acquaintance and visual behaviour between two interactants. their communicative function for the...

17
European Journal o f Social Psychology, Vol. I I , 409425 (I 981) Acqualntance and Visual behavkur between two interactants. Their communtcatlve function tor the impredon formation of an observer* ANDREA ABELE University of Bielefeld Abstract Predictions concerning the formation of impressions on the urnis of the visual behaviour of an observed person are tested in two experiments. Observers were shown a video recording of a conversation between two persons in which person (A) looked at his partner (B) or not depending upon the experimental conditions. In addition, the observers were told whether or not the persons were acquainted with one another before this conversation. It was found, as predicted, that A was rated higher with regard to friendliness und openness if he looked at his conversation partner. Equally in agreement with prediction it was found that the significance of visual behaviour for the formation of impressions is greater if it is known that the observed persons are well acquainted. Ratings of A's dominance and activity are only dependent on the visual behaviour shown if it is known that the observed persons know eath other. In the case in which the partners are not acquainted, there is no such effect. These results show that the communicative function of visual behaviour over and above the ascription of friendliness and openness is increased if contextual in formation is given concerning the interpersonal relationship between the persons observed. The formation of impressions concerning the second person involved in the conversation is not influenced by the visual behaviour ofhis conversationpartner. INTRODUCTION Within the framework of research into non-verbal behaviour and its effect upon social judgement and social interaction processes, the investigation of visual behaviour, i.e. length of glances, mutual gaze and frequency of glances, occupies a central position (see Argyle and Cook, 1976; Harper, Wiens and Matarazzo, 1978). One branch of research with which this present paper is concerned analyses the visual behaviour of an observed person as an independent variable for 'This research was supported by a grant (number 86/2762) from the University of Bielefeld. I am grateful to K. Juhl and W. Nowack for critical comments on an earlier version of this paper and to W. Thiel, who did the computations. 0046-2772/81/040409-17%01.70 0 1981 bv John Wiley & Sons, Ltd. Received 20 February I980 Revised 20 February I981

Upload: dr-andrea-abele

Post on 11-Jun-2016

214 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Acquaintance and visual behaviour between two interactants. Their communicative function for the impression formation of an observer

European Journal o f Social Psychology, Vol. I I , 4 0 9 4 2 5 ( I 981)

Acqualntance and Visual behavkur between two interactants. Their communtcatlve function

tor the impredon formation of an observer*

ANDREA ABELE University of Bielefeld

Abstract

Predictions concerning the formation of impressions on the urnis of the visual behaviour of an observed person are tested in two experiments.

Observers were shown a video recording of a conversation between two persons in which person (A) looked at his partner (B) or not depending upon the experimental conditions. In addition, the observers were told whether or not the persons were acquainted with one another before this conversation.

It was found, as predicted, that A was rated higher with regard to friendliness und openness if he looked at his conversation partner. Equally in agreement with prediction it was found that the significance of visual behaviour for the formation of impressions is greater i f it is known that the observed persons are well acquainted. Ratings of A's dominance and activity are only dependent on the visual behaviour shown i f it is known that the observed persons know eath other. In the case in which the partners are not acquainted, there is no such effect.

These results show that the communicative function of visual behaviour over and above the ascription of friendliness and openness is increased i f contextual in formation is given concerning the interpersonal relationship between the persons observed. The formation of impressions concerning the second person involved in the conversation is not influenced by the visual behaviour ofhis conversation partner.

INTRODUCTION

Within the framework of research into non-verbal behaviour and its effect upon social judgement and social interaction processes, the investigation of visual behaviour, i.e. length of glances, mutual gaze and frequency of glances, occupies a central position (see Argyle and Cook, 1976; Harper, Wiens and Matarazzo, 1978). One branch of research with which this present paper is concerned analyses the visual behaviour of an observed person as an independent variable for

'This research was supported by a grant (number 86/2762) from the University of Bielefeld. I am grateful to K. Juhl and W. Nowack for critical comments on an earlier version of this paper and to W. Thiel, who did the computations.

0046-2772/81/040409-17%01.70 0 1981 bv John Wiley & Sons, Ltd.

Received 20 February I980 Revised 20 February I981

Page 2: Acquaintance and visual behaviour between two interactants. Their communicative function for the impression formation of an observer

410 Andrea Abele

impression formation and attributions about this person. The general question is, ‘What does visual behaviour communicate to the receiver?’ (see Ellsworth and Ludwig, 1972). Experiments work either with confederates who interact with the subject, displaying different visual behaviour, i.e. the receiver is in this case the direct interaction partner (Argyle, Lefebvre and Cook, 1974; Cook and Smith. 1975; Ellsworth and Carlsmith, 1968), or with film or video recordings of an interaction from which the subjects are asked to form their impression, i.e. the receiver is an external observer (Dorris, Lane and Sharkey, 1975; Thayer and Schiff, 1974). In most experiments, response is measured on semantic differential scales with respect to the dimensions of evaluation, activity and potency. In the interpersonal context, these dimensions have shown themselves to be significant not only for the evaluation of visual behaviour but also for the evaluation of other non-verbal signs (see Gitin, 1970; Schlosberg, 1954).

The present paper researches the formation of impressions by an external observer of’two persons who are observed in a video recording of a conversation. Depending upon the experimental conditions, one of the two persons displays varying visual behaviour. Beside analysing the general formation of impressions made by visual behaviour, this paper is particularly concerned with the potential moderating effect of contextual information concerning the length of the relationship between the persons observed. Furthermore, the connection between the supposed and actual effects of gaze on impression formation is studied.

Impression formation dependent upon the visual behaviour of the person observed

There already exists a whole series of results concerned with the effects of the gaze of a person observed upon the impression formation of the observer. One frequently recurring result is that a person is rated more positively when he looks at his partner frequently, than when he looks at him infrequently (Argyle et al., 1974; Cook and Smith, 1975; Exline and Eldridge, 1967; LeCompte and Rosenfeld, 1971; Kleck and Nuessle, 1968; Kleinke, Meeker and Fong, 1974). Thus we formulate hypothesis 1: When a person looks at his conversation partner frequently, he is rated more positively than when he looks at him infrequently.

