The effects of alcohol on subjects' perception and processing of consistent and inconsistent social communications


Material Information

The effects of alcohol on subjects' perception and processing of consistent and inconsistent social communications
Physical Description:
viii, 138 leaves : ; 28 cm.
Couchells, Stella Marie, 1954-
Publication Date:


Subjects / Keywords:
Alcohol -- Psychological aspects   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1982.
Includes bibliographical references (leaves 132-137).
Statement of Responsibility:
by Stella Marie Couchells.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000334544
notis - ABW4187
oclc - 09426499
System ID:

Full Text









I would like to thank all the members of my committee,

Drs. Suzanne Johnson, Rudy Vuchinich, Jalie Tucker, Eileen

Fennell, Randy Carter, and Frank Sieka for their critical

appraisals of, and contributions to, this project. Special

thanks are extended to my committee chairmen; to Dr. Suzanne

Johnson, who has been an outstanding advisor throughout my

graduate training, and to Dr. Rudy Vuchinich who taught me

about alcohol research through many months of reading and

discussion. I would also like to thank Dr. Jalie Tucker for

her particular help and interest in this project. In many

ways she unofficially co-chaired this research effort.

Significant portions of this investigation would not

have been possible without the help of my research assis-

tants, Francine Solomon, Sandy Jacot, Doug Haymaker, and

David Lombard. I would like to thank them one and all for

their time and for their fine work. I would also like to

extend my appreciation to Dr. Nathan Perry for kindly allow-

ing me to use his research laboratory, which greatly expe-

dited my data collection.

And, I would like to thank my parents, who provided me

with the funds to pay some subjects in this experiment, in

addition to everything else they've given me, starting with

good genes. Finally, I would like to thank my husband,

Joseph Gulotta, for telling me jokes when I needed to






ABSTRACT . . vii


Traditional Models of Alcohol's Effects 1
Psychosocial Evidence . 5
Balanced Placebo Design Studies 5
Towards a Model of Alcohol's Socially
Significant Effects . 10
Summary of Empirical Evidence 10
Alcohol's Effects on Tasks of
Divided Attention .. .12
Application of Divided Attention
Findings to Social Situations 13
Summary of Relevant Communication
Literature . 15
Rationale of Proposed Research 17
Overview of Experiment . 18

II METHOD . . 22
Development of Experimental Videotape
Stimuli ...... 22
Pilot Study I: Selection of
Actors for Videotapes 22
Construction of Videotapes 25
Pilot Study II: Selection and
Verification of Experimental
Videotape Segments .. 26
Independent Variables . 35
Dependent Measures . 36

Subjects . . 41
Procedure ... .. . 42



Overview of Data Analyses .
Manipulation Checks ... .
Beverage Consumption Estimates .
Intoxication Ratings ..
Final BALs . .
Dependent Measures .
Scaled Ratings of Videotapes .
Recall Measures .. .
Behavioral Measures .

Summary of Results .
Conclusions . ..
Beverage Administration .
Beverage Content Instructions .
Effects of Videotape Condition and
Sex of Subject .
Methodological Limititations and Implica-
tions for Future Research ..














S 106

S 107

S 109


S 112


S 117

S 119




TAPES . . 125




REFERENCES . . .. 132


Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



Stella Marie Couchells

December 1982

Chairman: Suzanne B. Johnson
Co-Chairman: Rudy E. Vuchinich
Major Department: Clinical Psychology

This experiment investigated the effects of alcohol,

expectancy, sex of subject, and exposure to consistent and

inconsistent verbal and nonverbal videotape and in vivo in-

teractions on measures of normal drinkers' subjective im-

pressions, recall accuracy, eye contact and proximity. In a

2x2x2x4 split-plot factorial design, 48 males and 48 fe-

males were (a) given an alcoholic (.60 g/kg body weight)

or nonalcoholic beverage and (b) instructed that they were

receiving either an alcoholic or nonalcoholic beverage and

then (c) exposed to four consistent and inconsistent ver-

bal/nonverbal videotaped interactions which served as the

within-subjects factor in this investigation. Subjects'

ratings of the opposite-sex actors in the videotapes and


their verbal, nonverbal, and total recall of the videotape

interactions were the dependent measures. Subjects were

then exposed to a confederate behaving in a positive ver-

bal/negative nonverbal manner and dependent measures of

eye contact duration, eye contact frequency, and proximity

to the confederate were recorded.

Of central importance was a result involving beverage

effects for the eye contact duration measure with given al-

cohol subjects engaging in longer durations of eye contact

with the confederate. The a priori hypothesis that alcohol

consumption would result in greater interpersonal attrac-

tion toward the confederate was supported. Subjective rat-

ings, recall, and proximity were unaffected by the bever-

age manipulations. Instructions regarding beverage con-

tent, sex, and videotape condition main and interactive

effects were found for the dependent measures of subjec-

tive ratings and recall. The divided attention hypothesis

advanced in this study was not supported. However, the

potential utility of adopting an information-processing

approach to investigate post-consumptive social-behavioral

changes was supported. Results were interpreted and dis-

cussed within this conceptual framework. Implications

for future research were discussed as well.



Traditional Models of Alcohol's Effects

Traditionally, explanations of alcohol's effects on

socially significant human behaviors (e.g., aggression,

heterosexual interaction) have focused on pharmacological

considerations and have held that the drug produces certain

neurological and physiological changes generally of a de-

pressant nature. These changes presumably lead to direct

and specific emotional or motivational changes. Such inter-

nal, physiologically-based disruptions due to alcohol con-

sumption then are purported to be expressed behaviorally by

decreases in "anxiety" or "stress" and/or increases in so-

cially undesirable sexual or aggressive behavior.

Two related notions, the tension-reduction and disin-

hibition theories, have dominated psychological approaches

to drinking behavior over the past century. The tension-

reduction theory (TRT) of alcohol consumption states that

alcohol's depressant properties produce a reduction in anx-

iety or tension in humans, and more importantly, that when

stressed, humans will consume alcohol in order to ameliorate

adverse emotional states.

Three independent reviews of the empirical evidence

bearing on the TRT, however, found only equivocal support

for the theory. Instead, results that contradict a strict

tension-reduction interpretation of alcohol consumption pre-

dominate (Cappell, 1975; Higgins, 1976; Marlatt, 1976).

These reviews each emphasize the need to consider the medi-

ating effects of social context, individual differences,

and learning in human drinking behavior.

The disinhibition theory differs from the TRT in its

emphasis on how people behave following alcohol consumption,

rather than why people drink. In its most general form, the

disinhibition theory holds that existing portions of an in-

dividual's behavioral repertoire (thoughts, emotions, or be-

haviors), often offensive in nature, which are presumed to

be ordinarily held in check by neurological or intrapsychic

mechanisms, are released from the inhibiting mechanism and

allowed overt expression as a result of alcohol consumption.

The various neurological, psychodynamic, and behavioral for-

mulations of the disinhibiting effects of alcohol differ pri-

marily in their emphasis on the nature of the inhibiting

mechanism specified to be selectively disrupted by alcohol

(i.e., cortex, super-ego, anxiety).

As reviewed in a recently published article (Tucker,

Vuchinich, & Sobell, 1982), a major difficulty with either

the tension-reduction or disinhibition theories of drinking

behavior is that a complete account of the physiological

changes resulting from alcohol consumption is unavailable.

Moreover, the pharmacological effects that have been docu-

mented are by no means consistent or uniform within or

across physiological systems or functions. Some systems

are depressed by alcohol (reticular formation and cerebel-

lar motor coordination centers), but others evidence bi-

phasic responsivity to alcohol (evoked potentials, EEGs,

heart rate, respiratory rate and volume, catecholamine and

other biogenic amine release), and others (GSR, muscle ten-

sion levels, and interneuronal synaptic transmission) show

inconsistent patterns of responding (Kalant, 1970, Naitoh,

1972). These authors concluded that pharmacologically based

theories of drinking behavior disinhibitionn and tension-re-

duction) that postulate a one-to-one correspondence between

physiological changes and resulting social or emotional be-

haviors appear to be overly simplistic, not only in this

"assumption of identity," but also in their assumption that

alcohol induces consistent or uniform changes in physiology.

Psychosocial Evidence

Many recent studies have demonstrated that a direct

relationship does not exist between alcohol's physiological

effects and behavioral changes. Experiments that have as-

sessed alcohol's effects on anxiety have variously found no

changes in anxiety (Kastl, 1969; Nathan & O'Brien, 1971;

Smith, Parker, & Nobel, 1975), decreases in anxiety (Menaker,

1967; Vannicelli, 1972; Warren & Raynes, 1972; Williams,

1966), and increases in anxiety and dysphoria (McGuire,

Stein, & Mendelson, 1966; McNamee, Mello, & Mendelson, 1968;

Mendelson, LaDou, & Solomon, 1966; Steffen, Nathan, & Tay-

lor, 1974; Tamerin, Weiner, & Mendelson, 1970; Vannicelli,

1972; Williams, 1966). With respect to aggression, alcohol

consumption has been found to increase the aggressive con-

tent of TAT stories by males, but only in the presence of

an attractive female (Kalin, 1972) and at high doses of al-

cohol (McClelland, Davis, Kalin, & Wanner, 1972). Likewise,

studies of sexual arousal following alcohol consumption

have yielded equivocal results. Physiological indices of

sexual arousal have been found to decrease with increasing

doses of alcohol in both males (Briddell & Wilson, 1976;

Farkas & Rosen, 1976; Wilson, Lawson, & Abrams, 1978) and

females (Wilson & Lawson, 1976). In the latter study, self-

reported sexual arousal increased as a function of increas-

ing alcohol dose, however. The only study to date that in-

vestigated the effects of alcohol on significant nonverbal

behaviors associated with interpersonal attraction employed

like-sex or opposite-sex interviewers who engaged male and

female subjects who were either given alcohol or given non-

alcohol in a dyadic interaction (Lindman, 1980). Gaze dura-

tion and "mood changes" were the dependent variables under

consideration. Contrary to the experimenter's hypothesis,

alcohol consumption did not result in longer durations of

eye contact. In fact, there was a definite trend for fe-

male subjects who had consumed alcohol to look at their

like-sex interviewers less than sober female subjects.

Clearly, the behavioral consequences following alcohol

consumption are diverse and often contradictory (McAndrew &

Edgerton, 1969). As suggested by the more recent evidence

discussed below, a number of nonpharmacological factors in-

cluding the setting in which alcohol consumption occurs,

the drinking history of the subject population, and subjects'

beliefs about alcohol's effects (Marlatt, 1976) all influ-

ence the social-behavioral and emotional effects observed

following alcohol consumption. For traditional pharmacolog-

ical explanations to hold true, such factors as these ought

to have little to no bearing on the emotional and behavioral

consequences observed. Yet, as discussed below, these

"extrapharmacological" factors often have a rather powerful

impact on the socially significant behavior observed follow-

ing at least moderate alcohol consumption.

Balanced Placebo Design Studies

A methodological innovation known as the balanced pla-

cebo design (BPD) directly addresses the question of how

much intoxicated behavior can be attributed to individuals'

beliefs and expectations about alcohol's effects and how

much can be attributed to any pharmacological effects of al-

cohol itself. Specifically, in a 2x2 factorial design, the

BPD manipulates beverage administration (given alcohol or

given nonalcohol) and instructions given to subjects re-

garding the alcoholic contents of their beverages (told al-

cohol or told nonalcohol). The procedures found to con-

vincly maintain the deception in the BPD are detailed in

Marlatt and Rohsenow (1980).

Although a few studies employed a balanced placebo de-

sign in investigating the notions of "craving" and "loss of

control" drinking by alcoholics (Engle & Williams, 1972;

Merry, 1966), it was not until Marlatt, Demming and Reid

(1973) employed this design in a similar investigation of

drinking behavior that the BPD gained attention in other

areas of alcohol research. In their now classic analogue

study, Marlatt and his colleageus (1973) used a balanced

placebo design with both alcoholic and nonalcoholic sub-

jects and then measured their ad lib alcohol consumption

during a bogus "taste test." Contrary to traditional no-

tions of craving and loss of control, subjects' belief of

having consumed alcohol, rather than actual alcohol consump-

tion, was the primary determinant of subjects' subsequent

alcohol intake.

Since Marlatt and others' experiment, the BPD has been

employed in many investigations of postconsumptions changes

in various social behaviors and emotions, including changes

in anxiety, sexual arousal, and aggression. Indeed, as dis-

cussed below, the BPD has become the sine qua non of re-

search on human drinking behavior.

Anxiety studies

Alcohol's pharmacological and expected effects on so-

cial anxiety have been investigated in four separate stud-

ies that employed the BPD (Abrams & Wilson, 1979; Polivy,

Schueneman & Carlson, 1976; Wilson & Abrams, 1977; Tucker,

Maisto, Vuchinich, & Blumenthal, 1979). In the Abrams and

Wilson study (1979), heart rate, skin conductance, overt

behavioral responses, and self-reports of anxiety were mon-

itored in female subjects following beverage administration

during a social interaction with a male confederate. The

belief of having consumed alcohol significantly influenced

subject's physiological and behavioral responses during the

brief social interaction, but not in the predicted direc-

tion (i.e., women who believed that they had consumed al-

cohol, regardless of whether their drinks contained alcohol,

showed significant increases in levels of physiological

arousal compared to those who believed that they consumed

only tonic water). In addition, subjects were rated as ex-

hibiting greater discomfort when they believed they had con-

sumed alcohol, while self-report measures of anxiety failed

to differentiate the experimental groups (Abrams & Wilson,


Wilson and Abrams (1977) used an interaction situation

with males that was procedurally similar to their study with

females (Abrams & Wilson, 1979). Results showed that sub-

jects' belief that alcohol had been consumed led to

significantly smaller changes in physiological arousal

(heart rate), regardless of the alcoholic content of sub-

jects' beverages. A marginally significant instructional

main effect also was found for self-reported anxiety; male

subjects who believed they consumed an alcoholic beverage

reported experiencing less anxiety than those who believed

they consumed a nonalcoholic beverage.

