The study involved the use of two different methods of measuring emergent leadership: self-reference and other-reference. The utilization of both self-perceived and other-perceived leadership measures introduced a convergent validation model into the research design. Results indicated that the two measures are significantly related and have similar construct validities. When using both self-report and other-report data, the correlates and patterns of prediction hold up for both types of data. It can be concluded that the predictors of emergent leadership are similar under a variety of leadership measurement strategies. However, the criterion measures, self-perceived leadership and other-perceived leadership, may represent variations of emergent leadership and both may be needed to more fully understand the leadership emergence phenomenon.

Self-report and rater-data approaches have long been recognized as two basic methods of acquiring data in the behavioral sciences. Spitzberg (1986a, 1986b) calls these "self-reference" and "other-reference" instruments. Both methods rely on self-disclosure, and both are subject to distortions (Cronbach, 1984). Particularly, distortions of reality can be peculiar to both kinds of data (Cattell, 1965). Both are subjective types of data, but the major advantage of self-report data is that it is objective in the sense that all respondents react to the same stimuli.

With rater data, rating scales can help achieve this kind of objectivity. However, rater data can be doubly subjective in that they can be distorted by not only how the ratee reveals himself/herself, but by the observer who obtains, records, and evaluates the information. The reliability of rater judgment can be brought to acceptable levels by careful training of observers (Cronbach, 1984).

There is considerable support for utilizing both methods, rater/observer and self-report, in gathering information about a certain trait or behavior pattern (Campbell &;Fiske, 1959; Cronbach et al., 1972; Dawes, Faust, &;Meehl, 1989).

The present study involved the use of two different methods of measuring emergent leadership: self-reference (self-report) and other-reference (rater data) (Spitzberg, 1986a, 1986b). The utilization of both self-report, or self-perceived leadership (SPL) and the observer or other-perceived leadership (OPL) measure introduces a convergent validation model into the research design.

This approach is somewhat similar to the Campbell and Fiske (1959) multitrait-multimethod matrix and the Cronbach, Glesser, Nanda, and Rajaratnam (1972) generalizability convergent validity models. These reliability models are based on the idea that there are different sources of variation for the measurement of different traits and that when these different sources converge (or discriminate) in the measurement of these traits, then the reliability and validity of measurement is enhanced.

Historically, emergent leadership has been investigated by operationalizing the phenomenon as the extent to which an individual is perceived by others (other-reference) to have emerged as the group's leader. This study was interested in: (1) determining whether one's self-perception (self-reference) of emergent leadership would correlate with other-perceived emergent leadership; and (2) discovering the extent to which the same variables correlate with and predict both self-perceived and other-perceived emergent leadership. From the standpoint of convergent validation, if the same independent variables predict SPL that predict OPL, this indicates that the two measures of leadership emergence have similar construct validities. Such an approach would also allow one to determine how these two leadership measures do differ, if they do in fact have different construct validities (discriminate validities).

Method

Predictor Variables
Much of the research on leadership emergence has been concerned with the actual "talking behavior" of elected or achieved leaders (Bales, 1951; Larson, 1971; Morris &;Hackman, 1969; Reynolds, 1984). One meta-analytic review of leadership emergence literature has concluded that "generally, the tendency for the individual with the highest level of verbal participation to be chosen as the leader was significant and of strong magnitude" (Mullen, Salas &;Driskell, 1989, p. 545).

While the actual talking behavior of perceived leaders has received much attention, not much attention has been paid to the self-perceived talking behaviors of emergent leaders. Such intrapersonal perceptions may also be related to the phenomenon of emergent leadership.

Recent research in the area of self-perceived interpersonal talk, using the Talk Assessment Survey (Wanberg, 1987; Wanberg &;Goldberg, 1990,1991), provided a variable base for investigating the relationship between talk and the measures of emergent leadership for this study. Wanberg and Goldberg found that when individuals were asked to describe the verbal expressions they used when talking with others, they invariably formulated their responses in what were identified as categories of talk statements. Wanberg and Goldberg labelled these categories as "comemes" or self-perceived intrapersonal structures which are used in talking with others. Whereas talk expressions are considered to be specific statements used in talk communication, the comeme represents the structuring of these expressions.