Moreover, a person who looks frequently at his partner, or who looks with few, but long glances at him, is often rated as being more dominant than a person who looks at his partner infrequently, or with many, short glances (Argyle et al., 1974; Argyle, 1975; Cook and Smith, 1975; for female, but not for male observers: Kleck and Nuessle, 1968). So we formulate hypothesis 2: When a person looks at his partner frequently, he is rated as being more dominant than when he looks at him infrequently.

Finally, there are certain results which show that ratings of activity and attentiveness also vary with gaze. These ratings are higher when the observed person looks frequently at his partner (Argyle et al., 1974, only for men; Kleinke, Bustos, Meeker and Staneski, 1973). Hypothesis 3 is: When a person looks at his partner frequently, his activity in the interaction is rated higher than when he looks at him infrequently.

Page 3: Acquaintance and visual behaviour between two interactants. Their communicative function for the impression formation of an observer

Visual behaviour 41 1

Contextual information and impression formation dependent on visual behaviour

It is generally assumed that the effect of gaze upon impression formation is modified by the context; i.e. it may happen, under certain quite specific contextual conditions, that a person who gazes frequently at his partner is rated more negatively than a person who gazes infrequently at him (Ellsworth and Carlsmith, 1968). The question, ‘What does visual behaviour communicate to the receiver?’ is formulated more specifically as, ‘What does visual behaviour communicate to the receiver, and under what conditions?’ Earlier studies have demonstrated such moderating effects to a varying degree (Argyle et al., 1974; Dorris et al., 1975; Ellsworth and Carlsmith, 1968; Ellsworth, Friedman, P e r k and Hoyt, 1978), but no criteria have been mentioned for the selection of contextual stimuli worth studying. Which contextual stimuli are in fact relevant to the effects of visual behaviour upon impression formation? Corresponding to the results showing gaze to be a signal for liking, attentiveness and dominance, it may be supposed that a moderating effect is exercised in particular by such contextual stimuli as convey information concerning the interpersonal relationship of the persons observed; for by such means on another plane, additional information may be conveyed concerning liking and/or dominance. It is by means of such information that standards of conventional non-verbal behaviour are set up which, serving as a basis of comparison for observed behaviour, emphasize unconventional non-verbal behaviour all the more. This supposition does not, of course, imply that such comparison is made consciously. It is possible that visual behaviour functions as a well-learned discriminative stimulus which is particularly accentuated by contextual information concerning the interpersonal relationship.

Information concerning the interpersonal relationship may be either qualitative (e.g. r61e relationships, differences in status) or quantitative. The present study concerns itself with one quantitative aspect of the interpersonal relationship, the length of the acquaintance between the two persons observed. Infrequent visual contact between persons who know each other well is possibly much more striking than belween persons who are talking to one another for the first time, for in general, well-acquainted persons have a higher non-verbal intimacy, of which visual behaviour is a part, than strangers do (Coutts and Schneider, 1975; Gatton, 1970; Rubin, 1970). Up to now, one study by Dorris et al. (1975) has been conducted, which shows that observers’ ratings of the mutual liking of two persons observed are dependent on the non-verbal intimacy shown if they have been told that the persons know each other well. If. on the other hand, they have been told that the two persons are speaking to each other for the first time, ratings of liking are the same for different levels of non-verbal intimacy, i.e. non-verbal intimacy does not function as a signal for liking. Hypothesis 4 reads: If the observer has been told that the persons observed know each other well, then the observed persons’ visual behaviour is more important for the formation of impressions than in the case in which he has been told that the persons do not yet know each other.

The amount of gaze of one of the interaction partners does not only convey information about that person but also concerning the relationship between the two persons. It may also happen that frequent gazing by one interaction person makes the second person involved appear to an external observer to be friendlier and niore likeable than would otherwise be the case. Hypothesis 5 reads: If a person is

Page 4: Acquaintance and visual behaviour between two interactants. Their communicative function for the impression formation of an observer

412 Andrea Abele

frequently gazed at by his interaction partner, he is evaluated more positively than would be the case if he were gazed at little. With regard to the other two response dimensions, dominance and attentiveness, it is not possible to formulate any hypotheses.

EXPERIMENT I

Method

Subjects

Forty male students (age range between 20 and 31 years) served as subjects in the first experiment. They took part in groups of two or three.

Procedure

In the instruction, subjects were told to look at a 3-minute video recording showing a talk between two fellow students called A and B and, after that, to give their impressions about the observed persons. Ratings about person B had to be given first. One trial lasted for about 10 minutes.

Stimulus material

Video recordings on a talk between two students who did not know the experimental hypotheses served as stimulus material. Both partners were talking and listening for about the same amount of time. It was an everyday talk about student life.

The interactants were seated at adjacent sides of an 80-cm square table. Interactant B could be seen in profile, interactant A full face. It was clearly distinguishable whether A looked at B or not. Whether B looked at A or not was not clearly distinguishable. However, he always looked at A, to make sure that every gaze from A to B was a mutual gaze. Except for A’s visual behaviour, all the other non-verbal behaviours of the interactants were kept as similar as possible throughout the recordings. This was especially difficult with smiling and facial expression, because little gaze usually means less smiling and expressiveness than frequent gaze. Also, it was not possible to keep A’s head position completely constant between the two films.

Independent variables

In a 2 x 2 factorial design, A’s visual behaviour towards B and the information whether A and B knew each other before the conversation or not were varied. There were 10 subjects in each condition. Every subject saw one of the recordings under one of the acquaintance conditions. The information about acquaintance was given as follows: No prior acquaintance: ‘ . . . We met the students in the cafeteria and asked them to take part in this study. They had not talked to each other before. They do not know each other. . . .’ Prior acquaintance: ‘. . . We met the students in the cafeteria and asked them to take part in this study. They had talked to each other many times before. They know each other well. . . .’

