, while Form State is appropriate only in
communication settings. Consult Spielberger et al. (1970) for a description of
the reliability and validity of the x-1 scale.

Data Analysis
Within each condition, the criterion variable of state-CA was regressed
upon the four predictors. The regression was conducted in a hierarchical
fashion; predictors were entered into the equation in an order pre-
determined by the analyst. This method was selected over the more popular
stepwise method since our hypotheses demand specific state-CA and con-
text-CA relationships. Thus, in the analysis for the INTERPERSONAL condi-
tion order of the predictors was IP, GA, GP, and PS. For the SMALL GROUP
condition the order was GP, PS, IP, and GA. For the PUBLIC SPEAKING
condition the order was PS, GP, IP, and GA. Finally, in the CONTROL
condition the order of the predictors was GA, IP, GP, and PS.

Results and Discussion

Generally the results confirmed the main hypothesis. The appropriate
context-CA predictor was significant in the proper experimental condition. In
the INTERPERSONAL condition the IP context-CA was the best predictor (F =
4.57 [1,11] p = .0698). Although this was clearly not statistically significant, it
was very close and much greater than the other three predictors.
In the SMALL GROUP condition, the GP context-CA {F 12.77 [1,13],
p = .0073) and the IP context-CA {F = 8.73 [1, 13], p .0183) were
significant. In the PUBLIC SPEAKING condition only the PS context-CA (F =
6.68 [ 1 , 14], p = .0295) was significant. Finally, in the control condition the
GA-trait factor ( f = 6.90 [1, 10], p = .0392) was significant. ,
We also analyzed the full models with all predictors in the equation. Only
one (SMALL GROUP) of the four models met the .05 level of significance, yet
we think that the results from these equations are noteworthy. The relation-

Communication Quarterly Spr\ng^98f) 201

ships between the set of predictors and state-CA by condition show very large
effects. The f^ shows: 1.19 in the INTERPERSONAL condition, 3.27 in the
SMALL GROUP condition, 1.06 in the PUBLIC SPEAKING condition, and 1.95
in the CONTROL condition. As a point of comparison, Cohen (1977) recom-
ends f^ = .35 as a "large" effect.
The obtained effect sizes are more clearly demonstrated in the MR^
values. All four models explain a minimum of 50% of the state-CA variance.
Explained variance ranges from 51.6% (PUBLIC SPEAKING), to 54.5%
(INTERPERSONAL), to 66.2% (CONTROL), and 76.6% (SMALL GROUP). The
absence of statistical significance in the three of the four models appears to be
attributable to the small sample size and the accompanying low power. While
it is certainly possible that these could be explained by sampling error, such
large values so consistently demonstrated across four conditions warrant our
attention and examination.
The obtained means from this study offer confirmation for the "imaginary"
experiment carried out in the validation of the CAI. The obtained mean
state-CA score in the PUBLIC SPEAKING condition was 43.85. During the
validation procedure the obtained mean was 44.06. Thus, it appears that
people can recall or project their state-CA response with acceptable accura-
cy.
Another encouraging finding was the correlation between Form State and
X - 1 . The overall correlation was .69 (N = 39). When analyzed by condition
the correlations varied from .36 in the INTERPERSONAL condition to .81 and
.91 in the PUBLIC SPEAKING and SMALL GROUP conditions.
The pattern of these correlations when combined with the results of the
regression analyses suggests that: Form State and x-1 are strongly related and
thus tapping a similar construct, that context-CA and state-CA vary loosely
and strongly by setting, and that something did not work as well in the
INTERPERSONAL condition compared to the other conditions.

Study Two
The goal of this study was twofold: to clarify the relationship between the
dyad context-CA and state-CA and to evaluate the convergent-discriminant
relationship of the context-CA with test anxiety. Study One clearly demon-
strated that the context-CA scores converged and discriminated appropri-
ately within communication situations. It is important that a similar relation-
ship obtain when the context-CAs are compared to a related anxiety
construct. Thus, we hypothesize that the context-CAs will predict state-CA in
communication settings, but that test anxiety will have only a limited relation-
ship.