The Talk Assessment Survey (TAS) is comprised of 99 comemes. Eleven primary and three second order common factors have been reliably identified among these comemes across a variety of samples to demonstrate factor invariance (Wanberg &;Goldberg, 1990). Scales have been constructed to estimate these latent structures, and these scales have demonstrated to have very good internal consistency reliabilities and construct independence. (Wanberg &;Goldberg, 1990). Table 1 provides a summary of the 14 TAS scales, sample items used in each scale, and internal consistency reliabilities (Cronbach's alpha) for two different samples.

Over the years, the concepts of power and leadership have been closely linked (Bass, 1981). It is widely recognized that certain bases of power are important in bringing about effective leadership (French &;Raven, 1968). According to one review, those considered to have only a formal designation of leadership (legitimate power base) were scored lower in measures of performance and overall group satisfaction than those operating from more informal bases of power such as "referent" and "expert" (Hersey &;Blanchard, 1977).

This study was interested in determining if certain orientations toward power might be used to predict self-report and other-report emergent leadership. Cavanaugh (1979) defined a "power orientation" as "an individual's attitudes and beliefs about power" (p. 11). According to Goldberg et al., (1983) "the self-talk an individual engages in when focusing on the idea of power comprises that person's power orientation" (p. 90).

Cavanaugh's (1979) work led to the formation of an instrument called the Power Orientation Scale to measure a person's view or orientation to the concept of power (Goldberg, Cavanaugh, &;Larson, 1983). This approach to power recognizes the importance of intrapersonal communication in the context of interpersonal, group, and organizational life.

The six power orientations identified by Cavanaugh (1979) are: power as good, power as resource dependency, power as instinctive drive, power as political, power as charisma, and power as control &;autonomy.

In summary, the 11 primary dimensions of talk, the 3 second order dimensions of talk, and the 6 orientations toward power provided the variable base with which to investigate the correlations and patterns of predictors for both self-perceived leadership and other-perceived leadership in this study.

Leadership Emergence Criterion Measures
The research on emergent leadership is generally concerned with determining perceptions of leadership by members of an initially leaderless group. One's perceptions have been measured through a sociometric method where group members rank each group member in relation to others (Drecksel, 1985; Rees &;Segal, 1984; Reynolds, 1984), by a voting procedure in which the person acquiring the most votes is deemed the emergent leader (Baird, 1977; Wentworth &;Anderson, 1984), or by developing a set of Likert-type scales or semantic differential items in a questionnaire (Dobbins, Long, Dedrick &;Clemons, 1990; Kent &;Moss, 1990; Lord &;Alliger, 1985).

While each system serves a variety of purposes, Likert type scales offer a couple of advantages. First, such scales allow one to more adequately quantify the extent of leadership perceived to exist in groups. Second, such a determination is amenable to comparison and higher levels of statistical analysis.

The instrument used to measure both self-report and other-report emergent leadership is an adaptation of a Likert-type instrument used by Lord (1977), Lord and Alliger (1985), and Lord, Phillips, and Rush (1980) in their studies of leadership perceptions. The instrument was developed by Lord (1976) to study aspects of group performance as a function of leadership, behavior, and task structure. It has been found to be strongly related to objective behavioral measures and other leadership perception measures and has been utilized in subsequent research to determine general leadership perceptions (Ellis, Adamson, Deszca, &;Cawsey, 1988).

The primary purpose of this instrument is the assessment of emergent leadership as measured by perceptions of leadership by other group members. Items asked subjects to indicate: the amount of ratee's contribution to task performance, the amount of leadership the ratee exhibited, how willing the rater would be to choose the ratee as the formal leader for a successive experimental session, the extent to which the ratee exerted control over group activities, and the extent to which the ratee exerted influence over other group members. For the purposes of the present study, these five items are referred to as the "Leadership Emergence Measures" or LEM.