Page 5: Acquaintance and visual behaviour between two interactants. Their communicative function for the impression formation of an observer

Visual behaviour 4 13

In order to vary A’s visual behaviour, the conversation was recorded in two versions. Under the condition of ‘no gaze’ A did not look into B’s face. Under the condition of ‘frequent gaze’ A looked at his partner continuously. B always looked at A. These relatively artificial conditions-normally there is about 60 per cent gaze averaged over talking and listening in a dyadic interaction-was chosen in order to make sure that the different gaze patterns really were distinguishable.

Measures

The dependent variables were ratings of the two interactants on ten 7-point bipolar adjective scales. The scales were chosen both to represent the dimensions of evaluation, dominance and activity (see Osgood, Suci and Tannenbaum, 1957) and to refer to aspects of social interaction. After inspection of the literature (Cohen, 1969; Hofstatter, 1963; Tedeschi and Lindskold, 1976), the evaluative dimension was operationalized by ‘friendly-unfriendly’, ‘comradely-uncomradely’ and ‘open-reserved’. The dominance dimension was operationalized with ‘powerful-powerless’, ‘superior-inferior’ and ‘dominant-submissive’. The activity dimension was formulated as compliance respectively interaction readiness and was operationalized with ‘compliant-uncompliant’, ‘talkative- not talkative’, ‘helpful-unhelpful’, and ‘reliable-unreliable’. Additionally, subjects had to rate several dimensions of the interactants’ non-verbal behaviour (gaze, facial expression, gesture, smiling), and they had to indicate how much they had been influenced in their ratings by these behaviours. These measures, on the one hand, were taken to test the equivalence of the two video recordings. On the other hand, they were taken to study subjects’ attentiveness towards the visual behaviour of person A. Finally, subjects had to indicate how difficult it had been for them to rate the observed persons and how pleasant the experiment had been for them. At the end of the experiment, subjects had to tell whether the interactants had been acquainted with each other or not.

Results

Manipulation checks

Both the question concerning gaze of interactant A and the question concerning the degree of acquaintance between the observed persons showed highly significant differences between the experimental conditions in the direction of the manipulations.

Control variables

Under the condition of frequent gaze, smiling of A was rated as having been more intense than under the no gaze condition (F(1/36) = 20,85, p < 0.001; frequent gaze: M = 3,75, no gaze: M = 2,OO). Smiling of person B, who had shown the same behaviour through all conditions was rated as having been more intense under A’s frequent gaze too (F(1/36) = 5 ’ 7 6 , ~ C 0.05; frequent gaze: M = 2’75, no gaze: M = 1’95). Also the information concerning prior acquaintance between the interactants was significant for the ratings of A’s and B’s d i n g . Under the no prior acquaintance condition, ratings of A’s and B’s smiling were higher than under

Page 6: Acquaintance and visual behaviour between two interactants. Their communicative function for the impression formation of an observer

414 Andrea Abele

the prior acquaintance condition (A: F(1/36) = 4,40, p < 0.05; acquainted: M = 2’00, not acquainted: M = 2’80; B: F(1/36) = 4S5, p C 0.05; acquainted: M = 2,50, not acquainted: A4 = 3’35).

Analyses of the other control variables did not show any effects of the experimental conditions. With the exception of smiling ratings, it thus can be assumed that the reception of the two video recordings had been very similar. The ratings for the ease of impression formations as well as for the pleasantness of the experimental situation did not differ between the conditions.

Correlations between the ratings

In order to test whether the a prwri formulated dimensions of ‘positive evaluation’, ‘dominance’ and ‘compliance’ really do show up, correlations between the respective variables were computed. There was no factor analysis of the ratings, because factor-analytically determined dimensions would not have been independent of the specific experimental situation. The mean correlation between the ratings of how friendly, comradely and open the interactants had been was r = 0.48 (p < 0.01). The mean correlation between ‘powerful’, ‘superior’ and ‘dominant’ was r = 0.46 (p < 0.01). The mean correlation between ‘compliant’, ‘reliable’, ‘talkative’ and ‘helpful’ was r = 0.37 (p < 0.01). In the following, both the results upon the adjective scales and upon the combined dimensions are mentioned.

Ratings of interactant A

The results for the ratings of person A are shown in Table 1. In order to test the hypotheses 1 to 4 univariate comparisons between cell means were also computed (see Table 2). The results in Tables 1 and 2 show that hypothesis 1 has been

Table 1. Analyses of variance for ratings of person A (N = 40)

Factor

A (acquaintance B (gaze Variable condition) df 1/36 condition) df 1/36 A x B MSE,

Friendly Comradely

‘Positive evaluation’ Powerful Superior Dominant ‘Dominance’ Compliant Talkative Helpful Reliable ‘Comdiance’

Open

F < 1 F = 5.55’ F < 1

F = 2,16 F < 1 F = 1,96 F < 1 F < 1 F < 1 F < 1 F < 1 F < 1 F < 1

F = 4,20’ F = 16.30t F = 7,12t

F = 15,OOt F = 2.22 F = 1,31 F < 1 F < 1 F = 2,22 F < 1 F = 2,79 F < 1 F = 1.40

F = 2,14 F = 2,83 F < 1

F = 2,57 F < 1 F < 1

F = 2.07 F = 11,46t F < 1 F = 5.18’ F = 3,50 F = 6.69’

r = 2,5i

42,OO 31,80 50,60

22,42 79,40 55.70 70,20 40,68 49.10 108,80 39,lO 56,19 29.27

*p < 0.05; t p C 0.01.

