An Exploratory Investigation of

Characteristics of Compulsive Talkers


Robert N. Bostrom

Nancy Grant Harrington


 


Robert N. Bostrom (Ph. D., 1961, University of Iowa) is a professor at the University of Kentucky.

Nancy Grant Harrington, (Ph. D., 1992, University of Kentucky) is an assistant professor at the University of Kentucky. The authors wish to thank Ruth Anne Clark and two anonymous reviewers for their help in this study.
 
 

The "talkative" person may well be as much of a communication problem as the reticent one. In this study, 28 undergraduates who were identified by acquaintances as talking too much were compared to 224 "normals" on a series of variables thought to be associated with compulsive talking. Talkers differed from "normals" in their self-reports of verbal activity (PVB scores). Talkers also reported that they were more argumentative, less apprehensive, and had more positive attitudes about communication than normals. No differences in self esteem or locus of control were found between these groups. Keywords: compulsive talking, communication attitudes, apprehension, argumentativeness, locus of control
 


An Exploratory Investigation of the Characteristics of Compulsive Talkers


 


Few persons are more irritating than those who talk too much. While many individuals are "talkative," most are within the bounds of social acceptability. However, there is a degree of talkativeness that goes far beyond the bounds of social acceptability and that many persons find to be highly unpleasant. Some use the term "compulsive"(1) when describing such individuals. When a compulsive talker is involved in an interpersonal interaction, participants react with irritation, impatience, and disgust. Compulsive talkers seem to overvalue the quality of their contribution and signal (through refusal to listen) that others' ideas are unimportant. Worse, they often seem unaware of their problem. Regardless of cause, there seems to be universal agreement that talkers are a problem, both to themselves and to others. Accordingly, information about talkers would be useful in helping talkers control this behavior.

In an earlier investigation (Bostrom, Grant, Davis, & Einerson, 1990), two factors emerged that seemed to serve as reasonable criteria for the designation of "talkers:" (1) in an interpersonal interaction, they literally talk "nonstop," and only give way when another interactant begins talking, and (2) their talking is perceived by others as a problem. The behavior of talkers is a serious problem, not only affecting the talkers but those who work and live with them, as evidenced in two examples of "talkers" taken from that investigation:

Case One: This "talker" holds a mid-level supervisory position in a large Midwestern corporation. While he has worked at this corporation for over eighteen years, he has remained in his present position for the last 12 of those 18 years. An intelligent, capable man, he has been passed over for promotion time and time again because of the widespread knowledge of his compulsive talking. He is known to "trap" people in his office and on the telephone, telling them about his wife, children, golf game, etc., ad nauseam. In fact, it has become common practice for co-workers, when they have to talk to him, to preface the conversation with a statement like, "Now I only have a minute I can give you," or "I have to be in a meeting in five minutes, so we have to make this fast" (Bostrom, et al., 1990).
 

Case Two: This "talker" is a middle-aged woman who is the last surviving member of a prominent, wealthy, local family. Single, and with few social outlets, she makes her church the focus of her life. She attends every dinner or special committee meeting, arrives early, and leaves late. Typically she is the first to arrive at a function, and fastens her attention on whomever has the ill luck to be the second arrival. She begins with any topic and switches with skill. If not interrupted, she will talk continually for as long as fifteen or twenty minutes. Individuals cope with her by avoidance, and by attempting to foist her off on other members. Her true position in the church could only be described as that of a social outcast (Bostrom, et al, 1990).

Both of these "talkers" share a single trait, but seem to be completely different persons --- in training, in gender, in outlook, and in socio-economic status. They share a common failing, however, and inflict it on their coworkers and friends.

The persistence of the phenomenon is puzzling. Naive theories of behavior would predict that activities that are not reinforced should diminish in frequency and eventually disappear. This reinforcement effect has been demonstrated a number of times and in a variety of communicative settings (Bostrom, 1962; Verplanck, 1955). McCroskey and Richmond's (1995) contention that talking is positively valenced in our society would certainly explain ordinary talkativeness, and even "excessive" talking, but not the compulsiveness of many talkers. In the instances of compulsive talking that we discovered, the behavior seemed to be non-reinforced, i. e., disliked by friends and co-workers.

Unfortunately the literature in communication tells us little about the compulsive talker. Communication researchers have traditionally concerned themselves with reticence, rather than its opposite. Most have assumed that since communication is obviously of great value, anything that interferes with it is harmful. Researchers such as Spitzberg and Cupach (1984) typically have assumed a universal "motivation" to communicate --- a general tendency to approach or avoid communicative situations. Avoidance, of course, has attracted the most attention from researchers, and communication apprehension, shyness, and reticence, have been assumed to constitute a significant social and educational problem (Daly & McCroskey, 1984). Whether these feelings of avoidance have a corresponding psychological opposite is still an open question.