Two types of measures were taken using the Leadership Emergence Measures (LEM) instrument: self-perceived leadership (SPL) and other-perceived leadership (OPL). The SPL represents a self-report measure and involved the subject rating himself/herself on the five LEM items. A single SPL score was assigned to each subject by summing across all five of the SPL items. The OPL measure involved each individual in the group being rated by the other three group members across the five LEM items. Each subject was assigned a single OPL score by summing across all ratings for the five OPL items. Thus, the SPL measure represents self-report data and OPL essentially represents rater data.

Subjects

Subjects for this study were undergraduate and graduate students enrolled in journalism, mass communication, and speech communication classes. The sample included 90 male and 90 female subjects.

Procedure

Participants in the study first completed the instruments which measured their power orientations and self-perceived use of talk. Both the Power Orientation Scale or the Talk Assessment Survey were used as measures to ascertain general orientations toward power and talk and were not referenced as pertaining to specific aspects of leadership. Therefore, the study was essentially correlating general power and talk measures against a specific leadership behavior assessment.

Approximately two weeks after completing the POS and TAS instruments, the subjects participated in four-person, task-oriented groups comprised of two males and two females.

Since group output may be confounded by gender differences in the composition of groups (Aries, 1976; Eskilson, 1974; Eskilson &;Wiley, 1976), a proportional mixed-sex composition was used to control for possible confounding effects of gender in the make up of the groups.

Groups were not assigned leaders and were given one of three tasks. The tasks involved having groups either (1) produce a letter which informs passengers about a threat on an airliner and what they can do about it, or (2) discuss the pros and cons of warning the public about such threats, or (3) come up with a plan to solve the problem of when such warnings should be issued and who should be responsible for making warnings of such threats public. The tasks were based on an earlier task type category system which included production tasks, discussion tasks and problem-solving tasks (Hackman, 1968; Morris, 1966). Task types were controlled as separate factors in the analyses, however results showed that no differences with regard to task were present in this study.

At the end of a 25 minute task session, the assessment of leadership perceptions was obtained by five items (Lord, 1976) in a post-experimental questionnaire which asked respondents to score their own leadership behavior and to rate others in their groups. The post-experimental questionnaire was also used to acquire demographic information about the subjects.

Analyses

Internal consistency reliability using Cronbach's alpha (Cronbach, 1951) were calculated for both the self-perceived and other-perceived leadership measures.

Simple linear relationships between the emergent leadership measures (SPL and OPL) and the talk and power scales were studied using Pearson's product moment correlations. Multiple linear relationships between the leadership measures and the power and talk scales were studied using a stepwise multiple regression analysis.

Because the second order scales of the Talk Assessment Survey contain items which are part of the primary scales, there was a concern that the overlap would artificially increase the magnitude of the correlations and confound covariances between the primary talk and second order talk scales. The high correlation among these predictor variables would thus reduce the probability that one or both of the predictors would enter into the equation of predicting the criterion. Therefore, it was necessary to separate the primary scales and second order scales into different regression analyses.

Results

Correlations among the five SPL items were very high resulting in an internal consistency reliability (ICR) of .92. The ICR for the OPL summed item scores across raters was .95. This indicates that the items measuring other-perceived leadership were highly intercorrelated and internally consistent.

The correlation between the SPL and OPL measures was .54, indicating the two measures share 29 percent common variance.

At the simple linear (correlational) analysis level, both self-perceived and other-perceived leadership were significantly correlated with the primary talk factors SUPPORT, RETICENT (negative), and INFORM. (See Table 2) The talk factor FEEDBACK was significantly positively related to both SPL and OPL, while HELPSEEK was positively related to self-perceptions of leadership. The power orientation Charisma had a highly significant positive correlation with SPL. Significant positive correlations were also found between the second order scales INVOLVEMENT and INFORMATIVE and both self-perceived leadership and other-perceived leadership.