Page 7: Acquaintance and visual behaviour between two interactants. Their communicative function for the impression formation of an observer

Visual behavwur 415

Table 2. Mean ratings of person A in the four experimental conditions

Acquaintance No acquaintance

Rating Gaze No gaze Gaze No gaze

Friendly 5,40* 4’20 4,80 4,60 Comradely 5,60* 3.90 4.40 3,70 Open 5,30* 4.10 4.80 4.00 Powerful 4,lO 3.80 4.60. 330 Superior 4.60 4.50 4,40 3.60 Dominant 3.50 4,40 4.40 3.90 Compliant 5.30(a) 3,50(b) 4.10 4.80 Talkative 4,80 4.40 4.60 4.60 Helpful 5.00(a) 3.70(b) 4.10 4,30 Reliable 4,90 4,lO 3,90 4.60

Dimensions: Positive evaluation 5.43. 4,07 4,67 4,lO Dominance 4.07 4,23 4,47 3.67 Compliance 5.V 3,93 4.1 8 438

*Cell means are significantly different by Duncan multiple range test. Each cell-mean is based on N = 10 subjects.

confirmed: Under the ‘frequent gaze’ condition interactant A is rated as being significantly friendlier, more comradely and more open than under the ‘no gaze’ condition. Accordingly, means on the dimension of positive evaluation differ significantly in the expected direction (see Table 2).

However, hypothesis 2 cannot be confirmed: None of the three dominance ratings (powerful, superior, dominant), and thus also the dominance dimension, differ between the gaze conditions (see Table 1). There is only one univariate comparison for the means between the two gaze conditions which is significant in the expected direction (see Table 2): Under the no prior acquaintance condition A is rated as being more powerful when he looks at his partner than when he does not look at him.

Hypothesis 3 on the dimension of compliance or interaction readiness cannot be confirmed as well. Ratings of A’s talkativeness and reliability do not differ between the gaze conditions; for the ratings of compliance and helpfulness, there is an interaction between the two independent variables which also shows up in the ‘compliance’ dimension (see Table 1). Under the information of prior acquaintance between the interactants A is rated as being more compliant and helpful when he looks at his partner than when he does not look at him (see Table 2). Under the ‘no prior acquaintance’ condition there is a-non-significant-reverse effect.

Hypothesis 4 stated that gaze is more important for impression formations under the prior acquaintance condition than under the no prior acquaintance condition. This hypothesis can be tested by the univariate comparisons between the means of the two gaze conditions under prior versus no prior acquaintance (see Table 2).

Page 8: Acquaintance and visual behaviour between two interactants. Their communicative function for the impression formation of an observer

4 1 6

There are 6 significant differences between the gaze conditions. One of them shows up under the information of no prior acquaintance between the interactants, the other 5 can be found under the acquaintance condition. The differences between the means of the two gaze conditions in 8 out of 10 cases are greater under prior than under no prior acquaintance between the interactants (exceptions: power- ful and superior; see Table 2). A one-tailed sign test yields a significant result (p = 0.05). Thus, hypothesis 4 has been confirmed.

Additionally, Table 1 shows a significant main effect of the acquaintance manipulation for the rating of ‘comradely’. As can be seen from the means in Table 2, under the information of prior acquaintance A is rated as more comradely than under the information of no prior acquaintance.

Andrea A bele

Ratings of interactant B Table 3 shows the results for the ratings of person B.

There is no confirmation of hypothesis 5 : The evaluation of B is not influenced by A’s visual behaviour. Indeed the means differ in the expected direction (positive evaluation of B under A’s frequent gaze: M = 4,60; positive evaluation of B under A’s no gaze: M = 4,13), but there is no statistically significant difference (see Table 3). Also, the other ratings of person B do not differ dependent upon A’s visual behaviour.

In accordance with the rating of interactant A, B is rated as being more comradely under the information of prior acquaintance between the two than under the information of no prior acquaintance (F(1/36) = 6,70, p < 0.05; prior acquaintance: M = 4,90; no prior acquaintance: M = 4.10; see Table 3).

Table 3. Analyses of variance for ratings of person B (N = 40)

Factor

A (acquaintance B (gaze Variable condition) df 1/36 condition) df 1/36 A x B MSEITor

Friendly Comradely Open ‘Positive evaluation’ Powerful Superior Dominant ‘Dominance’ Compliant Talkative Helpful Reliable ‘Compliance’

F < 1 F = 6.70* F = 1,60

F = 2.57 F < 1 F < 1 F < 1 F < 1 F = 2.22 F < 1 F < 1 F < 1 F = 1.20

F = 3.07 F = 3.77 F < 1

F = 2.98 F = 1,86 F < 1 F < 1 F = 1.38 F < 1 F < 1 F < 1 F = 3.62 F < 1

F = 1,47 F = 1.67 F < 1

F < 1 F < 1 F = 1.03 F < 1 F < 1

F < 1

F < 1 F < 1

F < 1

F < 1

49,50 34,40 68.40

26.29 69,80 55.80 88.70 38,32 40.60 99,lO 33,lO 48.90 20,76

‘ p < 0.05.

Page 9: Acquaintance and visual behaviour between two interactants. Their communicative function for the impression formation of an observer

Visual behavwur 417

Assumed influence of the interactants’ non-verbal behaviour upon the impression formation

The assumed influence the non-verbal behaviours of A and B (smiling, gaze, facial expression, gestures) had upon subjects’ impression formations was rated as being the same throughout all the experimental conditions.

EXPERIMENT I1

In order to test the reliability and the generalizability of these results, a second experiment was run with the same stimulus material. There were only some minor changes in the rating scales used.

Method

Subjects

Forty male students who had not taken part in the first experiment served as subjects (age range between 18 and 30 years). They were tested in groups of two or three persons.

Procedure, stimulus material and independent variables

Procedure, stimulus material and independent variables were the same as in Experiment I.