McCroskey and Richmond (1995) suggest that our culture rewards talkativeness. For example, Daly, McCroskey, and Richmond (1976) found that individual's level of talk was highly correlated with judgments of leadership and influence in a small group. Arntson, Mortenson, and Lustig (1980) found that "strong talkers" were perceived as being more influential. This might indicate that compulsives were those who anxiously seek leadership and influence. On the other hand, it is possible that the talker may be only a person with bad social habits. Turn-taking in conversation has been characterized as an important cultural skill (Wiemann & Knapp, 1975), and those who "talk too much" may simply be those whose knowledge of their own communicative culture is deficient. In other words, talkers may be normal in other respects, but just ill-mannered, or with a poor attitude about other persons.

Some research in "approaching" communication has occurred in other contexts. General positive feelings toward communication are implicit in Mortensen's "predisposition" to verbal behavior (Mortensen, Arntson, & Lustig, 1977) and "willingness" to communicate (McCroskey & Richmond, 1987). These predispositions are negatively correlated with apprehension, but little else is known of their etiology or other relationships. Approach and avoidance may be true bipolar opposites, but whether or not this actually is the case is not known. The relationship between approach and avoidance behavior and level of communicative skill could be conceptualized as an inverted "U" curve, with skill along the vertical axis and "amount" of communication along the horizontal. An increase in talk might represent an increase in skill, but only to a certain point. After that point, continued talk may reflect a lack of skill.

Another way of examining the relationship between approach and avoidance of communicative behavior can be conceptualized as an attitude/behavior relationship. Approach could be caused by an underlying positive attitude toward communication (Bostrom, 1962, 1970, 1981). In addition, some individuals might see communication as a recreational activity. On the other hand, it is possible that compulsive talkers might well be very different kinds of people, with extremely different motivations, skills, and attitudes. Lastly, it is possible that talkers are the product of a combination of many of these different factors.

McCroskey and Richmond (1995) studied compulsive talkers by examining responses to a "talkaholic" scale. They discovered a modest relationship between a "talkaholic" scale score and willingness to communicate and assertiveness. They also discovered negative relationships between this scale and shyness, introversion, and communication apprehension. McCroskey and Richmond interviewed those with high scores on their scale and, unlike the adults in our first study (Bostrom, et al., 1990), reported that "compulsive communication was not a problem for them" (McCroskey & Richmond, 1995, p. 49).

Clearly there are many unanswered questions about compulsive talkers. Talkers differ from
"normal" persons in their verbal behavior, and it would seem logical for them to differ in other characteristics. McCroskey and Richmond's data showed differences in shyness, introversion, and apprehension, but left many unanswered questions. First, identifying talkers on the basis of scores on a questionnaire is problematic, especially if the criteria of nonstop talking, coupled with social problems, are applied. Also, it is not clear whether talkers may lack inhibiting negative affect (apprehension) or possess an abundance of positive affect (pleasurable attitudes). Predispositions toward talking are more complex than might be supposed, with the predisposition consisting of factors of time taken (frequency), dominance, and inhibition (Bostrom, Prather, & Harrington, 1998). In addition, other aspects of communication, such as argumentativeness and attitudes about communication might be associated with compulsive talking. The purpose of this study was to determine whether individuals who are indentified as talking too much differ from "normals" on variables that there is some reason to believe may be assiciated with compulsive talkers: predisposition toward verbal behavior, communication apprehension, argumenatativeness, self-esteem, and locus of control.
 


Method


 


Participants

Since compulsive talking is viewed so negatively by many, we asked a large number of undergraduates (786) to nominate someone of their acquaintance who was a compulsive talker. A questionnaire was distributed to classes which asked "Do you know someone who talks TOO MUCH? Would you be willing to tell us who they are?" This was followed by a series of blanks with places for names, addresses, and telephone numbers. Respondents were assured of anonymity and that the process was voluntary.

A list of 54 names and addresses was generated from these responses. Most were undergraduates, but some were members of the larger community. No one was nominated more than once. These persons were called and were asked to volunteer for a "communication study." No mention was made of the basis for their selection. Thirty-two were eventually contacted and of these 32, 28 (including 23 undergraduates) came to the research site and responded to a questionnaire booklet. These 28 individuals were designated as the "talkers" sample(2). In the group were 17 females and 11 males, with an average age of 22.3.

Data from another study (Bostrom, Prather, & Harrington, 1998) were used as a "normal" group for comparison. This group consisted of 224 undergraduates enrolled in upper division communication classes. In the group were 118 females and 106 males. Their average age was 20.6.
 

Scales

To measure the amount of talking, the Predisposition toward Verbal Behavior scale (Mortensen, Arntson, & Lustig, 1977) was selected.(3) This scale (PVB) is a self-report which measures the respondents' verbal activity. There is strong evidence of correspondence among such self-reports of behavioral tendencies, levels of verbalization, and subsequent impressions of verbal behavior (Mortensen et al., 1977). The factor structure reported by Mortensen and his coworkers did not appear (Bostrom, et al., 1998). Rather, factors that indicate conversational dominance, frequency, and inhibition were found.(4) These factors are presented in Table 1.