When combining only the second order talk scales and the power scales, INFORMATIVE, Power as Charisma and Power as Resource Dependency were the best predictors of self-perceived leadership. (Table 5) The talk factor INFORMATIVE was the only predictor of other-perceived leadership. (Table 6)

The best linear prediction model for self-perceived leadership includes a linear combination of low scores on RETICENT, high scores on SUPPORT, high scores on Power as Charisma, and low scores on Power as Resource Dependency. (Table 3). The regression analysis using second order talk scales indicated that INFORMATIVE also contributed significant variance. (Table 5).

These findings indicate that the perception that one is apt to emerge as leader in an initially leaderless group will most likely be predicted by the perception about self that one is not hesitant to speak, is supportive and concerned for the welfare of others, holds attitudes about power that says people possess something different from others, and does not tend to view power as something where one should try to control resources to make others dependent on them. These findings support earlier research on emergent leadership which indicates that active involvement and talkativeness are two of the more salient predictors of emergent leadership (Bales, 1951; Larson, 1971; Reynolds, 1984).

Since emergent leadership involves a free exchange of information, ideas, and recommendations, those viewing power as Resource Dependency are less likely to perceive themselves as leaders. People with views of power as Resource Dependency try to create dependencies in which others must come to them for certain resources such as information. This may work to gain compliance in a designated or formal position of leadership. However, creating resource dependencies by withholding information is not likely to help one achieve a leadership position in the process of emergent leadership. Therefore, those who view power as Resource Dependency may not look on themselves favorably as emergent leaders.

The prediction models for other-perceived leadership (OPL) were somewhat similar. The primary talk factors SUPPORT and RETICENT and the second order talk scale INFORMATIVE were the only factors to contribute significant variance to the prediction of OPL. (Tables 4 and 6).

Someone wishing to assess the best linear combination of power and talk factors to predict other-perceived leadership would conclude that those who can be oriented towards others, as evidenced by high scores on SUPPORT, are most likely to be perceived by others as better leaders. Those who express support, kindness, and sympathy tend to be more "other" oriented than those using other forms of talk. These talk patterns, along with a high overall degree of communicative interaction, seem to be most highly related to perceptions by others as to who will emerge as leaders in initially leaderless groups. These findings also support previous findings of research in the area of emergent leadership which indicates that emergent leaders tend to express more positive, approving behaviors (Baird, 1977), tend to talk more than others (Bass, 1949; Bostrom, 1970; Slater, 1955), and are consistently involved in conversations (Fisher, 1980; Reynolds, 1984).

The underlying assumption of this study is that when a group of individuals are brought together and given an assigned task with no designation of a specific leader, a leader will emerge in the process of the group performing the assigned task. Although this research did not address whether a leader did in fact emerge in such a group, the results do indicate that individuals reliably rate themselves across criteria which define leadership and they reliably rate others across these same criteria. There was a concept of leadership operating in these groups. Not only did they discriminate among themselves, both as observers of self and others (there was good variability in the SPL and OPL scores), they demonstrated reliability as to the rating of self and others on these measures. The final ranking of scores reflects this agreement as evidenced by high intercorrelations among variables and among raters, both self and others. Some have high scores on SPL/OPL, others have low scores, and there was generally agreement among subjects as to who were high, who were medium, and who were low. In essence, the concept of emergent leadership operates and is a reliable concept.

One limitation of the study has to do with the issue of previous interaction. To the greatest extent possible the groups were zero-history groups. However, certain groups were less likely than others to fit the criteria of the zero-history, leaderless group discussion. It may be that subjects in smaller classes had more previous interaction with those eventually chosen to take part in their groups, whereas the larger classes were much less likely to have had a history of previous interaction. At least one study has shown that there may be differences in leadership impressions between zero-history groups and those with some history of interaction (Pavitt &;Sackaroff, 1990).

Also, the study was limited in the amount of time groups were given to interact. Twenty-five minutes may only have been enough time for group members to form early impressions of emergent leadership. The phenomenon of emergent leadership is probably a process that evolves and changes over the course of more than one meeting and such developmental effects were unaccounted for in this study.