Polivy et al. (1976) used a balanced placebo design to

investigate self-reported anxiety in males following the

threat of painful electric shock contingent on motor task

performance. Subjects who were administered an alcoholic

beverage reported significantly less anxiety than those who

were administered a nonalcoholic beverage. Additionally,

subjects who believed that they had received alcohol showed

significant increases in self-reported anxiety, relative to

subjects who believed that they had consumed only tonic.

In the only balanced placebo design to date which em-

ployed both male and female subjects, Tucker et al. (1979)

assessed anxiety using radial pulse measures and self-re-

port questionnaires in response to threats of evaluation of

their performance on either a pursuit motor task or a con-

versational social skills task. The results of this exper-

iment precluded inferences concerning alcohol's pharmaco-

logical and expected effects because the deceptions asso-

ciated with the BPD were not fully effective. Nevertheless,

the results obtained highlighted the critical role of the

drinking context (motor vs. conversation task) and subjects'

sex in determining affective responses to alcohol. Males

who consumed alcohol evidenced greater anxiety in the motor

task condition and less anxiety in the conversational task

condition than males who were not given alcohol. In con-

trast, females who consumed a nonalcoholic beverage reported

greater anxiety in the motor task condition than females who

had consumed an alcoholic beverage.

In sum, alcohol administration, the belief that alco-

hol had been ingested, sex, and contextual variables have

all been shown to influence social anxiety in humans.

Aggression studies

Lang, Goeckner, Adesso, and Marlatt (1975) conducted

the only BPD study to investigate postconsumption changes

in aggression as measured by a Buss "aggression machine."

After the beverage administration and instruction manipula-

tions, male subjects were either provoked by a male confed-

erate or were involved in a neutral interaction with him.

Then subjects were permitted to aggress against the confed-

erate by administering shocks of various intensities to him

in a teacher-learner paradigm. Results showed that subjects

who were told that they had received alcohol administered

significantly more shocks of a higher intensity to the con-

federate than those subjects who did not believe that they

had consumed alcohol, regardless of the alcohol content of

their drinks.

Studies of sexual arousal

Balanced placebo design studies of postconsumptive

changes in sexual arousal in male subjects have been quite

consistent in finding greater sexual arousal under condi-

tions in which subjects believed that they had consumed an

alcoholic beverage regardless of the actual alcoholic con-

tent of their drinks (Briddell, Rimm, Caddy, Krawitz, Sho-

lis, & Wunderlin, 1978; Wilson & Lawson, 1976). However,

the only study employing a balanced placebo design to as-

sess sexual arousal using female subjects found that only

actual alcohol administration increased both subjective re-

ports and physiological indices of sexual arousal in re-

sponse to erotic slide presentations (Wilson & Lawson, 1978).

Towards a Model of Alcohol's
Socially Significant Effects

Summary of Empirical Evidence

The above studies amply demonstrate that socially sig-

nificant behaviors are mediated not only by the pharmacolog-

ical properties of alcohol, but by contextual characteris-

tics, cognitive variables (i.e., expectancy), and the drink-

er's sex. How then may these results be utilized to delin-

eate a mechanism by which alcohol influences and effects

such changes in social behaviors and emotions?

Based on their review of cross-cultural studies of in-

toxicated behavior, McAndrew and Edgerton (1969) offered an

explanation in terms of social learning theory. Noting that

"drunken comportment" varies cross-culturally, particularly

with respect to normally "taboo" behaviors, these authors

suggest that the types of inebriated behavior a society will

tolerate, and the imparting of this information to its mem-

bers, may be far more predictive of the types of intoxicated

behaviors observed than those predicted by traditional phar-

macological theories.

Although social learning theory may account for cross-

cultural variations in intoxicated behaviors and may par-

tially explain how beliefs about alcohol's effects influence

intoxicated behaviors, it, too, seems overly simplistic to

account fully for the results obtained in research on alco-

hol's effects on social behaviors. Recall that at least

some of the BPD studies showed strong beverage effects

(Polivy et al., 1976; Tucker et al., 1979; Wilson & Lawson,

1978) suggesting that alcohol's pharmacological properties

influence social responding in some way.

Thus, while "expectancy" effects have been instrumen-

tal in refuting traditional pharmacological explanations of

alcohol's effects on human social behavior, it may be pre-

mature and erroneous to adopt any new theories that are

based solely on nonpharmacological, cognitive-expectancy

variables to the exclusion of considering pharmacological

variables as determinants of postconsumption social behav-

iors. Therefore, the more fruitful efforts might be spent

in attempting to ascertain how pharmacological changes

which are induced by alcohol may influence the many mediat-

ing processes involved in normal social functioning. So-

cial behavior is a complex phenomenon involving many dif-

ferent cognitive processes related to the perception and

interpretation of social cues. The divided-attention lit-

erature reviewed below suggests that alcohol's pharmacolog-

ical effects may interfere with the effective processing of

social stimuli and affect social behaviors accordingly.

Alcohol's Effects on Tasks of Divided Attention

One line of investigation on the pharmacological ef-

fects of alcohol on attentional-perceptual processes has

involved studies of divided attention. This research has

consistently demonstrated that the number of environmental

stimuli perceived by subjects while under the influence of

alcohol is substantially reduced relative to conditions in

which no alcohol has been consumed (Miles, 1933; Gruner,

Ludwig, & Donner, 1964; Moskowitz & DePry, 1968). This re-

lationship has been particularly apparent when the stimuli

involved emanated from peripheral sources (Hamilton & Cope-

man, 1970). The observed behavioral deficits have not been

due to response impairment, but rather because the relevant

stimuli have not been perceived by subjects. When subjects

have been tested under alcohol and no alcohol conditions on

these attentional tasks separately in laboratory investiga-

tions, no decrements in performance have been found

(Moskowitz & DePry, 1968). Rather, it is the ability to

simultaneously attend to and monitor two or more channels

of information that is disrupted by alcohol (Gruner et al.,

1964). Data from an in-depth study of automobile accidents

indicated that intoxicated drivers were more likely to have

been engaged in some pre-accident activity that was secon-

dary to the driving task (Brewer & Sandow, 1980). These

naturalistic data were used to support the notion of re-

sponse impairment under conditions of alcohol ingestion and

divided attention as well.

Such pharmacologically-induced deficits in tasks of di-

vided attention have been further assessed and supported in

a BPD study in which males were administered a task of di-

vided attention (Vuchinich & Sobell, 1978). Error scores

were significantly increased both by alcohol administration

and the belief of consuming alcohol. Subjects who were told

that their beverages contained alcohol performed equally

poorly, regardless of whether they had actually received al-

cohol or not. However, of those subjects who believed that

their beverages were nonalcoholic, those who actually re-

ceived alcohol performed less accurately than those who re-

ceived a nonalcoholic beverage.

Application of Divided Attention Findings to
Social Situations

As suggested by Tucker and Vuchinich (in press) it may

be plausible to apply these laboratory findings on divided

attention processes to social situations. A typical social

situation requires an individual to simultaneously attend

to and process information from several sources. Such pro-

cessing may be expected to affect an individual's overall

assessment of the social situation, which may, in turn, in-

fluence his or her subsequent behavior. However, if the in-

dividual's capacity to attend to or perceive all relevant

stimuli were to be impaired by alcohol, as suggested by the

divided attention literature, the individual's social judg-

ment and subsequent behavior may be hypothesized to change

according to those cues which were (mis)perceived.

Divided attention tasks generally involve simple audi-

tory, visual, motor, or cognitive tasks. An analogous so-

cial attentional-perceptual task would require an individ-

ual to attend to and process communicative responses from

several different "channels" while interacting with another

individual. The dimensions of interest would be the sub-

jects' inferences about the other individual's intentions,

affect, sincerity, and so forth. At the most basic level,

the channels of communication used by the other individuals

involve verbal responses (i.e., the content of the communi-

cation) and nonverbal responses (i.e., how this content is

communicated behaviorally).

To date, no studies have assessed alcohol's effects on

individuals' abilities to perceive and process social cues

from different communication channels, nor has any research

addressed how the meaning of such behaviors may be altered

by alcohol administration. Nevertheless, a large body of

research on verbal and nonverbal communication exists in

the social-psychological literature. Some of this litera-

ture speaks to the way in which normal individuals who have

not consumed alcohol perceive and interpret the behavior

of others when their verbal and nonverbal communications

are conflicting. This research provides a methodological

paradigm for the proposed experiment and is reviewed below.

Summary of Relevant Communication Literature

The social communications literature will be reviewed

with respect to (1) sex differences in the ability to "de-

code" nonverbal behaviors, eye contact, and proxemics, and

(2) how an observer interprets inconsistent modes of commu-


Research studies using the empirically developed Pro-

file of Nonverbal Sensitivity (PONS) have found that fe-

males are consistently more accurate in decoding nonverbal

communicative cues presented in isolation than males at all

age levels. Females also are particularly proficient when

body cues are available and when social stimuli communicate

negative affect (Hall, Rosenthal, Archer, DiMatteo, &

Rogers, 1977). Studies of the frequency of eye contact and

proxemics between the sexes have consistently shown that

women maintain greater overall eye contact than males

(Aiello, 1972; Exline, Gray, & Schuette, 1965; Levine,

1972, Levy, 1972; Neville, 1974; Rubin, 1970; Russo, 1975),

as well as closer interpersonal distances (Hare & Bales,

1963; Liebman, 1970; Long, Ziller, & Henderson, 1968; Norum,

Russo, & Sommer, 1967) when interacting with another indi-


Research addressing the effects on social perception

of consistent vs. inconsistent verbal and nonverbal messages

has shown that nonverbal behaviors may significantly alter

the impact of a verbal message (Mehrabian, 1968; Mehrabian

& Diamond, 1971; Mehrabian & Ferris, 1967). Nonverbal be-

haviors have been shown to enhance or detract from the con-

tent of communication depending on the amount of congruence

between these two channels (Hasse & Tepper, 1972; Shapiro,

1966), with the verbal message being somewhat discredited

or weakened when the nonverbal behavior of the speaker con-

tradicts what is being said. Two separate investigations

that systematically crossed positive and negative verbal and

nonverbal behaviors (Bugenthal, Kaswan, & Love, 1970; Graves

& Robinson, 1976) demonstrated that both male and female

viewers judged the "communicator" in these studies to be

less "sincere" or "genuine" when the verbal message was at

odds with his/her nonverbal behavior than when the verbal

and nonverbal aspects of communication were in accord.

Graves and Robinson (1976) also found that subjects main-

tained greater interpersonal distances under conditions in

which inconsistent messages were presented.

Rationale of Proposed Research

A question not addressed by both investigators of non-

verbal behavior and alcohol researchers concerns the ef-

fects of alcohol on individuals' ability to attend to, per-

ceive, process, and integrate messages emanating from dif-

ferent social communication channels. Although sexual

arousal following alcohol consumption has been conceptual-

ized from an information-processing standpoint,that is some-

what different from the approach taken here (Lansky & Wil-

son, 1981). Alcohol research on tasks of divided attention

suggests that alcohol might exert a disruptive influence on

subjects' abilities to process verbal and nonverbal mes-

sages simultaneously. Relevant social stimuli may be more

likely to be overlooked or misperceived the more contradic-

tory the social situation becomes. The literature on social

communication suggests a methodology that may permit inves-

tigation of the processing of social stimuli by subjects

who consume alcohol by exposing them to consistent and in-

consistent verbal and nonverbal messages and then measuring

their social responses.

Given the contradictory extant research on alcohol's

effects on social behaviors and emotion, it seems reason-

able to examine whether pharmacological changes induced by

alcohol may disrupt perceptual-attentional aspects of so-

cial functioning in a manner similar to alcohol's disrup-

tive effects on tasks of divided attention. Furthermore,

if alcohol disrupts an individual's ability to attend simul-

taneously to and process social cues from verbal and nonver-

bal sources, particularly when these cues are incongruent,

one might expect the intoxicated individual to be less aware

of the inconsistency and therefore to assess and respond

differently to conflicting messages than sober individuals.

In any laboratory study of this phenomenon, the belief that

alcohol has been ingested would not be predicted to exert as

great an influence on these discriminatory behaviors as

would actual alcohol consumption.

Overview of Experiment

The present experiment employed a balanced placebo de-

sign to investigate whether alcohol consumption alters male

and female subjects' perception and processing of video-

taped interactions. It also considered alcohol's effects

on specific interpersonal behaviors duringin vivo social in-

teractions. The videotaped and live social interactions

were heterosexual interactions involving consistent and in-

consistent positive and negative verbal and nonverbal behav-

iors. Therefore, the data obtained in this exploratory in-

vestigation may have important implications for understand-

ing social behavior following alcohol use, which has not

been adequately explained by traditional pharmacological

models. Previous findings on social communication and di-

vided attention deficits following alcohol use were applied

to investigate the effects of alcohol on subjects' informa-

tion processing capabilities during such social situations.

This was assessed using self-report and recall measures as

well as behavioral measures associated with interpersonal


If it can be shown that the subjects in the given al-

cohol conditions recall less verbal and/or nonverbal stimuli

than given nonalcohol subjects in the inconsistent stimuli

manipulations, one may tentatively conclude that the simul-

taneous processing of verbal and nonverbal social cues was

disrupted by alcohol. If, in addition, the inferences made

by given alcohol subjects about the communicators of incon-

sistent messages were simplified relative to those made by

given nonalcohol subjects, one may tentatively conclude that

on the basis of this disruption in processing, subjects be-

came less sensitive to inconsistencies in verbal and non-

verbal behaviors and misperceive the intentions of others.