Method
Fifty-Six undergraduates enrolled in an introductory communication
course participated voluntarily in this study. Ages ranged from 18 to 56 (mean
was 27, median was 23); the sexes were evenly divided. During the first week
of the class, the students completed Form Trait, Y-2 (the revised version of

202 Communication Quarterly Spring 1986

X-2. see Spielberger, Vagg, Barker, Donham, and Westberr>|, 1980 for scale
development and validation), and an experimental Test-Anxiety Trait mea-
sure. ^ Later the Y-1 scale was administered on two occasions: after an
interpersonal activity on self disclosure and then directly after the mid-term
exam.

Experimental Conditions
In the INTERPERSONAL condition, students were randomly assigned dyad
partners. They then completed a disclosure activity which instructed them to
" . . . practice your skills and get to know your partner." A list of 20 topics,
ranging from public to personal information, was provided. At the conclusion
of the activity, the students completed Y - 1 . In the TEST condition, students
completed their mid-term exam and then filled out the Y-1 scale.

Predictor and Criterion Variables

I

As in the previous study. Form Trait operationalized the three communi-

cation context-CAs. The Test-Anxiety Inventory operatiofialized the trait
test-anxiety. This measure is a 20 item self-report scale formatted like the
STAI. In this study the obtained alpha was .945. We employed the new,
revised STAI (consult Spielberger, Vagg, Barker, Donham, and V\/estberry,
1980 for validity testing) to operationalize the general socio-evaluative
anxiety trait and to operationalize state-anxiety in both the INTERPERSONAL
and the TEST condition. We employed the more versatile Y-1 scale in this
instance rather than our Form State because our scale is appropriate only in
communication situations.

Data Analysis
Data were analyzed under a canonical correlation model. This model
allowed us to directly and simply test two hypotheses: a) two solutions should
emerge with state-CA loading on one variate and state-test anxiety loading
on the other, and b) the communication contexts should load with state-CA
while trait test-anxiety should load with state-test anxiety.

Results and Discussion

Table 2 displays the obtained results of the canonical analysis. Expectedly,
two statistically significant solutions emerged. The first solution captured most
of the variance in both set of variables, yet clearly, state-CA and communica-
tion context-CA relationship is the strongest. The picture clarifies with the
second canonical variate. State-test anxiety and trait test ahxiety load most
heavily, while the communication variables load only moderately and nega-
tively.

Discussion
The obtained results from the two studies demonstrate and confirm the
relationship between a person's level of fear or anxiety in a particular

Communication Quarterly Spring ^98(> ! 203

TABLE 2 Form Trait Descriptive Statistics and Reliabilities
Canonical Variates
Predictors XI X2
IP .7090 .5689
GP .5645 .1177
PS .3512 .0135
GA .8179 .0629
TA .4306 .7514
percent trace 35.98 18.12
Criterion
DYAD State-CA 9717 .2361
Test A-State .4917 .8707
percent trace 59.29 40.69
Solution RC RC^ Lambda X' df p
1 .6634 .4402 .4625 39.23 10 .000
2 .4168 .1737 .8262 9.73 4 .045

communication setting and the appropriate context-CA. Furthermore, this

relationship is quite large. The models show that, on average, 50% of the
variance in the level of fear or anxiety is explained by one and occasionally
two context-CA scores. Typically, researchers are quite pleased to report
significant correlations of .33 which explain 10% of the criterion variance. It
would appear that the reconceptualization works nicely in this application.
The studies also demonstrate the validity of the CAI (and, indirectly, the
PRCA-24) and the reconceptualization. As noted earlier, it is critical that
operational measures of trait- and context-CA clearly and systematically
predict the state-CA response. We think that this is an essential test and any
scale that cannot meet it is not valid. State-CA scores should vary with
different levels of context-CA or trait-CA. The CAI, when placed within the
framework of the reconceptualization, meets the test.
A point of empirical comparison illustrates the relative power of the CAI. A
recent study (McCroskey & Beatty, 1984) analyzed trait-CA as the accumula-
tion of state-CA experiences. Using a regression model, they found that four
state-CA scores explain nearly 50% of the variance in trait-CA as measured
by the PRCA-24. Most interesting, however, was that the interpersonal and
group context-CA scores correlated at less than .35 with state-CA as
measured by x-1 (Spielberger et al., 1970). These results were unexpectedly
low and the authors felt it necessary to account for them. While their solutions
may be correct, the problem could be within the X-1 scale. That scale is
designed to measure a more general state anxiety response. Form State,
designed strictly to measure communication state anxiety, correlates above .7
with context-CA in most situations assessed thus far. When combined with
the PRCA-24 and used to replicate the McCroskey & Beatty study, the results
might improve.
In summary, the studies reported here offer strong construct validity for
the reconceptualization of CA. Furthermore, the studies also provide direct