Measures

Because of their low importance in experiment I, the scales ‘reliable-unreliable’ and ‘talkative-not talkative’ were not used again. Also the scale ‘helpful-not helpful’ was discarded. The scale ‘dominant-submissive’ was replaced by ‘strong-weak’. The new scales of Experiment I1 were ‘tense-relaxed’, ‘attentive-unattentive’, ‘active-passive’ and ‘strong-weak’. The remaining scales were taken from Experiment I (‘friendly-unfriendly’, ‘comradely-uncomradely’, ‘open-reserved*, ’compliant-uncompliant’, ‘powerful-powerless’ and ‘superior-inferior*). The dimension of positive evaluation again is formed from the ratings of friendliness, comradeliness and openness. Interaction readiness is formed by ‘compliant’, ‘active’, ‘attentive’ and ‘relaxed’. In order to separate this dimension connotatively from the respective dimension in the first experiment, it is called ‘social participation’. The dominance dimension is operationalized by ‘powerful’, ‘superior’ and ‘strong*. This partial change of dependent variables means a limited comparability of the two experiments. However, it also means an extension upon other aspects of impression formation. All the other measures were taken over from Experiment 1.

Page 10: Acquaintance and visual behaviour between two interactants. Their communicative function for the impression formation of an observer

4 1 8 Andrea Abele

Results

Manipulation checks

In accordance with Experiment I, both gaze of interactant A and the information about the acquaintance between A and B were perceived in the intended way.

Control variables

There were two significant results in respect to aspects of the interactants’ non-verbal behaviours. As in Experiment I ‘smiling’ of A was rated as being more intense under the ‘frequent gaze’ condition (F(1136) = 23,48,p < 0.001). Contrary to Experiment I, there was no such effects for the rating of B’s smiling. The second significant result refers to B’s facial expression: B’s expressiveness was rated as higher under the condition of A’s frequent gaze (F(1/36) = 4 ’ 1 7 , ~ < 0.05). All the other control variables were not different between the experimental conditions.

Correlations between the ratings

The mean correlation between the scales of the ‘positive evaluation’ dimension was r = 0.42 (p < 0.01). The mean correlation between ratings of the dominance dimension was r = 0.67 (p < 0.01). The mean correlation for the dimension of social participation was r = 0.52 (p < 0.01).

Ratings of interactant A

Table 4 shows the results for the ratings of person A.

conditions. In Table 5 , there are the cell means of the ratings in the 4 experimental

Table 4. Analyses of variance for ratings of person A ( N = 40)

Factor

A (acquaintance B (gaze Variable condition) df 1/36 condition) df 1/36 A x B MS Enor

Friendly Comradely

‘Positive evaluation’ Powerful Superior Strong ‘Dominance’ Compliant Relaxed Active Attentive ‘Social participation’

Open

F < 1 F = 1,48 F = 1,23

F < 1 F < 1 F < 1 F = 2,07 F = 1,37 F = 2.06 F = 2,11 F = 1,20 F = 1.45

F = 3,09

F = 10.80t F = 1,48 F = 8,53t

F = 9,66t F = 1,38 F = 2,87 F = 6,17* F = 4,06 F = 2,06 I: = 9,65t F = 4,30* F = 1,44

F = 7,09’

F < 1 F < 1 F < 1

F < I F = 534’ F = 2,11 F = 1,38 F = 3,69 F < 1 F < 1 F = 130 F < 1

F < 1

44,lO 49,lO 88,70

38,52 65,OO 61.40 52,70 45,50 63,OO 86,60 60,SO 62,20

38,86

Page 11: Acquaintance and visual behaviour between two interactants. Their communicative function for the impression formation of an observer

Visual behaviour 419

Table 5. Mean ratings of person A (N = 40)

Acquaintance No acquaintance

Rating Gaze No gaze Gaze No gaze

Friendly Comrade 1 y

Powerful Superior Strong Compliant Relaxed Active Attentive

Open

Dimensions: Positive evaluation Dominance Social participation

5.20* 4,30 5.50. 4.70’ 5,10* 5.30’ 4.80 5,00* 5,60’ 5.20

5 ,OO* 5,03*

5.15

4,Oo 4,lO 3.80 3.20 3,80 3,90 430 3,70 4.20 430

3,97 3,63

4,23

5,20* 5 ,oo 4,70* 3,40 4.10 4.30 4.50 4.50’ 4.60 4.50

4,97* 3.93

4,53

4.10 4,30 3.50 3.90 4.00 3.80 3.60 2.70 4,30 4,20

3.97 3,90

3,70 ~~~~ ~ ~~ ~ ~~

*Cell means are significantly different by Duncan multiple range test. Each cell-mean is based on N = 10 subjects.

The results of Tables 4 and 5 again show a confirmation of hypothesis 1: Person A was evaluated more positively when he looked at his partner than when he did not look at him. This is true for the ratings of ‘friendly’ and ‘open’, as well as for the dimension of positive evaluation, but it is not true for the rating of ‘comradely’ (see Tables 4 and 5).

Hypothesis 2 on dominance ratings is only confirmed for the variable ‘strong’; as the means in Table 5 show, A is rated as being stronger when he frequently looks at his partner. In addition, there is an interaction between the experimental factors on the ‘powerful’ rating which also shows up on the dominance dimension (see Table 4); under the information of prior acquaintance, A is rated as stronger, more powerful, and superior when he frequently looks at B than when he does not look at him. Under the information of no prior acquaintance, there is-except for the rating of ‘strong’-no such effect (see Table 5) .

Hypothesis 3 can be confirmed for two variables (‘relaxed’ and ‘active’) and for the dimension of social participation (see Table 4): Under ‘frequent gaze’ A is rated as more relaxed, more active and thus more socially participatory than under ‘no gaze’ (see Table 5) . For the ratings of attentiveness and compliance, there are no differences between the gaze conditions.