As stated, the correlation between scores subjects gave themselves (SPL) and the sum of scores given by others (OPL) was .54, indicating that the two measures share 29 percent common variance. Although this represents a statistically significant and moderate relationship, 71 percent of the variance of these two measures is separate from and cannot be explained by their relationship. This suggests that SPL and OPL may be measuring different aspects of emergent leadership, and both are needed in an examination of the emergent leadership phenomenon.

Similarly, some of the correlations between the independent variables in the study and the leadership measures were somewhat low. This could be explained by the problem of trying to correlate general measures of power and talk, which were administered two weeks prior to the group interaction, with a very specific leadership behavior assessment. It is likely that stronger relationships might have been discovered if the power and talk instruments would have asked for responses in the context of leadership positions.

Even with the relatively low correlations, the fact that most of the correlates of the power and talk scales with SPL, and most of the predictors of SPL, were also the same correlates and best predictors of other-perceived leadership (OPL) in this study does tend to at least partially support the convergent validity approach to testing a construct. That is, among a set of variables (power and talk), the same variables had significant correlations with a leadership measure derived from two different methods (SPL and OPL). As well, similar predictors emerged in the regression study for a leadership measure derived by two different methods. These findings would suggest that SPL and OPL have common sources of variance and that they do measure something similar about leadership emergence. These findings indicate that the two measures (SPL and OPL) have similar construct validities. In essence, there was good agreement between the self-report and other-report data in this study.

Andrews, P. H. (1984). Performance-self-esteem and perceptions of leadership emergence: A comparative study of men and women. The Western Journal of Speech Communication, 48, 1-13.

Aries, E. (1976). Interaction patterns and themes of male, female, and mixed groups. Small Group Behavior, 7, 7-18.

Baird, J. E., Jr. (1977). Some nonverbal elements of leadership emergence. Southern Speech Communication Journal, 42, 352-361.

Bales, R. F. (1951). Interaction process analysis: A method for the study of small groups. Cambridge: Addison-Wesley Press, Inc.

Bass, B. M. (1949). An analysis of the leaderless group discussion. Journal of Applied Psychology, 33, 527-533.

Bass, B. M. (1981). Stogdill's handbook of leadership. New York: The Free Press.

Bormann, E. G. (1975). Discussion and group methods: Theory and practice, 2d ed. New York: Harper &;Row.

Bostrom, R. N. (1970). Patterns of communicative interaction in small groups. Speech Monographs, 37, 257-263.

Campbell, D. T., &;Fiske, D. W. (1959). Convergent and discriminant validation by the multitrait-multimethod matrix. Psychological Bulletin, 56, 81-105.

Cattell, R. B. (1965). The scientific analysis of personality. Baltimore: Penguin Books.

Cavanaugh, M. S. (1979). A formulative investigation of power orientations and preliminary validation of relationships between power orientation and communicative behavior. Unpublished doctoral dissertation, University of Denver.

Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16, 297-334.

Cronbach, L. J. (1984). Essentials of psychological testing. New York: Harper &;Row.

Cronbach, L. J., Glesser, G. C., Nanda, H., &;Rajaratnam, N. (1972). The dependability of behavioral measurement: Theory of generalizability for scores and profiles. New York: Wiley.

Dawes, R. M., Faust, D., &;Meehl, P. E. (1989). Clinical versus actuarial judgment. Science, 243, 1668-1674.

Dobbins, G. H., Long, W. S., Dedrick, E. J., &;Clemons, T. C. (1990). The role of self-monitoring and gender on leader emergence: A laboratory and field study. Journal of Management, 16, 609-618.

Drecksel, G. L. (1985, May). Interaction characteristics of emergent leadership. Paper presented at the Annual Meeting of the International Communication Association, Honolulu, HI.

Ellis, R. J., Adamson, R. S., Deszca, G., &;Cawsey, T. F. (1988). Self-monitoring and leadership emergence, Small Group Behavior, 19, 312-324.

Eskilson, A. (1974). Sex composition and leadership in small groups. Dissertation Abstracts International, 35, 5694.