If, in addition to these findings, it can be demonstrated

that subjects given alcohol respond differently to the con-

federate than subjects in the given nonalcohol conditions,

one may conclude that alcohol-induced changes in some so-

cial behaviors result in part from alcohol's disruptive

influence on the perceptual processing of social information.

Such results may not demonstrate conclusively that dif-

ferences in social comportment commonly observed following

alcohol consumption are unequivocally mediated by the

proposed mechanism. However, if positive results are ob-

tained across recall responses, self-reported inferences

and/or behavioral responses, they would lend credence to

this model and encourage future research efforts using an

information-processing approach.

A Priori Predictions

Although this study is basically exploratory, several

predictions can be advanced on the basis of the existing

relevant research. Subjects who consume an alcoholic bever-

age are predicted to show greater disruptions in the process-

ing of the inconsistent social cues than subjects who con-

sume a nonalcoholic beverage. The directionality of the ef-

fects of this alcohol-induced disruption cannot be predicted

for all of the dependent variables. Nevertheless, it is pre-

dicted that subjects given alcohol will rate actors in the

verbal positive/nonverbal negative and verbal negative/non-

verbal positive conditions as more "sincere" than subjects

given a nonalcoholic beverage. Similarly, it is hypothe-

sized that the recall accuracy for both verbal and nonver-

bal aspects of the actors' behaviors will be lower for the

alcohol groups, particularly for the inconsistent videotapes.

Greater durations of eye contacts with the confederate and

closer seating proximities to the confederate are also pre-

dicted for given alcohol subjects.

The instructional manipulations have been included

largely for methodological rather than conceptual reasons.

Based on the relevant divided attention literature (i.e.,

Vuchinich & Sobell, 1978), instructional effects that may

occur are predicted to interact with the effects of alco-

hol administration.

The sex differences documented in the literature on

nonverbal communication predict that females may be more

sensitive to nonverbal messages than males. How such

heightened sensitivity might manifest itself in the ratings

of inconsistent messages and might possibly interact with

alcohol and/or instructions is difficult to say. However,

one might expect that the females in this study would be

more proficient overall at specifying nonverbal behaviors

on the films, and thus exhibit a higher recall score for

nonverbal behaviors than males, based on the previous lit-

erature (Hall et al., 1977).

The literature also indicates that women maintain

closer interpersonal distances and engage in more frequent

eye contact than males. It is not unlikely then, that a

significant main effect for sex reflecting these differences

in responding will be found between the male and female sub-

jects in this study for the two behavioral ratings of prox-

imity and eye contact.


Development of Experimental Videotape Stimuli

Since no standardized experimental videotape stimuli

existed from any previous investigations which employed the

types of verbal/nonverbal interactions necessitated by this

study, preliminary studies were conducted to develop the

necessary experimental videotapes. The experiment called

for eight videotape segments, including four segments fea-

turing male actors and four equivalent segments featuring

female actresses. Thus, two equivalent segments (one using

a male actor and one using a female actress) were needed in

each of the following verbal/nonverbal portrayals: positive

verbal/positive nonverbal, positive verbal/negative nonver-

bal, negative verbal/positive nonverbal and negative verbal/

negative nonverbal interactions. The process by which these

videotape conditions were constructed is detailed below.

Pilot Study I: Selection of Actors for Videotapes

Sixteen potential actors (eight males and eight fe-

males) were initially selected on the basis of similarity

of age, coloring and facial features to participate in mak-

ing the videotapes. To select four male and four female

actors of similar attractiveness on an empirical basis, a

pretest study was conducted in which 20 female undergrad-

uates rated the 8 potential male actors and 18 male under-

graduates rated the attractiveness of the 8 potential fe-

male actresses.

Subjects received one hour of experimental course

credit for their participation in individual 30 minute ses-

sions. Full body photographs of all potential opposite sex

actors were presented to subjects in counterbalanced order.

Photographs were presented for 30 seconds, during which time

subjects were asked to make their attractiveness ratings,

using the bipolar scales shown in Appendix I. There was a

10 second delay between presentations.

The data points on the attractiveness rating scale in

Appendix I were placed at one inch intervals. The number

of inches from the zero midpoint to subjects' marks consti-

tuted their ratings. The datawere transformed by adding

four inches to each rating for the analyses. Thus, the

range of possible values extended from 0-8, with 4 being a

neutral rating.

Attractiveness ratings were then analyzed separately

for male and female subjects using a repeated measures anal-

ysis of variance for each sex with one within subjects fac-

tor (photo) which had eight levels. Significant differences

between photos were found within each sex [in male subjects:

F(7.119)=9.77, p=.001, female subjects: F(7,133)=17.1,

p=.001]. Duncan's Multiple Range Procedure was used to select

the four photos for each sex which received mean attractive-

ness ratings that were not significantly different from one

another. These data can be found in Tables II.1 and 11.2,

in Appendix II.

The mean ratings shown in Table II.1 for the first

three male photos (MP, MH, and GG) are not significantly

different from one another. Female subjects did, however,

judge the photo of RW to be significantly less attractive

than those for MP and MH. Nevertheless, because the mean

ratings for RW and GG did not differ significantly, a deci-

sion was made to use actor RW in constructing the videotape

stimuli. All four mean ratings for female photos were not

significantly different, as seen in Table 11.2. These

eight actors were thus used in constructing the videotapes.

The attractiveness ratings for the eight photos of act-

ors to be used for the experimental stimuli were then sub-

jected to a 2x8 repeated measures ANOVA with one between

subjects factor (sex) and one within subjects factor (photo),

in order to establish comparability between the two sets of

male and female photos along the attractiveness dimension.

Sex of subject did not significantly differentiate between

ratings, F(1,36)=1.81, p=.18, although significant differ-

ences between photos were found, F(7,107)=2.31, -=.03.

Mean attractiveness ratings for photos were analyzed

using Duncan's procedure for pairwise contrasts. These re-

sults are presented in Table III.1, in Appendix III. The

data in Table III.1, and the fact that no significant main

effects were found for sex, illustrate that with only a few

exceptions, male and female subjects gave comparable at-

tractiveness ratings to all eight photos.

Construction of Videotapes

Each actor was trained, individually, to simulate the

four verbal/nonverbal response conditions. Each actor was

then taped portraying these four conditions (verbal accep-

tance/nonverbal acceptance; verbal acceptance/nonverbal re-

jection; verbal rejection/nonverbal acceptance and verbal

rejection/nonverbal rejection) with an opposite-sex individ-

ual who remained identical within each sex and across tape

conditions. Sixteen videotape segments using female actors

and 16 videotape segments using male actors were thus ini-

tially available from which to choose the final 8 video-

tapes to be used in the study.

In each videotape interaction, an opposite-sex indi-

vidual approached a seated actor who was seen in full view

throughout the interaction. The opposite sex individual

took a seat facing the actor and remained with his/her back

to the camera throughout each filming. The positive and

negative scripts for these segments can be found in Appendix

IV. In training for the nonverbal segments, positive non-

verbal behaviors included drawing closer to the opposite-

sex individual, smiling, head nodding, frequent eye contact,

and sitting face forward. Negative nonverbal behaviors

included turning at a 900 angle from the seated opposite-

sex individual, frowning, sighing, shrugging, and relatively

infrequent eye contact. These nonverbal behaviors have been

found in a review of the literature to reflect positive and

negative nonverbal communications, respectively (Harper,

Wiens, & Matarazzo, 1978).

Pilot Study II: Selection and Verification of Experimental
Videotape Segments

In order to select the eight experimental videotapes

to be used in the experiments from 32 segments available,

two female undergraduates were recruited to rate the 16

videotapes portraying males and two male undergraduates

were used to rate the 16 videotapes portraying females.

Raters were not informed of the specific nature of the con-

ditions portrayed. These raters first viewed their respec-

tive videotapes without sound, and then made a scaled rat-

ing as to how positive or negative the actor appeared. They

then listened to the audio portions of the segments and made

identical rating. Raters then viewed the composite video-

tapes and made a third global rating of how positive or neg-

ative the actors' responses were, to the opposite-sex indi-

vidual on the videotape. These rating scales can be found

in Appendix V. Again, all segments were viewed and heard in

counterbalanced order.

Interrater reliability was fair to excellent for the

16 segments of videotape depending on the stringency of the

agreement criteria. When reliability was defined as a

1-point or less discrepancy between raters, 44-69% agree-

ment was achieved for both sets of raters for the audio,

video, and composite conditions. Allowing a 2-point or less

discrepancy resulted in reliability indices of 75-100%.

Fifty-six to 100% agreement was achieved when reliability

was defined as the number of ratings between raters which

were made in the same direction from the midpoint of 4.

These data are presented in Table VI.1, Appendix VI.

Of the two inconsistent verbal/nonverbal conditions

portrayed by each actor (positive verbal/negative nonverbal

and negative verbal/positive nonverbal), the actors' por-

trayals that were given the most discrepant and reliable

ratings in the isolated audio and video conditions by the

four raters were selected for use in the experiment. The

two remaining consistently positive and negative conditions

portrayed for each sex were then chosen from the remaining

unselected actors' portrayals, according to the most reli-

able and extreme positive and negative ratings made by rat-

ers. These data are presented in Table VII.1, Appendix VII.

This procedure yielded eight videotapes featuring four dif-

ferent males and four different females of similar attrac-

tiveness in each of the four verbal/nonverbal conditions

necessitated by the study.

It should be noted that the male and female actors se-

lected by raters for each conditions received mean

attractiveness ratings that were not significantly differ-

ent from one another (Table III.1, Appendix III). For ex-

ample, the male and female (MH and DM) selected for the pos-

itive verbal/positive nonverbal condition, received mean at-

tractiveness ratings that were not significantly different

from one another. These data, coupled with the comparable

training experiences each sex underwent in making the films,

support the notion of equivalency between male and female


A final pretest procedure then was conducted to verify

the selection of experimental videotapes. Twenty-two male

and 19 female subjects (in groups of 4-5) simultaneously

viewed both the audio and video portions of their respective

four interaction stimuli in counterbalanced order. The six

self-report measures to be used in the actual experiment

were collected following each presentation. These data

were used to verify that the four verbal/nonverbal condi-

tions had been adequately constructed and that the rating

scales to be used in the experiment discriminated between

tapes. These procedures also assessed the extent to which

subjects generally attend to information from the verbal

and nonverbal channels.

The six rating scales used by female subjects are pre-

sented in Appendix VIII. An identical set of scales with

the actor's sex identified as female was used for male sub-

jects. The number of inches from the zero midpoint to where

subjects had placed their mark constituted their rating for

each scale. These data were transformed by adding five

inches to subjects' ratings. Thus, the range of possible

values extended from 0-10, with 5 being a neutral rating,

for each of the six dimensions assessed.

All six self-report measures were subjected to a 2x4

repeated measures ANOVA in a split-plot design with one be-

tween-subjects factor (sex) and one within subjects factor

(the four videotape conditions).

Highly significant main effects were found for video-

tape condition on all six dependent measures (p=.0001), in-

dicating that this manipulation was quite powerful. Mean

differences between tapes were likewise in the predicted

directions, thus establishing the validity of the intended

manipulations. These results are presented in Table IX.1,

Appendix IX.

Subjects judged the consistently positive tapes (PP)

as significantly more positive than any other verbal/non-

verbal condition with respect to (1) how much the actor

seemed to like the opposite-sex speaker, (2) how much (s)he

wanted to go to dinner, (3) how sincere (s)he was about what

was said, (4) how personally likable subjects found the

actor to be, and (5) how likely subjects would be to make

a similar request in the future. With respect to how at-

tracted subjects judged the actors to be toward the oppo-

site-sex speaker, the consistently positive videotape

condition condition (PP) and the negative verbal/positive

nonverbal condition (NP) received mean ratings that were

not significantly different from one another. This sug-

gests that in evaluating attraction, when the nonverbal be-

havior of the actor was positive, subjects tended to disre-

gard the verbal content of the interaction. This was not

true, however, for the case where the actor's nonverbal be-

havior was negative with positive verbal content (PN). In

assessing attraction in this condition, it would appear that

the positive verbal content influenced subjects' ratings to

some extent, given the significant mean difference between

this condition and the NN condition on attraction ratings.

The consistently negative condition, NN, received mean rat-

ings that were significantly more negative than all other

three conditions for all six self-report ratings.

With the exception of the attraction rating previously

discussed, all subjects tended to rate the two inconsistent

conditions significantly more positively than the NN tape

and more negatively than the PP condition. In evaluating

sincerity (Rating 4), likability (Rating 5), and intentions

(Rating 3), any inconsistency in verbal/nonverbal channels

resulted in no significant differences between the two in-

consistent tapes, suggesting that male and female subjects

responded equally to both channels of communication in as-

sessing these qualities. However, in evaluating how much

the actor liked the speaker (Rating 1), and the probability

of their asking again (Rating 6), condition NP received

significantly more positive ratings than condition PN, indi-

cating that subjects relied more on the nonverbal behavior

of the actor in making these judgements than on the verbal

content of the interaction.

Marginally significant main effects for sex were found

for evaluations of sincerity (Rating 4), F(1,39)=3.8, E=

.0580, likability (Rating 5), F(1,39)=4.01, p=.052, and

probability of asking again (Rating 6), F(1,39)=3.05, E=

.088. Table IX.1, Appendix IX, shows that male subjects

rated the female actors more positively than the female sub-

jects rated the male actors with respect to both sincerity

and likability. The means for probability of asking again

were not significantly different from one another.