Hypothesis 4 again can be tested by comparing the mean differences between the gaze conditions under the information of prior versus no prior acquaintance (see Table 5). Out of 10 significant differences between means, 7 can be found under the prior acquaintance condition (friendly, open, powerful, superior, strong, relaxed, active), 3 under the no prior acquaintance condition (friendly, open, relaxed). The differences between the means of the two gaze conditions are in 7 out of 10 cases greater under the prior acquaintance than the no prior acquaintance

Page 12: Acquaintance and visual behaviour between two interactants. Their communicative function for the impression formation of an observer

420 Andrea Abele

condition. Under application of a sign test, this difference just fails to be significant

As a main effect the acquaintance manipulation has no influence upon the ratings about person A. The interaction between the acquaintance and the gaze manipulations upon the rating of ‘powerful’ (see Table 4 ) means that contrary to the prior acquaintance condition under the condition of no prior acquaintance, the rating of A’s power is-statistically unreliably-lower under frequent than under no gaze (see Table 5 ) .

(p -= 0.10).

Ratings of interactant B

Table 6 shows the results for the ratings of person B. There are no effects of A’s visual behaviour upon the ratings of person B. The

differences between the means of positive evaluation are in the expected direction, but they are very small (frequent gaze: M = 4,63; no gaze: M = 4,37). Hypothesis 5 on a more positive evaluation of B under frequent gaze of A thus has to be rejected.

The two significant effects of the acquaintance manipulation mean that B was rated as being more attentive (F(1/36) = 4,75,p C 0.05; prior acquaintance: M = 5 , 1 5 ; no acquaintance: M = 4,lO) and compliant (F(1/36) = 4,75,p C 0.05; prior acquaintance: M = 4,70; no acquaintance: M = 3,80) under the acquaintance condition. This effect also shows up with the dimension of social participation.

Assumed influence of the interactants’ non- verbal behaviour upon the impression formation

Also in Experiment 11, there were no differences between these ratings in the different experimental conditions.

Table 6. Analyses of variance for ratings of Person B (N = 40)

Factor

A (acquaintance B (gaze Variable condition) df 1/36 condition)df 1/36 A x B MSElTW

Friendly Comradly

‘Positive evaluation’ Powerful Superior Strong ‘Dominance’ Compliant Relaxed Active Attentive ‘Social participation’

Open

F < 1 F = 3,72 F < 1

F = 1,96 F < 1 F < 1 F < 1 F < 1 F = 4,75’ F = 1,46 F = 1,98 F = 5.78’

F = 6.74’

F < 1 F = 1,90 F = 1,45

F < 1 F < 1 F < 1 F < 1 F < 1 F < 1 F = 1,46 F = 2,59 F = 1.06

F < 1

F = 1,22 F = 2,73 F < 1

F = 1,00 F < 1 F < 1 F < 1 F < 1 F < 1 F < 1 F < 1 F = 2.95

F = 2,Ol

59,90 47,40 75,lO

40,OO 68.40 34,30 32,50 30.38 61,40 103,90 89,OO 68.70

36,34

‘ p c 0.05.

Page 13: Acquaintance and visual behaviour between two interactants. Their communicative function for the impression formation of an observer

Visual behavwur 421

GENERAL DISCUSSION

Awareness of the observed persons’ non-verbal behaviour

There was a fair ability of the present experiments’ subjects to register different gaze patterns of person A who had been facing them. On a 7-point scale and averaged over the 2 experiments, gaze ratings between the ‘frequent gaze’ and the ‘no gaze’ conditions differed about 3 points (frequent gaze: M = 5’40; no gaze: M = 2’30). Thus, the result of good discrimination abilities with extreme visual behaviours is confirmed (see Cook and Smith, 1975; Kendon and Cook, 1969). In both experiments, there is a covanation between the ratings of A’s smiling and the gaze condition. This covariation probably partially reflects real differences between the video recordings, because it is especially difficult to keep the mouth position independent from one’s looking or not looking at another person. But this result might also reflect a learned covariation scheme which means that somebody who does not look at his partner also does not smile. Additionally, a change of the mouth position is only then interpreted as smiling when the eyes are not kept down (Argyle, 1975). However, it should be stated that differences in the impression formations also might have been determined by the perceived amount of smiling by person A.

The other significant results for the ratings of the observed persons’ non-verbal behaviours (smiling of person €3 dependent on gaze condition in Experiment I; smiling and facial expression of person B dependent upon acquaintance condition in Experiment 11) only appear in one of the experiments and are not discussed any further.

Subjects’ awareness of the visual behaviour of the observed persons, however, does not mean that they also knew about the influence it had upon their impression formations. Independent of the respective conditions, the present experiments’ subjects rated the influence of the observed persons’ visual behaviour as relatively small. The correlation between the ratings of the importance of the visual behaviour for impression formation and the ratings of the frequency of glances was r = 0.26, i.e. on the average, subjects only assumed a moderate relation.

Although it is not possible from the present data to draw inferences upon the influence the different visual behaviour patterns had on individual judgements because every subject only saw one recording, the results, nevertheless, show that the visual behaviour’s importance was underestimated in the group mean.

Influence of the visual behaviour Hypothesis 1 on the positive evaluation of person A was confirmed in both experiments. In accordance with the results cited in the literature gaze is a signal of friendliness, openness and-to a certain extent-comradeliness.

The results uppn hypothesis 2, the relation between gaze and dominance ratings, are ambiguous. ‘In Experiment I, there was-with one exception (‘powerful’; see Table 2)-no effect of the gaze manipulation upon the dominance ratings. However, in Experiment If, there was a main effect of the gaze manipulation for the rating of ‘powerful’ (see Table 4). For the combined ratings of the dominance dimension there was also an effect of the gaze manipulation under the condition of

Page 14: Acquaintance and visual behaviour between two interactants. Their communicative function for the impression formation of an observer

422 Andrea Abele

prior acquaintance; person A was rated as more powerful, as stronger and superior, when he looked frequently at his partner (see Table 5 ) . The different effects of the gaze manipulation on the dominance ratings in both experiments partially can be attributed to the different operationalizations of this dimension. Owing to the identical stimulus material used, the differences over and above this only can be interpreted as reflecting a low stability of the relation between gaze and dominance ratings in the context of video recordings. These data thus mean that compared to dominance ratings of a real interactant (see Argyle et al., 1974) gaze is of smaller importance for the dominance ratings of a person observed via a video recording.