Eskilson, A., &;Wiley, M. G. (1976). Sex composition and leadership in small groups. Sociometry, 39, 183-194.

Fisher, B. A. (1980). Small group decision making. 2d ed. New York: McGraw-Hill.

French, J. R., &;Raven, B. (1968). The bases of social power. In D. Cartwright &;A. Zander (Eds.), Group dynamics: Research and theory. New York: Harper &;Row.

Geier, J. G. (1967). A trait approach to the study of leadership in small groups. Journal of Communication, 17, 316-323.

Goldberg, A. A., Cavanaugh, M. S., &;Larson, C. E. (1983). The meaning of power. Journal of Applied Communication Research, 11, 89-108.

Hackman, J. R. (1968). Effects of task characteristics on group products. Journal of Experimental Social Psychology, 4, 162-187.

Hersey, P., &;Blanchard, K. (1977). Management of organizational behavior: Utilizing human resources. Englewood Cliffs, N.J.: Prentice-Hall.

Kent, R. L., &;Moss, S. E. (1990). Self-Monitoring as a predictor of leader emergence. Psychological Reports, 66, 875-881.

Larson, C. U. (1971). The verbal response of groups to the absence or presence of leadership. Speech Monographs, 38, 177-181.

Lord, R. G. (1976). Group performance as a function of leadership and behavior and task structure: Toward an explanatory theory. Organizational Behavior and Human Performance, 17, 76-96.

Lord, R. G. (1977). Functional leadership behavior: Measurement and relation to social power and leadership perceptions. Administrative Science Quarterly, 22, 114- 133.

Lord, R. G., &;Alliger, G. M. (1985). A comparison of four information processing models of leadership and social perceptions. Human Relations, 38, 47-65.

Lord, R. G., Phillips, J. S., &;Rush, M. C. (1980). Effects of sex and personality on perceptions of emergent leadership, influence, and social power. Journal of Applied Psychology, 65, 176-182.

Morris, C. G. (1966). Task effects of group interaction. Journal of Personality and Social Psychology, 4, 545-554.

Morris, C. G., &;Hackman, J. R. (1969). Behavioral correlates of perceived leadership. Journal of Personality and Social Psychology, 13, 350-361.

Mullen, B., Salas, E., &;Driskell, J. E. (1989). Salience, motivation, and artifact as contributors to the relation between participation rate and leadership. Journal of Experimental Social Psychology, 25, 545-559.

Pavitt, C., &;Sackaroff, P. (1990). Implicit theories of leadership and judgments of leadership among group members. Small Group Research, 21, 374-392.

Rees, C. R., &;Segal, M. W. (1984). Role differentiation in groups: The relationship between instrumental and expressive leadership. Small Group Behavior, 15, 109-123.

Reynolds, P. D. (1984) Leaders never quit: Talking, silence, and influence in interpersonal groups. Small Group Behavior, 15, 404-413.

Slater, P. E. (1955). Role differentiation in small groups. American Sociology Review, 20, 300-310.

Spitzberg, B. H. (1986a). A critical review of communicative competence measures I: Self-reference instruments. Paper presented at the Speech Communication Association Convention, Chicago, IL.

Spitzberg, B. H. (1986b). A critical review of communicative competence measures II: Other-referenced instruments. Paper presented at the Speech Communication Association Convention, Chicago, IL.

Wanberg, K. W. (1987). Dimensions of talk: A multiple- construct view of human communication. Unpublished doctoral dissertation, University of Denver.

Wanberg, K. W., &;Goldberg, A. A. (1990). Self-perceived dimensions of interpersonal talk. Paper presented at the Speech Communication Association Annual Convention, Chicago, IL, November, 1990.

Wanberg, K. W., &;Goldberg, A. A. (1991). Differential assessment of self-perceived interpersonal talk: The scales of the Talk Assessment Survey. Unpublished manuscript.

Wentworth, D. K., &;Anderson, L. R. (1984). Emergent leadership as a function of sex and task type. Sex Roles, 5/6, 513-524