Although no significant Sex x Condition interactions

were found, some marginally significant interaction effects

were obtained as seen in Table IX.1. In addition, to as-

sess any differences in the patterning of responses that

might occur between male and female subjects, the data for

each sex were analyzed separately in two repeated measures

ANOVAs with one within-subjects factor (videotape condition).

Videotape condition remained a highly significant factor for

both males and females on all 6 dependent measures (p=

.0001 in all cases). Significant mean differences between

conditions were analyzed using Duncan's procedure (Tables

X.1 and X.2, Appendix X). A comparison of Tables X.1 and

X.2 reveals that overall, male subjects were more inclined

to disregard the negative verbal content of the interaction

and responded almost exclusively to the nonverbal behavior

of the female actor when that behavior was positive, than

did female subjects. In contrast, female subjects appeared

to use data from both channels of communication to evaluate

actors on all measures with the exception of attraction.

These differences are more likely due to sex of subject

than to any differences between tapes in the NP condition.

This interpretation is supported by (1) the fact that the

actors for these segments (MP and RI) did not differ signif-

icantly on attractiveness ratings made earlier (Table III.1,

Appendix III), (2) the actors used identical scripts and re-

ceived identical training experiences, and (3) data pre-

sented later will demonstrate that the frequency of positive

nonverbal behaviors displayed in the two tapes was equivalent.

Intercorrelations using all subjects for the six depen-

dent measures were highly significant across conditions

(N=41, r=.34 to r=.88; p<.05) with the exception of the NN

condition where actors' likability did not correlate signif-

icantly with ratings that subjects made of the actors' be-

haviors toward the opposite-sex actor in the films (N=41,

r=.22 to r=.29, p<.067 to p<.16). Correlation matrices for

each sex showed that while the six ratings for female sub-

jects were significantly and positively intercorrelated

across all four conditions (N=19, r=.46 to r=.95; p<.05),

male subjects evidenced considerably less uniformity in

making their ratings. Generally, the ratings that male sub-

jects made regarding the female actors' behavior toward the

male in the tapes did not correlate significantly with eval-

uations of either likability or ratings of likelihood of

asking again (N=22, r=.19 to r=.40; 2<.06 to p<.40). This

finding occurred across all videotape conditions.

In sum, the data from this pilot investigation veri-

fied that the four videotaped conditions were adequately

constructed and established their validity as experimental

stimuli. In addition, it shows that all six dependent meas-

ures to be used in the study were effective in discriminat-

ing between conditions. Some sex differences were obtained

that suggest that male subjects generally attended more to

the nonverbal channel of communication when that behavior

was positive than to verbal channels. Female subjects gen-

erally used information from both channels in evaluating so-

cial interactions. Male subjects also evidenced greater

variability in assessing different aspects of the social in-

teractions portrayed than did females.

Neutral Videotapes

The greater variability in responding seen by male sub-

jects in the pilot investigation and recent data from a re-

lated study (Tucker & Vuchinich, in press) suggested that

some baseline measure of response to similar stimuli might

be advantageous for use as a covariate in the actual exper-

iment to control for individual variability that might ob-

scure treatment effects. Therefore, two "neutral tapes"

were constructed using a male and a female from the first

pilot study who had each obtained the next highest mean at-

tractiveness ratings from the experimental stimulus actors.

These actors (MM and AB) were trained by the experimen-

ter to behave in a "neutral" fashion during the interaction

(i.e., relaxed posture, facing forward in original position

relative to opposite-sex actor, nondescript facial expres-

sion, normal amount of eye contact). Care was taken to

eliminate all relevant positive or negative nonverbal social

cues emphasized in the experimental videotapes from these

sequences. Similarly, the script for these "neutral tapes"

was an abridged version of the experimental scripts, where

all positive and negative comments were eliminated and the

actor terminates the sequence by stating that (s)he'll have

to let the person making the dinner request know at a later

time if (s)he will accept the invitation or not.

Five of the male subjects and ten of the female sub-

jects in the second pilot study viewed their respective

"neutral" tapes and rated them on the six rating scales used

in the experiment. Those subjects who viewed the neutral

tapes did so at the end of the experimental session. The

script for these sequences and the subjects' data are pre-

sented in Appendix XI.

Given the small sample sizes, no statistical analyses

were performed. Nevertheless, the data presented in Appen-

dix XI show that fairly consistent neutral ratings were ob-

tained for the neutral male tape (AB) by the ten female sub-

jects asked to evaluate it. The five male subjects tended

to give more positive ratings to the neutral female tape

(MM) in evaluating attraction, sincerity, and likability.

These apparent differences could be due to (a) the rela-

tively smaller sample size for male subjects, which re-

stricted the range of observations; (b) the possibility that

male subjects were more influenced by the personal attrac-

tiveness of the (female) actor than were female subjects;

or, (c) that a true difference existed between portrayals in

the videotapes. Since these tapes were to be used merely

as covariates in the final analyses, these apparent devia-

tions from "neutrality" were not seen as critical.

Independent Variables

Four independent variables were manipulated in a 2x2x2

x4 split-plot factorial design with three between subjects

factors and one within subjects factor. Between subjects

factors were (1) beverage administration (given alcohol or

given nonalcohol), (2) instructions regarding beverage con-

tent (told alcohol or told nonalcohol), and (3) sex of sub-

ject (male or female). The levels of the within subjects

factor were the four experimental stimuli.

Prior to any manipulations, all male and female sub-

jects viewed the appropriate neutral videotape for their

sex, which was used as a covariate in the data analysis.

Following the beverage and instruction manipulations, all

subjects viewed, in counterbalanced order, the four 90 sec-

ond experimental videotape segments for their sex.

Dependent Measures

Six self-report measures, two recall measures, and

three behavioral measures were used in the study. After

viewing each of the four videotaped interactions, subjects

were asked to make scaled ratings of the actor featured in

each segment in terms of (1) how much the actor seemed to

like the person (s)he was speaking to; (2) how attracted

the actor was to the person he/she was speaking to; (3)

how much the actor seemed to want to go on the date; (4)

how genuine or sincere the actor was about what he/she said;

(5) how personally likable they found the actor on tape to

be; and, (6) if the subject were the person requesting the

date, how likely they would be to request another date with

the actor. The bipolar rating scales used in the actual ex-

periment were identical to those used in the second pilot

investigation (Appendix VIII).

Subjects' ratings were derived and transformed in the

same way as in the pilot investigation.

Following the last of the four videotape presentations,

subjects were asked to recall certain specific aspects of

the final videotape, including (1) a verbatim account of

the stimulus actor's verbal communication during the inter-

action, and (2) the nonverbal behavior of the actor in this

last videotape presentation, which was assessed using a

multiple-choice questionnaire that asked about specific

nonverbal behaviors. On the basis of this information, two

self-report dependent measures were calculated--the number

of "bits" of correctly recalled verbal communications and

the number of correctly recalled nonverbal behaviors in the

last film viewed. The verbal and nonverbal recall tasks

were presented in counterbalanced order across conditions

to control for the effects of delay on recall. In addition,

all videotape conditions were presented in counterbalanced

order across treatment conditions. Thus, verbal and non-

verbal recall data were available for each of the four video-

tape conditions presented last in the series.

The nonverbal recall multiple choice questionnaire was

constructed in the following way. The categories of nonver-

bal behaviors used in training the actors (i.e., eye contact)

were given to two female research assistants, who then

viewed the experimental tapes for each category. These re-

search assistants were not informed of which videotape con-

dition they were viewing. They were then asked to make in-

dependent "counts" of the behavior in each category for each

videotape. Any discrepancies in frequency counts between

the two assistants resulted in a re-viewing of that partic-

ular segment until agreement was reached. On the basis of

these data (Appendix XII), the 11 questions in the multiple

choice questionnaires for male and female subjects (Appen-

dix XIII) were formulated. Four choices were available for

each category of behavior, only one of which was correct

for any given tape.

Following these videotaped presentations subjects were

exposed to a confederate of the opposite sex who interacted

with each subject in a verbally positive but nonverbally

negative fashion. Three behavioral measures of subjects'

interpersonal attraction to the confederate following this

inconsistent communication were collected [(1) duration and

(2) number of eye contacts, and (3) proxemic distance]. The

duration and number of subjects' eye contacts during a two

minute interaction were measured by having the confederate

press an unobtrusive button that activated a recording de-

vice whenever the subject maintained eye contact. Subjects'

proxemic distance from the confederate was measured by the

number of inches subjects chose to place their chair from

the seated confederate.

Training of Confederates

Two male and two female students served as research

assistants in the final experiment. These students were

trained by the experimenter to interact with opposite-sex

subjects in a verbally positive/nonverbally negative fash-

ion. This condition was chosen because previous research

has shown that of the two incongruent verbal/nonverbal con-

ditions used in this study, it results in the most aversive

interpersonal experience for subjects and the greatest in-

terpersonal distancing (Bugenthal et al., 1970; Graves and

Robinson, 1976).

Research assistants viewed the positive verbal/nega-

tive nonverbal segments of the experimental videotapes as

a model for the type of nonverbal behaviors that they would

be required to simulate during the invivo interaction with

subjects. They were given scripts to recite verbatim as

they ostensibly checked over subjects' data (Appendix XIV)

during the positive verbal/negative nonverbal interaction.

They rehearsed with one another and clinical psychology

graduates who served as "practice subjects," until the ex-

perimenter was satisfied with their enactments. Videotaped

feedback was provided during training. A minimum of five

one-hour training sessions were held with each confederate.

Training for recording duration and number of eye con-

tacts was held in conjunction with these sessions. As

seen in Appendix XIV, a predetermined set of topics were

created for confederates to use in his/her two minute con-

versation with subjects. This conversation immediately fol-

lowed the positive verbal/negative nonverbal interaction.

Naive "practice subjects" were used in this training as

well. The experimenter sat behind a one-way mirror, just

behind the confederate and facing subjects with a stopwatch.

She recorded the duration of eye contacts subjects made with

the confederate, who was simultaneously recording the same

on the recording device to be used in the experiment. Re-

liability was defined as a five seconds or less discrepancy

between these two duration counts. Prior to running any ex-

perimental subjects, each confederate trainee was required

to reach a criterion of at least ten consecutive trials of

reliable recordings with the experimenter. Although reli-

ability was defined as a five second or less discrepancy,

all confederate trainees achieved less than three seconds

discrepancy with the experimenter on the ten trial criterion.

It was not possible to check reliability during the ac-

tual experimental procedure. Therefore, after each confed-

erate had run approximately half of the experimental sub-

jects that (s)he had been scheduled to run, independent re-

liability checks (five trials) with the experimenter were

performed as they had been during training, using naive

practice subjects. This procedure was undertaken to assess

any experimental "drift" from reliability that may have oc-

curred between confederates. All confederate trainees re-

mained reliable with the experimenter according to the cri-

terion set previously (a five second or less discrepancy).

Although all confederates trained together and every effort

was made to standardize their role in the experiment

through training, they were not equated on other relevant

variables (i.e., attractiveness). They were, however, coun-

terbalanced across groups of subjects in the experiment.


Subjects were 48 male and 48 female undergraduates,

aged 19 to 28 years enrolled at the University of Florida

(n=12 subjects per cell). Twenty-two of the 48 female sub-

jects and 44 of the 48 male subjects were recruited from

those responding to a research sign-up sheet for an intro-

ductory psychology course in which they received two hours

of experimental course credit for participating in the

study. Twenty-six of the 48 female subjects and 4 of the 48

male subjects were paid $5.00 each for their participation.

Paid subjects' participation was solicited through an ad

in the school newspaper. Only students enrolled at the

University, aged 19-25 years old, and who consumed alcohol

on a fairly regular basis were accepted from those respond-

ing to the ad. The data from two additional subjects (one

male and one female) were discarded due to subjects' knowl-

edge of the experimental deceptions (both in the told non-

alcohol/given alcohol group), as determined on postexperi-

mental questionnaires and during debriefing interviews.

For male subjects, each of the four Beverage x Instruc-

tion treatment groups contained six to eight "heavy,"

three to five "moderate" and one to three "light" social

drinkers, as assessed by the Drinking Practices Question-

naire (Cahalan, Cisin, & Crossley, 1972) and the brief form

of the Michigan Alcoholism Screening Test (SMAST; Pokorny,

Miller, & Kaplan, 1969). Each of the four treatment groups

of female subjects contained four to five "heavy," three

to five "moderate" and three to six "light" social drinkers.

Subjects with significant alcohol problems (as assessed by

the SMAST) or those with any medical conditions which con-

tradicted alcohol consumption were excluded from participat-

ing. All subjects were asked not to consume any alcohol or

use any drugs on the day of their session and to refrain

from eating for two to three hours or drinking any liquid

for 30 minutes prior to their session. Subjects were asked

to bring identification with them to their experimental ses-

sion, verifying that they were at least 19 years of age.


Eligible subjects participated individually in one and

a half to two hour experimental sessions. After the exper-

imenter verified that they were above the age of 19, sub-

jects signed an informed consent form and filled out the

above-mentioned drinking questionnaires. Subjects were

then told that the purpose of the investigation was to study

alcohol's effects on subjective impressions of social in-

teractions. They were told that following beverage

administration in which they would either receive an alco-

holic beverage consisting of vodka and tonic or a nonal-

coholic beverage consisting of tonic only, they would view

four 90 second videotaped interactions in which a same-sex

individual would be asking an opposite-sex actor to dinner.

They were told that after viewing each segment, they would

be asked to make subjective judgments about the actor re-

sponding to the dinner invitation. Subjects were then

given a copy of the rating scales to be used in the experi-

ments and told that they would have an opportunity to prac-

tice this task prior to any beverage administration. They

were told that they would view a videotaped interaction

that was similar in form to the four experimental videotapes.