Hypothesis 3 on the relation between ratings of social participation and gaze led to partially different results between both experiments. In Experiment I, person A was rated as more compliant and helpful when he looked frequently at his partner and when the subjects had the information of prior acquaintance between the observed persons (see Table 2). The two remaining ratings for this dimension (talkative, reliable) were not influenced by the experimental conditions. In Experiment 11, gaze was significant for the ratings of ‘compliant’ and ‘attentive’. An inspection of the univariate comparisons between cell means (see Tables 2 and 5) , however, shows that consistently in both experiments, gaze was more important for the ratings of social participation when there was the information of prior acquaintance between the interactants. The only exception is the rating of ‘relaxed’ where gaze was important under both acquaintance conditions. The results for hypothesis 3 can be summarized as follows: When there were differences in the social participation ratings, they were always in the expected direction of higher ratings under the condition of frequent gaze. However, these differences did not show up with all scales used, and in general, they were only then statistically significant when there was the information of prior acquaintance between the persons observed.

Hypothesis 5 on the ratings of person B could not be confirmed. In fact, B was rated more positively when A looked at him than when there was no mutual gaze, but the differences were very small. Thus a generalization of the impression formation from person A upon person B could not be observed.

Acquaintance information

Hypothesis 4 stated a greater importance of the gaze manipulation for the impression formation when there is the infomation that the persons observed know each other well. This hypothesis was confirmed in Experiment I, and it just failed to reach the 5 per cent significance level in Experiment 11. Combining the results of both experiments, in 15 out of 20 comparisons between cell means, the differences were greater under the prior acquaintance condition. Under application of the sign test, this difference is significant (p = 0.05). Hypothesis 4 thus in general has been confirmed. The contextual information about the acquaintance between the observed persons intensifies the effect of the visual behaviour. Looking at the ratings’ variances in the different experimental conditions, one can see that in experiment I, they are in 4 out of 10 cases smallest under ‘prior acquaintance, frequent gaze’; in Experiment 11, they are smallest under this condition in all 10 cases. This result might be interpreted as showing that the subjects of the ‘prior acquaintance, frequent gaze’ condition had the lowest difficulties to rate person A,

Page 15: Acquaintance and visual behaviour between two interactants. Their communicative function for the impression formation of an observer

Visual behaviour 423

because both information, the knowledge about the acquaintance and the observed non-verbal behaviour, were fitting. Consistent with this interpretation, the results for the acquaintance manipulation show that the subjects had implicit assumptions about the mutual behaviour of persons who know each other well (in Experiment I, A and B were rated as more comradely; in Experiment 11, B was rated as more attentive and compliant, when there was the information of prior acquaintance).

Contextual informations, visual behaviour and impression formation

Summarizing the results of the present research it can be seen that the visual behaviour of an observed person does have influence on the impression formation of an external observer. Gaze is a signal of friendliness and openness also in this situation. Compared to results with directly interacting partners, ratings of dominance and social participation are influenced to a lower degree by the observed person’s visual behaviour. In general, these ratings only differ then when there is the contextual information about the prior acquaintance between the observed persons. The present results thus confirm the assumption that the communicative function of the visual behaviour is higher when one has the information that the interactants know each other well. This result was obtained even though the operationalization of the acquaintance information was not very strong in the present experiments. A stronger accentuation of the interpersonal relationship might possibly increase the effects.

The extension of the research on the formation of impressions from the visual behaviour of an observed person should be done in three directions which partially refer to the general process of impression formation, partially to impression formation dependent upon different contextual informations: For the general process of impression formation based on the observed person’s visual behaviour specific hypotheses have to be tested on those aspects which are not influenced by the visual behaviour; further, which are the differences between the impression formations of a direct interactant and an external observer. According to research upon the influence of contextual informations, it has to be tested whether-as has been assumed in this article-contextual informations about the interpersonal relationship between the observed persons are of greater significance for the communicative function of the visual behaviour than contextual informations about other aspects of the situation respectively the persons observed.

REFERENCES

Argyle, M. (1975). Bodily Communication, Methuen, London. Argyle, M. and Cook, M. (1976). Gaze and Mutual Gaze, Cambridge University Press,

Cambridge. Argyle, M., Lefebvre, L. and Cook, M. (1974). ‘The meaning of five patterns of gaze’,

European Journal of Social Psychology, 4: 125-136. Cohen, R. (1 969). Systemalische Tendenzen bei Pers~nlichkeitsbeuril~gen . Huber, Bern. Cook, M. and Smith, J. M. C. (1975). ‘The role of gaze in impression formation’, British

Journal of Social and Clinical Psychology, 11 : 19-25. Coutts, L. M. and Schneider, F. W. (1975). ‘Visual behavior in an unfocused interaction as a

function of sex and distance’, Journal of Experimental Social Psychology, 11: 64-77.

Page 16: Acquaintance and visual behaviour between two interactants. Their communicative function for the impression formation of an observer

424 Andrea Abele

Doms, W., Lane, P., and Sharkey. J. (1975). Causal schemata and the perception of implicit communication. Unveroffentlichtes Manuskript, Chicago.

Ellsworth, P. C. and Carlsmith. J. M. (1968). ‘Effects of eye contact and verbal content on affective response to a dyadic interaction’. Journal of Personality and Social Psychology.

Ellsworth. P. C., Friedman, H.. P e r k D. and Hoyt, M. (1978). ‘Some effects of gaze on subjects motivated to seek or to avoid social comparison’, Journal of Experimental Social

Ellsworth, P. C. and Ludwig, L. M. (1972). ‘Visual behavior in social interaction’, The Journal of Communication. 22: 375-403.