204 Communication Quarterly Spring 1986

support for the validity of the CAI and indirect support for the validity of the
new PRCA.
Note
'We thank Dr. Spielberger for his permission to employ this scale. Interested researchers may contact
the authors for more information or Dr. Spielberger at the University of South Florida, Tampa, FL 33620
where he is the Director for the Center for Research in Community Psychology.

References
Behnke, R., & Beatty, M. (1981). A cognitive-physiological model of speech anxiety. Communication
Monographs, 48,158-163. |
Behnke, R., & Carlile, L. (1971). Heart rate as an index of speech anxiety. Speech Monographs, 38, 65-69.
Behnke, R., Carlile, L., & Lamb, D. (1974). A psychophysiological study of state and trait anxiety in public
speaking. Central States Speech journal, 25, 249-253.
Buros, O. (1978). The eighth mental measurements yearbook. Highland Park, NJ: The Gryphon Press.
Cohen, J. (1977). Statistical power analysis for the behavioral sciences (2nd ed.). New York: Academic
Press.
Freimuth, V. (1976). The effects of communication apprehension on communication effectiveness. Human
Communication Research, 3, 289-298.
Lamb, D. (1972). Speech anxiety: Toward a theoretical conceptualization and preliminary scale develop-
ment. Speech Monograpiis, 39, 62-67.
Lamb, D. (1969). The effect of public speaking on self-report, physiological, and behavioral measures of
anxiety. Unpublished doctoral dissertation, Florida State University.
Lustig, M. (1974, November). Verbal reticence: A reconceptualization and preliminary scale development.
Paper presented at the convention of the Speech Communication Association, Chicago, IL.
McCroskey, J. (1984). The communication apprehension perspective. In ). Daly & J. McCroskey (Eds.),
Avoiding Communication (pp. 13-38). Beverly Hills, CA: Sage.
McCroskey, J. (1981, April). Oral communication: Reconceptualization and a new look at measurement.
Paper presented at the convention of the Central States Speech Association, Chicago, IL.
McCroskey, j . (1977). Oral communication apprehension: A summary of recent theory and research.
Human Communication Research, 4, 78-96.
McCroskey, J. (1970). Measures of communication-bound anxiety. Speech Monographs, 37, 269-277.
McCroskey, J., & Anderson, J. (1976). The relationship between communicatipn apprehension and
academic achievement in college students. Human Communication Research, 3, 73-81.
McCroskey, M., & Beatty, M. (1984). Communication apprehension and accurnulated state anxiety
experiences: A research note. Communication Monograph, 51, 79-84. j
McCroskey, J., Daly, J. & Sorensen, G. (1976). Personality correlates of communication apprehension.
Human Communication Research, 2, 376-380.
Powers, W., & Smythe, M. (1980). Communication apprehension and achievement in a performance-
oriented basic communication course. Human Communication Research, 6, ji 45-152.
Richmond, V. (1978). The relationship between trait and state communication apprehension and interper-
sonal perceptions during acquaintance stages. Human Communication Research, 4, 338-349.
Spielberger, C. (1966). Anxiety and behavior. New York: Academic Press.
Spielberger, C , Gorsuch, R., & Lushene, R. (1970). STAI manual for the state-trait anxiety inventory. Palo
Alto, CA: The Consulting Psychologists Press. I
Spielberger, C , Vagg, P., Barker, L., Donham, G., & Westberry, L. (1980). The factor structure of the
state-trait anxiety inventory. In C. Spielberger and I. Sarason (Eds.), Stress andfanxiety (\/o\ume 7, pp.
1-30). Washington, DC: Hemisphere/Wiley.

Communication Quarterly Spring 1986 205