The experimenter then showed subjects their respective "neu-

tral" videotapes and they made the six scaled ratings (Nl-

N6) to this interaction. The experimenter then entered with

a tray containing a vodka bottle, a tonic bottle, and three


The experimenter instructed subjects that his/her as-

signment to either the alcohol or tonic "control" group

would be randomly determined by having them draw a slip of

paper from a cup containing both alcohol and nonalcohol as-

signment slips. In actuality, subjects in the told alcohol

conditions chose from a cup containing only slips that read

alcohol and subjects in the told nonalcohol condition chose

from a cup containing only slips that read nonalcohol.

Subjects' assignments to one of the four Beverage x

Instruction treatment groups had been randomly determined

prior to their experimental session.

The experimenter than prepared the drinks in the sub-

jects' presence following the procedures discussed in Mar-

latt et al. (1973). Subjects were weighed on a kilogram

scale. The experimenter then consulted a chart to deter-

mine what volume of one-part vodka to five-parts tonic was

required for a person of the subject's weight to reach a BAL

of .05% within 30-40 minutes (the equivalent of .6 g eth-

anol per kg body weight).

In conditions where the subject expected to receive al-

cohol, the experimenter poured 1/6 of the total quantity re-

quired from an 80 proof Smirnoff vodka bottle into a gradu-

ated cylinder and then distributed it equally into three

glasses. This procedure was repeated using five times the

volume from the chilled Schwepp's tonic bottles. Finally,

approximately one ounce of Rose's lime juice was added to

the three cups. When subjects expected tonic only, the

same procedure was followed, except all liquid was taken

from the tonic bottles.

All of the beverages used in the experiment were pre-

mixed according to subjects' actual beverage administration

condition (given alcohol or given tonic) and placed in bot-

tles consistent with subjects' instructional condition.

Thus, for subjects who received alcohol, all of the bottles

seen contained a one to five mixture of vodka and tonic.

For subjects who received only tonic, all of the bottles

seen by the subjects contained only tonic. The tonic water

used in vodka bottles was decarbonated.

Subjects were placed in a screened off area of the lab-

oratory to consume their beverages. This area contained a

chair, table, radio, and magazines. Subjects drank the con-

tents of the three glasses at a rate of one glass every

seven minutes while listening to the radio and reading the

magazines. Following beverage consumption, all subjects

waited 15-18 minutes to allow time for alcohol absorption

for those subjects in the given alcohol treatment group.

During this time, subjects were permitted to use the

restroom and were given false BAL feedback in a "rigged"

breathalyzer test (Alco-Analyzer, Luckey Laboratory). For

subjects in the told alcohol treatment groups, a standard

mixture of 200-proof ethanol and distilled water was ana-

lyzed so that subjects saw a positive reading of .05 on the

apparatus. Subjects in the told tonic groups were shown a

reading of 0 on the breathalyzer machine.

The experimenter then gave subjects four rating scales

(Appendix VIII) for each of the four videotapes. Subjects

were told that they would view the four videotapes, in which

an opposite-sex actor would be responding to a dinner invi-

tation from a same-sex individual. Following these presen-

tations, they would be required to rate the actor in the

film just seen, using the identical rating scales used in

the neutral presentation. Subjects were told to begin mak-

ing their ratings only after they had viewed the entire seg-


As a rationale for the live interaction, subjects were

told at this time, that following the very last videotape

segment, they would be asked to fill out three additional

questionnaires about this last segment of tape only. How-

ever, they would not be permitted to see those question-

naires until after they had completed the six ratings to

this last tape. They were told that many of the questions

in these last three questionnaires required responses which

would have to be coded into categories for the data analy-

sis. The experimenter then informed subjects that her re-

search assistant was doing this coding. They were told

that when they had completed all ratings and questionnaires,

the research assistant would be coming out from a secluded

part of the lab to check over subjects' data to make sure

it could be easily coded. Subjects were instructed that

they might be asked by the assistant to clarify any ques-

tionable responses. They were further informed that the

research assistant was not aware of their group status (al-

cohol or nonalcohol). It was explained that since the ex-

perimenter was aware of both the study's hypotheses and

subjects' group status, she could not question them about

their data, because to do so might lead to an inadvertent

biasing of results, hence the need for a "blind" research

assistant. Subjects were thus given a plausible rationale

for the research assistant's involvement in the experiment.

Subjects then were told that when the research assis-

tant had finished with them, they would be brought to an-

other screened off area to wait with the research assistant

for the experimenter. They would then be given a final

breathalyzer test before leaving the laboratory.

Subjects then viewed the four experimental videotapes

appropriate for their sex, that had been developed during

the pilot investigations, in counterbalanced order. Fol-

lowing each of the videotape presentations, subjects made

the six scaled ratings. For the last videotape presented,

subjects were asked to recall verbatim everything that the

opposite sex actor had said to the same sex individual in

responding to the dinner invitation. The experimenter re-

corded these responses verbatim. Subjects were also asked

to fill out the nonverbal recall questionnaire asking about

specific aspects of the actor's behavior in this last in-

teraction. As noted previously, the two recall tasks were

presented in counterbalanced order across subjects. Fi-

nally, they were asked to write a few lines about what it

was about this last actor that would make them more or less

likely to ask that actor to dinner again if they were mak-

ing the request for dinner. Subjects were told that these

responses were the ones that required coding later on.

Thus, this final task was administered merely to give fur-

ther plausibility to the research assistant's participation

in the study.

A confederate of the opposite sex who was indeed "blind"

to the subjects' experimental condition then entered the

room and enacted a positive verbal/negative nonverbal inter-

action of about two to three minutes in length with sub-

jects (see Appendix XIV). The confederate complimented the

subject's performance and commented positively on the clar-

ity of his/her responses according to a prearranged script,

but avoided eye contact with the subject, frowned, yawned,

sat stiffly, and leaned away from subjects while speaking.

The confederate then asked the subject to follow him/

her into another area to wait for the experimenter to re-

turn. The area that they entered had a desk and chair in

it with a folding chair along the wall by the breathalyzer

machine. The confederate research assistant took the desk

chair and had subjects pull up the folding chair. The dis-

tance between the midpoint of the subjects' chairs and the

middle of the desk constituted the behavioral measure of


While subjects waited approximately three to five min-

utes for the experimenter to return, the confederate inter-

acted "normally" with the subject, maintaining a constant

gaze throughout this interval, and discussing the topics in

Appendix XIV. At the beginning of the interaction, the

confederate research assistant signalled the experimenter

to begin timing the two minute interaction by pressing an

unobtrusive button on the side of their chair which acti-

vated a recording device. Each time the subject met the

research assistant's gaze directly, the research assistant

pressed the button and released it each time subjects

averted their gaze. The experimenter kept a record of each

time the assistant pressed the button and activated the re-

cording device over the two minute period, thus keeping a

running count of the number of eye contacts for each sub-

ject. At the end of the two minute interval, the total

duration of eye contacts was recorded on subject's data

sheets from the reading on the recording device.

The experimenter then entered the area in which sub-

jects sat with the research assistant and took the proxim-

ity measure using a tape measure. Subjects were then asked

to fill out a final questionnaire which asked them to indi-

cate (1) to what group they had been assigned, (2) to esti-

mate the number of ounces of alcohol they had consumed, and

(3) to indicate how intoxicated they felt. These measures

were used to check on the effectiveness of the beverage

content instruction manipulations. The experimenter then

administered a final breathalyzer test to subjects to de-

termine actual BALs, thoroughly debriefed subjects as to

the purpose of the study, the actual beverage they had con-

sumed, and all unobtrusive measures that were made. Any


questions subjects had about the study were answered and

they were permitted to leave.


Overview of Data Analyses

The six scaled ratings, (1) actor liking, (2) actor

attraction, (3) actor's willingness to go to dinner, (4)

actor's sincerity, (5) subject's rating of actor's likabil-

ity, and (6) subject's willingness to date actor) that sub-

jects made to the four videotapes were analyzed separately in

a 2x2x2x4 (Beverage x Instructions x Sex of Subject x Vid-

eotape Condition) repeated-measures analyses of covariance.

The four videotape conditions were a within-subjects factor

and the other three factors were between-subjects factors.

The six pretreatment ratings that subjects made to the neu-

trao tapes (N1-N6) were used as the covariates in these

analyses. The verbal and nonverbal recall measures and the

sum of these measures were examined in 2x2x2x4 (Beverage x

Instructions x Sex of Subject x Recall Videotape Condition)

analyses of variance. In these analyses, Recall Videotape

Condition represented a between-subjects factor. The be-

havioral measures of duration and number of eye contacts

were analyzed in 2x2x2 (Beverage x Instructions x Sex of

Subject) analyses of variance. The proximity measure was

subjected to a 2x2x2 (Beverage x Instructions x Sex of

Subject) analysis of covariance, using subjects' weight as

a covariate to control for the possible differential influ-

ence of subject size on this measure.

Pairwise comparisons of significant main effects and

interactions for between-subjects factors were made using

Duncan's Multiple Range Test with alpha = .05. Significant

interaction effects involving the within-subjects factor

(Videotape Condition) were analyzed in separate analyses of

covariance prior to the Duncan's analyses so that the appro-

priate error term and degrees of freedom were generated for

use in the separate pairwise comparisons involving this

within-subjects factor (Kirk, 1968). Main effect predic-

tions advanced in the introduction were analyzed separately

using planned t-test comparisons (one-tailed).

The slight variation in degrees of freedom for subjects'

ratings of actors' likability occurred because one subject

did not complete this dependent measure for one videotape

condition. For the dependent measure of proximity, the ex-

perimenter failed to record these data for two subjects. Two

other subjects misunderstood the experimenter's instructions

and sat on the other side of the room. Their proxemic dis-

tance from the confederate was well out of the range of that

obtained for all other subjects and could not be attributed

to the operation of any of independent variables in the

model. These data were discarded as well, resulting in a re-

duction of degrees of freedom in the analysis of this depen-

dent measure.

Manipulation Checks

Beverage Consumption Estimates

On the postexperimental questionnaire all told alcohol

subjects indicated that they thought they had received an

alcoholic beverage and all told nonalcohol subjects indi-

cated that they thought they had received a nonalcoholic

beverage, with the exception of two male subjects in the

given alcohol/told tonic group who indicated that they

thought they may have received some alcohol. Nevertheless,

their data were retained in the analyses because (1) unlike

those subjects whose data had been discarded, these two sub-

jects were still uncertain during questioning in the de-

briefing interview of their beverage consumption status and

(2) they indicated that during the data collection phase of

the experiment, following false BAL feedback, they had been

unsure as to what beverage had been consumed. Since all sub-

jects in the given tonic/told tonic control group gave zero

alcohol consumption estimates, a 3x2 (Group x Sex of Sub-

ject) analysis of variance was performed on the alcohol con-

sumption estimates for the 72 subjects in the three experi-

mental treatment groups (given alcohol/told alcohol/ given

tonic/told alcohol/ and given alcohol/told tonic) was per-

formed. This analysis resulted in a significant main effect

for group only, F(2,66)=79.54, -=.001. As predicted, the

pairwise comparisons using Duncan's procedure showed that

the mean estimates for the two groups of told alcohol sub-

jects were significantly higher than that of the given al-

cohol/told nonalcohol group (Given Alcohol/Told Alcohol

M=3.79 oz; Given Tonic/Told Alcohol M=2.42 oz; Given Alco-

hol/Told Tonic M=0.13 oz; MSE=1.04, df=66, p<.05). These

findings support the efficacy of the instruction manipula-

tion. Significant mean differences were also found for the

two told alcohol groups, with subjects in the given alcohol/

told alcohol group estimating that they had received greater

quantities of alcohol (M=3.79 oz) than those in the given

tonic/told alcohol group (M=2.42 oz).

Intoxication Ratings

All 24 of the given nonalcohol/told nonalcohol subjects

gave zero ratings of their intoxication level on the postex-

perimental questionnaire. Fifteen of the 24 given alcohol/

told nonalcohol subjects gave nonzero ratings of their intox-

ication level (group M=2.04), despite a preponderance of re-

ports of having received no alcohol in this group. In the

debriefing interviews with these subjects, they typically

attributed their feelings of intoxication to either expec-

tations associated with participating in an alcohol study,

rapid consumption of carbonated tonic water or both. All

subjects in the given alcohol/told alcohol treatment group

reported positive ratings of intoxication level (group

M=4.79). Five of the given tonic/told alcohol subjects

reported an intoxication level of zero, despite the fact

that all five had estimated that they had consumed some al-

cohol (group M=2.29). These five subjects stated during

the debriefing interviews that they assumed the alcohol had

either not had time to affect them (one subject) or that

they had high tolerance levels (four subjects). Thus, they

were not surprised by not feeling any subjective signs of


With the exception of the given nonalcohol/told nonal-

cohol group, which lacked any variance in their intoxication

ratings (all ratings = 0 in this group), the remaining three

experimental groups were subjected to a 3x2 (Group x Sex of

Subject) analysis of variance. Like subjects' estimates of

alcohol consumption, the only significant effect was a group

main effect F(2,66)=13.38, p=.0001. Pairwise comparisons

of group means resulted in no significant differences be-

tween the given tonic/told alcohol (group M=2.29) and given

alcohol/told tonic (group M=2.04) groups on ratings of in-

toxication level, while the given alcohol/told alcohol group

reported feeling significantly more intoxicated (group

M=4.79) than the two deception groups (MSE=4.15, df=66,

E<.05). When these data are considered in conjunction with

the data for consumption estimates and the verbal attribu-

tions subjects made during the debriefing interviews, they

indicate that, overall, the instruction manipulation was ef-


Final BALs

A 2x2 (Instruction x Sex of Subject) analysis of vari-

ance of given alcohol subjects' BALs at the end of the ex-

perimental session (approximately 50-55 minutes after the

end of the drinking interval) resulted in no significant ef-

fects (grand M=0.032%).