Exline, R. V. and Eldridge, C. (1967). ‘Effects of two patterns of a speaker’s visual behavior upon the perception of the authenticity of his message.’ Paper presented at the meeting of the Eastern Psychological Association, Boston.

Gatton, M. J. (1970). ‘Behavioral aspects of interpersonal attraction.’ Unpublished doctoral dissertation, Purdue University.

Gitin, S. (1970). ‘A dimensional analysis of manual expression’, Journal of Personality and Social Psychology, 15: 271-277.

Harper, R., Wiens, A., and Mataraizo, J. (1978). Nonverbal Communication: The State of the Art, Wiley, New York.

Hofstatter, P. (1963). Einfihrung in die Sozialpsychologie, Kroner, Stuttgart. Kendon, A. and Cook, M. (1969). ‘The consistency of gaze patterns in social interaction’,

British Journal of Psychology, 69: 481-494. Kleck, R. E. and Nuessle, W. (1968). ‘Congruence between the indicative and

communicative functions of eye-contact in interpersonal relations’, British Journal of Social and Clinical Psychology, 7 : 241-246.

Kleinke, C. L., Meeker, F. B., and Fong, C. (1974). ‘Effects of gaze, touch, and use of name on evaluation of engaged couples’, Journal of Research in Personality, 7 : 368-373.

Kleinke, C. L., Bustos, S. S., Meeker, F. B., and Staneski, R. A. (1973). ‘Effects of self-attributed and other-attributed gaze in interpersonal evaluations between males and females’, Journal of Experimental Social Psychology, 9: 154-1 63.

LeCompte, W. F. and Rosenfeld, H. M. (1971). ‘Effects of minimal eye contact in the instruction period on impressions of the experimenter’, Journal of Experimental Social

Osgood, C., Suci, G. and Tannenbaum, P. (1957). The Measurement of Meaning, University

Rubin. A. (1970). ‘Measurement of romantic love’, Journal of Personality and Social

Schlosberg, H. (1954). ‘Three dimensions of emotion’, Psychological Review, 61: 81-88. Tedeschi, J. and Lindskold, S. (1976). Social Psychology, Wiley, New York. Thayer, S. and Schiff, W. (1974). ‘Observer judgement of social interaction’, Journal of

10: 15-20.

PS.ychology. 14: 69-87.

Psychology, 7: 21 1-220.

Press, Urbana.

PsyChOlOgy, 16: 265-273.

Personality and Social Psychology, 30, 110-1 14.

Dans deux experiences on a test6 les prkdicticns concernant la formation d’impressions B partir du comportement visuel d’une personne observee. Pour cela des observateurs regardaient I’enregistrement magnCtoscopique d’une conversation entre deux personnes dans laqueIle la personne A regardait son partenaire B ou non, suivant les conditions experimentales. De plus les observateurs Ctaient inform& sur le fait de savoir si les pcrsonnes se connaissaient avant cet entretien ou non.

En concordance avec la prediction, on a trouvC que les dvaluations de la personne A en ce qui concernait sa gentillesse et sa sinc6ritC augmentaient lorque celle-ci regardait son partenaire. Egalement en concordance avec la prediction, l’importance du comportement visuel pour la formation d’impressions Ctait nettement plus grande si I’information sur I’existence de la connaissance entre les deux personnes observCes Ctait prCsente: les estimations de la dominance et de I’activitC de la personne A Ctaient diffbrentes seulement

Page 17: Acquaintance and visual behaviour between two interactants. Their communicative function for the impression formation of an observer

Visual behaviour 425

en relation avec le comportement visuel montrk si I’information etait donnke que les personnes se connaissaint auparavant. Dans le cas oii les partenaires ne se connaissaient pas, cet effet n’existait pas. Ces rtsultats ont montrC que. au deb de l’attribution de gentillesse et de sincerite, la fonction communicative du comportement visuel pouvait augmenter si les informations contextuelles sur la relation interpersonelle entre les deux personnes observees Ctaient donnCes. La formation d’impressions sur la deuxikme personne impliquee dans la conversation n’ttait pas influencke par le comportement visuel de son partenaire de conversation.

ZUSAMMENFASSUNG

In zwei Experimenten werden Vorhersagen zur Eindrucksbildung aufgrund des visuellen Verhaltens einer beobachteten Person getestet.

Beobachter sahen die Videoaufzeichnung eines Gesprachs zwischen zwei Personen, wobei eine Person (A) ihren Partner (B) je nach Experimentalbedingung anschaute oder nicht. Zusatzlich erhielten die Beobachter die Information dariiber,.. ob die Personen sich vor diesem Gesprach bereits gekannt hatten oder nicht. In Ubereinstimmung mit der Vorhersage ist die Bewertung von A hinsichtlich Freundlichkeit und Offenheit hoher, wenn er seinen Gesprachspartner anschaut. Ebenfalls in Ubereinstimmung mit der Vorhersage ist die Bedeutung des Blickverhaltens fiir die Eindrucksbildung wesentlich groBer, wenn die Information guter Bekanntschaft zwischen den beobachteten Personen besteht: Dominanz- und Aktivitatseinschatzungen von A unterscheiden sich nur dann in Abhangigkeit vom gezeigten Blickverhalten, wenn die Information besteht, da6 die beobachteten Personen sich kennen. Diese Befunde zeigen. da6 die kommunikative Funktion des Blickverhaltens uber die Zuschreibung von Freundlichkeit und Offenheit hinaus erhoht werden kann, wenn Kontextinformationen uber die interpersonelle Beziehung zwischen den beobachteten Personen gegeben werden. Die Eindrucksbildung uber die zweite am Gesprach beteiligte Person wird durch das visuelle Verhalten ihres Gesprachspartners nicht beeinfldt.

Author’s address: Dr Andrea Abele, Abteilung Psychologie der Universitat Bielefeld, D-4800 Bielefeld 1, Federal Republic of Germany.