Dependent Measures

Scaled Ratings of Videotapes

Actor liking

This measure involved subjects' ratings of how much

the actor in each videotape seemed to like the person he/

she was interacting with. The model accounted for 81.6%

of the variance [R =.816, F(119,264)=9.81, p=.0001, Grand

M=6.06]. The only significant effect was a main effect

due to the within subjects factor, videotape condition,

F(3,264)=338.17, p=.0001. Duncan's multiple range test re-

vealed significant mean differences between all four video-

tape conditions (see Table 1). The means and groupings in

Table 1 are almost identical to the data obtained in the

second pilot investigation for actor liking ratings (see

Table IX.1, Appendix IX).

No significant effects or interactions were obtained

for Beverage, Instructions or Sex of Subject on this meas-

ure. The covariate rating for the neutral tape did achieve

significance in the model, F(1,87)=11.28, p=.0009.

Table 1. Mean Differences between Videotape Con-
ditions for Actor Liking

Videotape Condition N Mean Grouping*

Positive Verbal/Positive Nonverbal (PP) 96 8.54 A

Negative Verbal/Positive Nonverbal (NP) 96 7.45 B

Positive Verbal/Negative Nonverbal (PN) 96 5.70 C

Negative Verbal/Negative Nonverbal (NN) 96 2.54 D

*Means with the same letter are not significantly different
from one another at p<.05 (MSE=1.95, df=264).

A marginal Beverage x Tape interaction, F(3,264)=2.39,

p=.068, appeared to be the result of given alcohol subjects

rating the actors in the NP condition more positively (M=

7.68) than given nonalcohol subjects (M=7.20).

Actor attraction

This rating asked subjects to judge how attracted the

stimulus actor in each videotape was toward the opposite-sex

person to whom he/she was speaking. The overall model ac-

counted for a significant 80.9% of the variance [R2=.809,

F(119,264)=9.38, p=.0001, Grand M=5.71]. The only signif-

icant effect was a main effect due to the within subjects

factor of videotape condition, F(3,264)=328.16, p=.0001.

The covariate also achieved significance in the overall

model F(1,87)=7.60, p=.006.

Table 2 shows the pairwise comparisons and mean differ-

ences between videotape conditions for the attraction

ratings using Duncan's multiple range test for pairwise


Table 2. Mean Differences between Videotape
Conditions for Actor Attraction

Videotape Condition N Mean Grouping*

PP 96 8.35 A

NP 96 7.45 B

PN 96 4.93 C

NN 96 2.13 D

*Means with the same letter are not significantly different
from one another at p<.05 (MSE=2.29, df=264).

As with ratings of actor liking, all four means for the

videotape conditions were significantly differentiated from

one another.

A comparison of the data in Table 2 and the data show-

ing the mean differences between videotape conditions in the

second pilot investigation in Table IX.1, Appendix IX, re-

veals a difference in means groupings. While the 41 sub-

jects in the pilot study did not differentiate between video-

tapes PP and NP, on ratings of attraction, the data for the

96 experimental subjects in this study suggest stronger con-

sideration of the verbal behavior of the actors in the PP

condition, in evaluating attraction. These 96 subjects

judged actors in the PP condition to be significantly more

attracted during the interaction than actors in the NP vid-

eotape condition.

Actors' willingness to go to dinner

On subjects' ratings of how much the stimulus actor in

each videotape condition wanted to go to dinner, the over-

all model accounted for 80% of the variance [R =.80, F(119,

264)=8.89, p=.0001, Grand M=5.48). A significant main ef-

fect was found for the within subjects factor, videotape

condition, F(3,264)=305.29, p=.0001.

No other significant effects were obtained and the

covariate did not achieve significance in the model, F(1,87)=

.28, p=.60. Duncan's pairwise comparisons for mean differ-

ences between videotape conditions on this measure are shown

in Table 3.

Table 3. Mean Differences between Videotape
Condition for Actor's Willingness
to Date

Videotape Condition N Mean Grouping*

PP 96 8.52 A

NP 96 6.02 B

PN 96 5.78 B

NN 96 1.61 C

*Means with the same letter are not significantly different
at p<.05 (MSE=2.57, df=264).

As seen in Table 3, subjects did not give significantly

different mean ratings of how much the stimulus actor wanted

to go to dinner to the two inconsistent verbal/nonverbal

videotapes. The positive verbal/positive nonverbal video-

tape actors received significantly higher mean ratings of

intention than all other three videotape stimulus actors

and the NN videotape actors received significantly lower

ratings than actors in all other three videotape conditions.

A marginal Sex x Videotape condition interaction,

F(3,264)=2.53, p=.056 was noted. The analysis of this in-

teraction indicated that it resulted from male subjects rat-

ing the actor in the NP condition more positively in terms

of willingness to go to dinner (M=6.45) than did female sub-

jects (M=5.59).

Actor sincerity

The model for subjects' ratings of the sincerity or

genuineness of the stimulus actor accounted for a signif-

icant 56.0% of the variance [R =.56, F(119,264)=2.80, p=

.0001, Grand M=6.18]. Results show a significant main ef-

fect for videotape condition F(3,264)=69.9, p=.0001, and a

significant Beverage x Instruction interaction, F(1,87)=

6.83, p=.01. The covariate, accounted for a marginally sig-

nificant portion of the variance, F(1,87)=2.59, E=.109.

Table 4 shows the pairwise comparisons between video-

tape condition means for subjects' sincerity using Duncan's

procedure. Table 4 shows that subjects did not differentiate

between the two inconsistent verbal/nonverbal videotape con-

ditions (NP and PN) on ratings of sincerity. The stimulus

actors in these two conditions were judged by subjects to be

significantly less sincere than the actors in the consis-

tently positive condition (PP), but more sincere than the

actors in the consistently negative condition (NN). These

data replicate results found in the pilot investigation

(Table IX.1, Appendix IX). They also replicate previous

findings concerning evaluations of sincerity of communica-

tion for incongruent verbal/nonverbal social interactions

(Graves & Robinson, 1976).

Table 4. Mean Differences between Videotape
Conditions for Sincerity Ratings

Videotape Condition N Mean Grouping*

PP 96 8.74 A

NP 96 6.09 B

PN 96 5.91 B

NN 96 3.96 C

*Means with the same letter are not significantly different
at p<.05 (MSE=5.29, df=264).

The Beverage x Instruction interaction was analyzed in

separate pairwise comparisons of the four group means ad-

justed for the covariate in a Least Square Means procedure.

Table 5 shows the results of these analyses.

Table 5. Mean Differences between Beverage and
Instructional Groups for Sincerity of
Communication Ratings

Comparison p value

GATA = 5.84 vs. GATT = 6.23 .040*
GATA = 5.84 vs. GTTA = 6.77 .006*
GATA = 5.84 vs. GTTT = 5.87 .833
GATT = 6.23 vs. GTTA = 6.77 .032*
GATT = 6.23 vs. GTTT = 5.87 .105
GTTA = 6.77 vs. GTTT = 5.87 .004*

*indicates a significant difference.
GA = given alcohol TA = told alcohol
GT = given tonic TT = told tonic

The results in Table 5 show that of the four Beverage

x Instruction treatment groups, subjects in the told alco-

hol/given nonalcohol group gave significantly higher ratings

of sincerity to the videotapes than subjects in any of the

other three groups and subjects in the given alcohol/told

tonic group gave higher sincerity ratings to the tapes than

subjects GATA.

The a priori prediction that alcohol consumption would

result in ratings of greater "sincerity" to the inconsistent

verbal/nonverbal videotape segments was not supported in the

planned one-tailed t-test, t(94)=.64, p=.26.

A marginal Sex x Videotape condition interaction was

found for ratings of actor's sincerity as well, F(3,264)=

2.42, E=.066. The inspection of this interaction suggested

a tendency for male subjects to rate the actor in videotape

condition NP as more sincere (M=6.59) than did female sub-

jects (M=5.59).

Subject's assessment of actor likability

For subjects' ratings of how personally likable the

actor was, the overall model accounted for a significant

69.8% of the variance [R2=.698, F(119,263)=5.11, p=.0001,

Grand M=5.12]. Significant main effects were found for

both videotape condition, F(3,263)=147, p=.0001 and the co-

variate, F(1,87)=4.46, 2=.036. A significant Sex x Tape in-

teraction was also found, F(3,263)=3.73, p=.01.

Table 6 shows the pairwise comparisons between video-

tape conditions for actor likability using Duncan's proce-

dure. These mean differences are qualified by the analysis

of the Sex x Videotape interaction which follows.

Table 6. Mean Differences Between Videotape
Conditions for Ratings of Actor

Videotape Condition N Mean Grouping*

PP 95 8.00 A

NP 96 5.48 B

PN 96 4.86 C

NN 96 2.15 D

*Means with the same letter are not significantly different
at p<.05 (MSE=3.73, df=263).

The pairwise comparisons for Rating 5 show that all

four videotape conditions received mean ratings that were

significantly different from one another, with the actors

in the consistently positive condition (PP) receiving the

highest ratings of likability and the stimulus actors in

the consistently negative condition (NN) receiving the low-

est ratings of likability. In contrast to the data obtained

during the pilot investigation (Table IX.1, Appendix IX),

subjects in the present experiment differentiated between

the two inconsistent videotape conditions and judged the

actors in the negative verbal/positive nonverbal condition

(NP) to be significantly more likable than the actors in

the positive verbal/negative nonverbal condition (PN).

The Sex x Tape interaction was analyzed in two sep-

arate ANACOVAs; one by Sex, for significant

Videotape effects and one by Videotape, for significant

Sex effects. In the ANACOVA by sex, videotape

condition was highly significant for both males [F(3,140)=

95.1, p=.0001] and females [F(3,141)=66.9, =.0001). Table

7 shows the pairwise comparisons between videotape condi-

tion means for both sexes.

The pairwise comparisons show that while male subjects

judged the stimulus actors in all four videotape conditions

to be significantly different with respect to likability,

and rated the NP actor to be significantly more likable than

the actor in the PN condition, females gave equivalent

likability ratings to the stimulus actors in the two incon-

sistent verbal/nonverbal videotape conditions. The analyses of

covariance by videotape condition resulted in a significant

sex effect for condition NP only F(1,93)=5.18, p=.025.

This illustrates that the Sex x Tape interaction found in

the overall model was due to male subjects rating the nega-

tive verbal/positive nonverbal stimulus actor to be signif-

icantly more likable than did female subjects.

Table 7. Mean Differences between Videotape Con-
ditions in Actor Likability for Male
and Female Subjects

Videotape N Mean Grouping* Videotape N Mean Grouping*
Male Female
PP 47 7.83 A PP 48 8.17 A

NP 48 6.15 B NP 48 4.81 B

PN 48 4.84 C PN 48 4.88 B

NN 48 2.09 D NN 48 2.20 C

*Means with the same letter are not significantly different
at p<.05.

Subject's willingness to date actor

Subjects' ratings of the probability of their extend-

ing another dinner invitation to the actor in each video-

tape was also highly significant, accounting for 70% of the

variance for Rating 6 [R2=.70, F(119,264)=5.21, p=.0001,

Grand M=5.11]. Significant main effects for videotape con-

dition, F(3,264)=149, p=.0001 and for the covariate, F(1,264)

=19.4, p=.0001, were found. In addition, there was a sig-

nificant Sex x Videotape interaction F(3,87)=5.87, E=.0008.

No other significant effects were found for any of the other

independent variables.

Mean differences between videotape conditions for this

measure are shown in Table 8. These mean differences are

qualified by the analysis of the Sex x Videotape interaction.

Table 8. Mean Differences between Videotape
Conditions for Rating 6

Videotape Condition N Mean Grouping*

PP 96 8.42 A

NP 96 5.57 B

PN 96 4.97 B

NN 96 1.46 C

*Means with the same letter are not significantly different
at p<.05 (MSE=5.27, df=264).

The pairwise comparisons between videotape condition means

show that subjects indicated they would be most likely to

extend another dinner invitation to the stimulus actors

whose verbal and nonverbal behavior during the interaction

were consistently positive (PP) and least likely to ask the

actors portrayed in consistently negative condition to din-

ner again (NN). However, these pairwise contrasts show

that, in general, subjects gave similar intermediate ratings

of probability of asking again, to the actors in the two in-

consistent conditions.

The Sex x Videotape interaction was analyzed in two

separate ANCOVAs, one by sex, for significant

videotape effects, and one by videotape, for significant

sex effects. Significant main effects for videotape condi-

tion were found in these analyses for both male,

F(3,141)=99.8, p=.0001, and for female subjects, F(3,141)=

68.6, p=.0001. The pairwise comparisons among videotape

condition means for subject's willingness to date the actor,

for male and female subjects, are shown in Table 9.

Table 9. Pairwise Comparisons between Videotape
Conditions for Male and Female Subjects
in Subjects' Willingness to Date Actor

Videotape N Mean Grouping* Videotape N Mean Grouping*
Male Female
PP 48 8.35 A PP 48 8.49 A

NP 48 6.66 B NP 48 4.48 B

PN 48 5.06 C PN 48 4.89 B

NN 48 1.34 D NN 48 1.57 C

*Means with the same letters are not significantly different
at p<.05.

The mean differences between videotape conditions pre-

sented in Table 9 indicate a rating difference between male

and female subjects to the stimulus actors portrayed in the

negative verbal/positive nonverbal videotape condition (NP).

Females gave comparable ratings of probability of extending

another dinner invitation to the actors in both inconsistent

verbal/nonverbal conditions, while male subjects indicated

that they'd be more likely to ask the negative verbal/pos-

itive nonverbal actor (NP) to dinner again than the positive

verbal/negative nonverbal (PN) actor. This apparent sex

difference between means for condition NP, was also sup-

ported in the ANCOVAs by videotape. In these

analyses, the only videotape condition of the four which re-

sulted in a significant main effect for sex was videotape

condition NP, F(1,93)=7.72, p=.007.

Recall Measures

Verbal recall

The verbal recall responses subjects made to the last

videotape condition seen were transcribed verbatim by the

experimenter. Of the 12 statements stimulus actors in each

videotape made to the opposite-sex actor, only those state-

ments which were recalled by subjects verbatim were scored

as correct. Scoring of verbal recall responses was done by

the experimenter, who was blind to the Beverage x Instruc-

tion status of subjects' verbal recall data.

In a 2x2x2x4 (Beverage x Instructions x Sex of Subject

x Videotape Recall Condition) analysis of variance of sub-

jects' verbal recall score, no significant main effects or

interactions were found [R =.15, F(19,76)=0.69, p=.82, Grand

M=5.35). However, a marginally significant effect for In-

structions was found,F(1,76)=3.48, p=.066, with told tonic

subjects (M=5.81) having higher verbal recall scores than

told alcohol subjects (M=4.90).

Nonverbal recall

For the nonverbal recall measure, the number of multi-

ple choice questions that subjects answered correctly out

of a possible 11, for the videotape condition presented

last, constituted subjects' nonverbal recall score. These

questionnaires can be found in Appendix XIV.

In an identical overall analysis of variance, 44% of
the variance was accounted for by the model [R =.44,

F(19,76)=3.10, p=.0002, Grand M=6.9). Significant main ef-

fects were found for both sex of subject, F(1,76)=6.19,

p=.015, and videotape recall condition, F(3,76)=13.8,

p=.0001. Male subjects achieved a significantly higher non-

verbal recall score (M=7.38) than female subjects (M=6.44).

The pairwise comparisons for the four videotape recall con-

ditions are presented in Table 10.

Table 10. Mean Differences between Videotape Re-
call Conditions for Nonverbal Recall Scores

Videotape Recall Condition N Mean Grouping*

PP 26 7.77 A

NP 25 7.16 A

PN 25 4.92 B

NN 20 7.95 A

*Means with the same letter are not significantly different
at p<.05 (MSE=3.41, df=76).

Table 10 shows that subjects correctly recalled fewer non-

verbal behaviors for the PN condition compared to all other

videotape conditions, which did not differ significantly.

Total recall score

The verbal and nonverbal recall scores were added to

derive a total recall score for each subject. Although the

two recall tasks differed procedurally, (one being a free

recall task and one involving a cued recall), because the

order of presentation of these tasks were counterbalanced

across subjects, it seemed reasonable to add the two scores

to assess subjects' total overall recall. The model in the

overall analysis of variance, accounted for only 23.6% of

the variance [R2=.238, F(19,76)=1.24, 2=.25, Grand M=12.26).

Nevertheless, significant main effects for Instructions,

F(1,76)=4.17, p=.045), and Videotape Recall Condition,

F(3,76)=3.61, p=.017, were obtained. Told nonalcohol sub-

jects had significantly higher mean total recall scores

(M=12.9) than did told alcohol subjects (M=11.63). The

pairwise comparisons among videotape recall conditions are

presented in Table 11.

Table 11 shows that subjects' total recall for the two

consistent verbal/nonverbal videotape conditions (NN and PP)

was significantly greater than for the two inconsistent ver-

bal/nonverbal videotape conditions. Of the two inconsistent

verbal/nonverbal videotape conditions, the mean recall for

the positive verbal/negative nonverbal condition (PN), was

significantly lower than that for the negative verbal/pos-

itive nonverbal videotape condition (NP).

Table 11. Mean Differences between Videotape Re-
call Conditions for Sum Recall Scores

Videotape Recall Condition N Mean Grouping*

PP 26 13.04 A

NP 25 12.20 B

PN 25 10.64 C

NN 20 13.35 A

*Means with the same letter are not significantly different
at p .05 (MSE=9.3, df=76).

Thus, for the dependent measures involving recall, no

significant beverage effects were obtained in the overall

model. In addition, the a priori hypothesis that alcohol

consumption would impair subjects' recall was not supported

in the planned one-tailed t-test comparisons for verbal re-

call, t(94)=0.245, p<.40, nonverbal recall, t(94)=0.610,

p<.27, or total recall, t(94)=0.57, p<.28.

Behavioral Measures

Eye contact duration

The model in the overall analysis of variance, for sub-

jects' total duration of eye contact during the two minute

interaction with confederates, accounted for only an

insignificant 6.2% of the variance [R =.062, F(7,88)=0.83,

Z=.56, Grand M=70.5). Only a marginal main effect for Bev-

erage was obtained, F(1,88)=3.01, p=.087. No other signif-

icant or marginally significant results were found for any

of the other factors in the model. The a priori hypothesis

that alcohol consumption would result in longer durations

of eye contact was subjected to a planned one-tailed t-test

analysis. The results of this comparison showed that, as

predicted, given alcohol subjects engaged in significantly

longer durations of eye contact than did given nonalcohol

subjects (see Table 12).

Table 12. Mean Differences between Beverage
Groups for Duration of Eye Contact

Beverage Group N Mean T DF p>T

Given alcohol 48 74.46 1.76 94 .04
Given tonic 48 66.67

The a priori hypothesis that female subjects would engage in

longer durations of eye contact than male subjects was not

supported in the planned one-tailed t-test comparison

t(94)=0.3, p<.38.

Eye contact frequency

The model for frequency of eye contact accounted for

only 9.6% of the variance [R =.096, F(7,88)=1.34, p=.24,

Grand M=26.2). Nevertheless, a significant Beverage x Sex

interaction was found, F(1,88)=5.25, p=.024. Duncan's pro-

cedures by beverage condition and by sex of subject were

performed to test significant differences among means. These

pairwise comparisons are presented in Table 13.

Table 13.

Pairwise Comparisons of Mean Differences
in Frequency of Eye Contact by Beverage
Condition and by Sex

N Grouping* N Grouping*

By Beverage: Given Given
Alcohol Tonic

Males 27.46 24 A 23.08 24 A

Females 25.92 24 A 28.25 24 B

By Sex: Males Females

Given alcohol 27.46 24 A 25.92 24 A

Given tonic 23.08 24 B 28.25 24 B

*Means with the same letter
at p<.05 (MSE=51.4, df=88).

are not significantly different

Table 13 shows that while males and females who con-

sumed alcohol do not differ on mean frequency of eye contact,

females who consumed a nonalcoholic beverage engaged in a

significantly greater number of eye contacts than did males

who consumed a nonalcoholic beverage. Moreover, beverage ad-

ministration influenced the frequency of eye contact of male

and female subjects in opposite directions; i.e., compared

to their respective given nonalcohol conditions, alcohol

consumption resulted in significantly more eye contacts for

males and significantly less eye contacts for females. No

other significant effects were found for this measure.

Average duration of eye contact

To obtain a measure of eye contact, taking frequency

of eye contact into account, subjects' duration of eye con-

tact was divided by number of eye contacts to obtain an

average eye contact duration score. This average was then

subjected to a 2x2x2 (Beverage x Instructions x Sex of Sub-

ject) analysis of variance. The model accounted for only

an insignificant 4.8% of the variance [R =0.48, F(7,88)=

0.63, p=.73, Grand M=3.06) and no significant main effects

or interactions were found.


The number of inches that subjects placed their chairs

from the confederate during the two minute interaction con-

stituted the dependent measure of proximity. Data from two

subjects (one female in the given alcohol/told tonic condi-

tion and one male in the given tonic/told alcohol condition)

were missing due to the experimenter's failure to record

these data. Proximity data from two other subjects (both

males in the given tonic/told alcohol groups) were dis-

carded because these subjects misunderstood that they were

to sit and converse with the confederate until the experi-

menter returned; instead, they placed themselves 88 and 97

inches away, near the breathalyzer machine, to await the

final BAL test. (The range of proximity values for all

other subjects was 3-27 inches.) Thus, proximity data for

only 92 subjects were included in the analysis.

Some subjects who were relatively tall indicated that

placement of their chairs relative to the confederate was

dependent on the length of their legs. No data on height

of subjects were available but weight had been recorded to

determine appropriate beverage quantities. It was assumed

that height and weight are highly correlated, and so weight

of subject was used as a covariate in the overall analysis

to control for any differential effects of subject size on

chair placement. A 2x2x2 (Beverage x Instructions x Sex of

Subject) analysis of covariance of the proximity data re-

vealed no significant group differences and the model did

not account for a significant portion of the variance, [only

3.5%, R =.035, F(8.82)=.37, =.93]. The a priori prediction

that alcohol consumption would result in closer proximities

by subjects was not supported in the planned t-test compari-

son t(90)=.27, p=.40, nor was the prediction that female

subjects would assume closer interpersonal distances than

male subjects, t(90)=.l, =.47.

Pearson correlational procedures were conducted to as-

certain if the behavioral dependent measures of eye contact

(duration and frequency) and physical proximity were assess-

ing a related or relatively independent type of interper-

sonal behavior. Although it had been assumed that the two

eye contact measures and the proximity measure all reflected

interpersonal attraction, Pearson correlations for the whole

sample resulted in no significant correlations between any

of the three behavioral measures. Separate analyses by bev-

erage and sex of subject similarly produced no significant

intercorrelations among any of the three variables.

Verification of confederate research assistants'

Several analyses were performed to verify that the four

individual raters did not elicit differential behavioral re-

sponses from subjects that might confound the treatment ef-

fects. No differences were predicted given the standardized

rater training experiences, rater training to criterion on

reliability, reliability checks, and counterbalancing of rat-

ers across treatment groups. Each behavioral measure (eye

contact, number of eye contacts, and proximity) was analyzed

separately in a four-way analysis of variance to test for

significant effects due to the four confederate trainees.

The results of these analyses showed that no significant ef-

fects due to confederate trainees were present for: dura-

tion of eye contact F(3,92)=.55, p=.65, number of eye con-

tacts F(3,92)=1.92, p=.13 or proximity F(3,88)=.72, p=.55.

In the pairwise contrasts using Duncan's procedure, all con-

federate trainees had nonsignificantly different means for

these three behavioral measures. These analyses verify that

any significant effects due to the independent variables


were not confounded by potential differential responses

that subjects may have made to confederates. They also es-

tablish comparability among confederate trainees.


Summary of Results

The primary focus of this study was to investigate the

potential utility of applying a divided attention paradigm

to study the social-behavioral effects of moderate alco-

hol consumption. Secondary goals were (a) the empirical

development of videotape stimuli portraying consistent and

inconsistent verbal and nonverbal social interactions, and

(b) the examination of any sex differences in responding to

such interactions under both alcohol and .nonalcohol condi-


The present results did not support the primary hypoth-

esis that alcohol consumption disrupts individuals' process-

ing of inconsistent verbal and nonverbal social messages,

in a manner similar to alcohol's effects on divided atten-

tion performance tasks. The a priori predictions concern-

ing the effects of alcohol on the scaled ratings that sub-

jects made to the four videotape conditions were not sup-

ported and no significant beverage main effects were found

on any of these measures. In the only significant interac-

tion involving beverage consumption that was obtained on

the self-report ratings, subjects given nonalcohol but told

alcohol rated all videotape actors as more sincere than did

subjects in either the given alcohol/told alcohol or given

nonalcohol/told nonalcohol groups. No obvious explanation

for this finding is apparent. However, other BPD studies

(e.g., Tucker and Vuchinich, in press) have occasionally

found that subjects in either inconsistent beverage/instruc-

tion condition deviate in responding from those subjects in

the consistent beverage/instruction conditions. Because no

similar results were found on any of the other subjective

ratings, this result will not be emphasized in the discus-

sion of the data. One conceptually interesting marginal

effect, a Beverage x Videotape interaction, was obtained for

actor liking. Given alcohol subjects tended to rate the neg-

ative verbal/positive nonverbal actor more positively than

did given nonalcohol subjects. No other subjective self-re-

port measures were affected by alcohol consumption.

Similarly, the a priori prediction that alcohol consump-

tion would impair subjects' recall of verbal and nonverbal

stimulus cues was not supported. Recall was impaired to

some extent by the belief of having consumed alcohol, but

not by the actual alcohol consumption.

The only a priori prediction that was supported occurred

on the behavioral measures. Compared to given nonalcohol

subjects, given alcohol subjects engaged in longer durations

of eye contact with the opposite-sex confederate, which pre-

sumably reflects a greater degree of interpersonal attraction.

A significant Beverage x Sex of Subject interaction was

found for frequency of eye contact as well. Males exhib-

ited an increase in eye contact frequency under conditions

of alcohol consumption, while females who consumed alcohol

showed a decrease in number of eye contacts with the confed-

erate, relative to nonalcohol conditions.

These results support the utility of investigating al-

cohol's social-behavioral effects within a general informa-

tion processing framework. However, the specific mechanism

proposed to underlie the disruptive influences of alcohol on

subjects' processing of social cues did not receive support

in this study.

Although the results of the analyses of the six self-

report scaled ratings did not support the a priori hypoth-

eses, they did generally replicate the results of the pilot

experiments. Highly significant main effects were found for

the within-subjects factor of videotape condition on all

six ratings.

Across all six dependent measures, the pairwise compar-

isons showed that the positive verbal/positive nonverbal vid-

eotape condition (PP) received significantly greater positive

ratings than all other videotape conditions and that the NN

videotape condition received significantly more negative rat-

ings than the other three videotape conditions. These

two consistent verbal/nonverbal social interactions were

uniformly differentiated from the two inconsistent verbal/

nonverbal videotape conditions across all ratings. These

results support the concurrent and predictive validity

of the experimental videotape stimuli and the replication

of the original pilot data establishes their reliability.

The pairwise comparisons between videotape conditions

were modified to some extent by sex of subject. Significant

or nearly significant Sex x Videotape Condition interactions

were present in the analyses of four of the self-report rat-

ings. These interactions resulted from male subjects rating

the negative verbal/positive nonverbal stimulus actor more

positively than did female subjects. For male subjects, the

positive nonverbal behavior of the actor in this particular

videotape condition was given more weight in these ratings,

which were significantly more positive than ratings made to

the positive verbal/negative nonverbal videotape. In con-

trast, female subjects gave equivalent ratings to both in-

consistent verbal/nonverbal stimulus actors on these four

dependent measures. These same sex differences were noted

in the pilot data.

In the analyses of the three recall measures, instruc-

tions regarding beverage content produced a marginal tendency

for told alcohol subjects to recall less of the verbal con-

tent of the videotape interactions and to obtain signif-

icantly lower total recall scores than told nonalcohol sub-

jects. Sex of subject had a significant effect on subjects'

nonverbal recall score, with males scoring significantly

higher on this task than females. This was surprising, in

light of the a priori prediction that female subjects would

recall more of the actors' nonverbal behaviors than male


The videotape stimulus condition that subjects were

exposed to also exerted a significant effect both on their

nonverbal recall scores and the sum of both verbal and non-

verbal recall scores. In general, subjects exposed to the

two consistent verbal/nonverbal videotape conditions accu-

rately recalled more of the nonverbal behaviors in these

videotapes and achieved significantly higher total recall

scores than subjects exposed to the two inconsistent video-

tapes. On both of these dependent measures, subjects who

responded to the positive verbal/negative nonverbal video-

tape (PN) scored significantly lower than subjects respond-

ing to the other three videotape conditions. While exposure

to the negative verbal/positive nonverbal videotape (NP) did

not depreciate subjects' nonverbal recall relative to the

two consistent verbal/nonverbal videotapes, the mean total

recall for this videotape condition was significantly lower

than the mean total recall of the two consistent verbal/non-

verbal videotapes (PP and NN).

In the results of the analyses involving the behavioral

measures, no significant beverage, instruction, or sex ef-

fects were found for the dependent measure of proximity. As

previously described, only actual beverage consumption had a

significant influence on subjects' duration and frequency of

eye contact with the confederate.


Beverage Administration

Alcohol consumption had virtually no effect on the sub-

jective inferences that subjects made about the behavior of

the actors in the four videotape conditions or on their re-

call of specific information from the interactions. These

self-report data contrast substantially with the behavioral

eye contact measure wherein alcohol consumption resulted in

longer durations of eye contact with the confederates. The

previous literature strongly suggests that this effect re-

flects subjects' greater interpersonal attraction to the con-

federate (Harper et al., 1978). This finding of alcohol-in-

duced behavioral changes contrasts with previous balanced

placebo design studies that assessed postconsumptive behav-

ioral responses associated with emotional changes (i.e., Lang

et al., 1975; Vuchinich, Tucker, & Sobell, 1979) and found

only instructional main effects. A possible reason for this

discrepancy concerns the stimulus materials used. In these

earlier studies, the incoming information or social stimuli

that subjects were exposed to were consistent and constant

across communication channels. In contrast, in this study,

the confederates' verbal and nonverbal behavior towards sub-

jects was notably inconsistent and significant beverage ef-

fects were obtained for eye contact duration. Given the ef-

fectiveness of the present instructional manipulations, this

finding strongly suggests that, at least at a behavioral

level, subjects were responding differentially to the in-

consistent stimulus information in the interaction with the

confederate, depending on whether they had consumed an al-

coholic or nonalcoholic beverage.

Because only one inconsistent interaction (PN) was in-

cluded, it is possible that alcohol consumption simply leads

to greater interpersonal attraction in general or that it

increases relative responsiveness to all verbal cues (posi-

tive or negative), rather than reflecting effects due to in-

consistencies in the verbal and nonverbal channels. To as-

sess these alternative interpretations, at a minimum, a con-

sistent verbal/nonverbal interaction with a confederate also

would have to be included as a basis for comparison. Al-

though future research should investigate these possibil-

ities by including the necessary stimulus conditions, a study

by Lindman (1980) tends to rule out the interpretation that

alcohol consumption generally increases attraction. Lind-

man exposed subjects to consistent stimulus cues and as-

sessed their eye contact duration under conditions of al-

cohol and nonalcohol administration. The results obtained

showed that intoxicated subjects tended to engage in less

overall eye contact than did sober subjects. Although Lind-

man's study (1980) differed from the present research in fun-

damental ways other than the absence of conflicting inter-

personal cues (i.e., higher dose levels, no instructional

set conditions, and pretreatment assessment of eye contacts),

these results do not refute the possibility that the bever-

age effects observed in this investigation for eye contact

duration may have been due to differences in subjects' pro-

cessing of inconsistent stimulus information, rather than

to differential changes in overall attraction.

The differential effects of alcohol on male and female

subjects' eye contact frequency with the confederate are

somewhat more difficult to interpret, primarily because the

meaning of this measure is less obvious than for the dura-

tion measure. Greater eye contact frequencies may imply

greater affiliative or avoidance behavior if one considers

that this measure also represents the number of averted

glances during the interaction. The lack of a significant

correlation in either direction between duration and fre-

quency of eye contacts for either the entire sample or

within relevant subgroups further obscures the meaning of

the eye contact frequency measure. Although duration of

eye contact is almost always associated with greater inter-

personal attraction, the interpersonal significance of eye

contact frequency or "glancing behavior" has been subject

to both positive and negative interpersonal interpretations

depending on the nature of the study (reviewed in Harper et

al., 1978). That differences exist between the sexes in

eye contact frequency and that frequency of eye contact may

serve different functions for males and females during an

interaction has also been suggested in previous studies in-

volving visual behavior (Exline & Winters, 1965; Coutts &

Schneider, 1975). In the Exline and Winters study (1965),

females were found to engage in more shifts in looking while

talking than males. Coutts and Schneider (1975) obtained a

similar sex difference when males and females were placed

together in an experimental situation where talking was pro-

hibited. These findings of greater eye contact frequency

for female subjects were replicated in the given nonalcohol

groups in the present investigation. Why alcohol consump-

tion resulted in an increase in this behavior for males and

a decrease for females relative to nonalcohol conditions is

difficult to say.

If one assumes that increases in this behavior signify

greater interpersonal attraction and comfort, the specula-

tion would be that males felt more at ease and attracted to

the confederate following alcohol consumption, and were more

likely to express this behaviorally than females. Con-

versely, females given alcohol may have experienced some

discomfort or embarrassment in interacting with an opposite

sex confederate which, in conjunction with the physiolog-

ical sensations of the alcohol, may have led them to dif-

fuse their anxiety by engaging in less of this behavior than

normal. On the other hand, if one accepts the premise that

frequent glancing represents an avoidance behavior, it

would appear that males given alcohol experienced greater

discomfort during the interaction than females. Another

possibility, which makes more sense when the eye contact

duration results are considered is that glancing serves

two separate interpersonal functions for men and women

(signifying positive affect for males and negative affect

for women). Thus, under conditions of alcohol consumption,

both sexes were engaging in more affiliative interpersonal

behavior with the confederate. This interpretation has less

appeal in light of the lack of significant correlations be-

tween duration and number of eye contacts within each sex.

Clearly, this issue will require further exploration. A co-

herent resolution cannot be made on the basis of these data

alone. Nevertheless, sex differences found in the present

investigation concur with earlier reports (i.e., Abrams &

Wilson, 1979; Wilson & Abrams, 1977) of male/female differ-

ences in laboratory studies of alcohol's social-behavioral

effects and they re-emphasize the need for further explora-

tion of sex differences in this area as well.

If subjects were responding differentially, at a behav-

ioral level, to the inconsistent behavior of the confederate

during the interaction under conditions of alcohol and non-

alcohol administration, one must question why these differ-

ences were not evidenced at an inferential level in the self-

report ratings made to the videotape or in recall? One ex-

planation could be that as passive observers, subjects' per-

ceptual processing of verbal and nonverbal stimulus

information was unaffected by the alcohol dose used in this

experiment. In contrast, subjects may have been more sensi-

tive to alcohol-induced disruptions in processing when they

were required to actively participate in such interactions.

In this latter situation, where subjects were required to

simultaneously attend to the incoming stimuli and respond to

the confederate, this added response requirement may have

served as additional interference to disrupt processing.

Another possibility is that the within-subjects video-

tape manipulation became obvious to subjects in the repeti-

tious viewings. Discrepancies in actors' verbal and nonver-

bal behavior may have been fairly easy to detect when pre-

sented in conjunction with the two consistent verbal/nonver-

bal interactions. These consistent videotape conditions may

have served as "anchors" to cue subjects about the discrep-

ancies in the two remaining videotape conditions, thus mak-

ing them less susceptible to the beverage manipulation. In

the isolated interaction with confederates, no comparisons

were possible. Thus, the discrepant nature of his/her com-

munication may have been more subtle to subjects.

The lack of significant beverage effects for the prox-

imity measure was also curious in light of the beverage ef-

fects obtained for the eye contact measures. Two possible

explanations for the lack of significant beverage effects

on this measure, other than its insensitivity to beverage

effects, involve certain methodological limitations. One

possibility is that individual variability in seating prox-

imity was so high as to obscure any alcohol treatment ef-

fects. No pretreatment assessment of proximity was made to

control for the potential effects of individual variability

and this may be important in future research. Another expla-

nation is that having the confederate sit behind a desk re-

sulted in enough of a "barrier" for subjects (given nonal-

cohol) who may have found his/her behavior somewhat aver-

sive, that further distancing became unnecessary and so no

discriminable response occurred across beverage conditions.

Beverage Content Instructions

In general, instructions regarding beverage content did

not influence subjects' responses, with the notable excep-

tions of its effects on subjects' recall and liking ratings.

For total recall, an instruction main effect was found.

Marginal instruction effects appeared in subjects' verbal

recall responses and in their ratings of actor liking. Re-

call accuracy was impaired in both cases by subjects' belief

of having consumed alcohol and ratings of actor liking were

relatively more positive for subjects who believed that they

had consumed an alcoholic rather than a nonalcoholic bever-


At a general level, these findings support the use of

instructional set control conditions in research on alco-

hol's effects and are consistent with other investigations

of postconsumptive behavioral changes (Lang et al., 1975;

Vuchinich et al., 1979). These instructional effects for

recall are also consistent with data obtained in a related

study on memory for social interactions following moderate

alcohol consumption (Tucker, Vuchinich, & Schoenhaut, un-

published manuscript).1 Typically, such instructional ef-

fects are explained in terms of subjects' expectations re-

garding alcohol's effects (Marlatt et al., 1973). In the

present recall tasks, where it was more obvious that correct

and incorrect responses were possible than on the subjective

ratings made to the videotapes, subjects apparently expected

alcohol to impair their recall accuracy. This expectancy

may have resulted in reduced effort to memorize the stimulus

materials or produced deficits at the level of retrieval to

depreciate recall scores. The lack of any significant main

or interactive beverage effects may indicate that alcohol

consumption did not affect subjects' recall. However, this

is surprising, given the extant research demonstrating imme-

diate memory deficits following alcohol consumption (Ryback,

1973). Quite possibly, the absence of any significant bev-

erage effects was again the result of repeated viewing of

four very similar videotape conditions. This repetition may

have simplified recall tasks for subjects to such an extent

that they were able to overcome any potentially debilitating

effects of alcohol consumption on memory.

Effects of Videotape Condition and Sex of Subject

In addition to the beverage and instruction effects

that were obtained, several other salient results were found

that were unrelated to either variable. First, the pretreat-

ment ratings that subjects made to the neutral videotapes

were significant as covariates in the analysis of all but

two of the six self-report measures. These results suggest

that pretreatment ratings were important in this study and

may be important to include in similar studies to control

for individual variability in subjective attributions. It

may be one reason why significant Sex x Videotape interac-

tions resulted in this study, but not in the pilot data

analyses. Second, the significant main effects for video-

tape condition in this study and in the pilot data estab-

lish the validity and reliability of the four experimental

videotape stimuli and support their use in future experiments

involving verbal and nonverbal channels of communication.

Visual inspection of the pairwise comparisons in the mean

self-report ratings for the pilot and experimental studies

suggest that there were slight differences in ratings of

actors' attraction, willingness to go to dinner, subjects'

ratings of actors' likability, and ratings of subjects' will-

ingness to date the actor portrayed in the videotape. These

differences basically amounted to differences in subjects'

ratings in both studies to the negative verbal/positive non-

verbal videotape condition. These results and the sex

differences discussed below suggest that this videotape con-

dition may be subject to the most variability among subjects'


The larger sample size used in the experimental study

and the use of the covariate in the analysis of these sub-

jects' data suggests that the patterning of mean differences

between videotapes for the experimental subjects probably

represents a closer estimate of the population means and

differences associated with the videotape conditions than

does the pilot data.

Other noteworthy findings occurred in the significant

and marginally significant Sex x Videotape interactions in

the experimental study. For the two evaluative ratings that

subjects made concerning their personal responses to actors'

likability and their willingness to date the actor in each

videotape condition, significant sex differences resulted

from the ratings made to the NP videotape condition. Males

gave the actor in this condition more positive ratings than

did female subjects. This same trend was noted for subjects'

ratings of actors' willingness to go out to dinner and the

perceived sincerity of communication. Given the equivalency

of the two NP videotape conditions shown to male and female

subjects, it seems reasonable to conclude that male subjects

tended to focus on, and give more weight to, the nonverbal

behavior of the actor when this behavior was positive than

did female subjects. The finding in this study that male