------------------------
Table 1. about here
-------------------------

Communication Apprehension is historically the most common indication of affective responses to communicative situations. Apprehension (McCroskey, 1970, 1977, 1978) has been shown to be correlated with introversion and verbal reticence. Daly's (1978) comparison of communicatively oriented affect measures clearly indicated that the McCroskey measure is the best indicator of these traits. We used the short form of the Personal Report of Communication Apprehension (PRCA) (McCroskey, 1978).

Argumentativeness is another characteristic that might well be associated with talkers. This characteristic differs from verbal aggressiveness (Infante & Wigley, 1986), which is more a measure of an individual's tendency to attack or humiliate another person. Argumentativeness was chosen over verbal aggressiveness because verbal aggressiveness need not be correlated with loquacity, while argumentativeness should exhibit this feature.

Self-esteem was measured with Rosenberg's (1965) scale. Research using this measure indicates that persons who have high self-esteem are inclined to feel comfortable with others, are not afraid of others' reactions, and are able to defend themselves against negative comments of others (Hamachek, 1982). Persons who have low self-esteem feel threatened by others, are sensitive to possible negative reactions, and have difficulty defending themselves against others' negative comments (Hamachek, 1982). Thus, a person with high self-esteem may be inclined to talk more than a person with low self-esteem.

Locus of Control was assessed with Rotter's scale (Rotter, 1966). This scale measures the extent to which individuals perceive themselves as having control over their environments. Persons with an internal locus of control consider themselves to have power over the events in their lives and to be responsible for any rewards or failures they experience. Those with an external locus of control believe that they are at the mercy of forces outside themselves, that any effort on their part to control events is pointless. Rotter's scale has proved to be related to a number of social characteristics in several studies (Ayabe, 1979; Lefcourt, 1982; Ollendick, 1979; Surlin, 1976 ), many of which relate to talking. Externals might well feel that since others control their lives, it is pointless to talk.

Since evaluations of communication activities in an "attitude" framework have been shown to be predictive of communicative activity (Bostrom, 1970, 1990), a "Communication Attitude Index" was constructed to measure the affective orientation toward talking. This index consisted of five statements designed to determine the extent to which individuals "enjoy" various aspects of talking.(5)
 

Procedure

Participants came to a designated room in groups of two or three, and were asked to respond in a test booklet. It was clear that these persons were indeed compulsive talkers.(6) The test booklets were introduced as questionnaires designed to assist the department in curricular matters, and various scales were ordered randomly within each booklet. The PVB scales, however, were consistently grouped together. Most respondents finished in approximately 35 minutes, and none took longer than an hour.

Analysis

All of the scales were entered into a 2 x 8 multiple analysis of variance. This procedures tested for main effects for "talkers" vs. "normals," differences among the various measures, and the interactions among these factors. A significant interaction appeared (Wilks' lambda = .867, F =4.78, df =7,244), indicating that the group effects varied by scale. Then means for talkers and "normals" were compared on each scale by means of t tests. The assumption of normality in the "talker" group was not fully met. However, Hays (1973, p. 410) contends that departures from normality in this situation are not serious.
 


Results


 


Talkers differed from the "normal" sample in dominance, frequency, inhibition, attitude, and apprehension. These means are presented in Table 2. In this table, positive
------------------------
Table 2. about here
------------------------

numbers indicate more dominance, more time taken, and less inhibition. Talkers differed from "normals" on all three factors. Talkers were more dominant, reported talking with greater frequency, and were less inhibited. This indicates that talkers, at least the ones in our sample, had an accurate perception of their own behavior and that the Mortensen PVB scale is sensitive to these differences. The largest difference is the simplest one --- reports of frequency or "time taken."

The other characteristics displayed Table 2 might be said to "mediate" talking in some way or another. Talkers were less apprehensive than the normal sample and more argumentative, but the differences due to argumentativeness do not actually reach significance (p < .06). Given the violation of normality in the "talker" group, Hays (1973) would not conclude that the argumentativeness differences were significant ones. The other differences were large enough to satisfy a conservative requirement. Talkers had a more positive attitude toward communication but did not differ on locus of control or in self-esteem, when compared to "normals.".
 


Discussion


 


The present study differs from the questionnaire method of identifying talkers (Arntson et al.,1980; McCroskey & Richmond, 1995). It shows that when "compulsives" are selected by acquaintances on the basis of their communicative behavior, they differ from "normals" on some variables. Strong effects were found for dominance and frequency of talk, but much weaker ones in lack of inhibition. Moderate differences occurred present in apprehension scores, but no differences were found in self-esteem and locus of control. The latter two findings are different from other studies of communicative affect (Daly, 1978).

Talkers in the present study differed from "normals" when overall PVB scores were compared, but displayed less difference from "normals" on individual factors. Dominance and frequency responses showed the greatest differences. This might indicate that "compulsive" behavior is more complicated than overall mean comparisons might indicate. McCroskey and Richmond's (1995) finding that "talkaholics" reported that their talking was not a problem is quite different from our earlier study (Bostrom et al, 1990). There are probably enough compulsives in the population in general to warrant further study of this phenomenon.

The notion that talkers continue with their behavior in spite of negative social reinforcement is puzzling, but the differences in attitude toward talking scores offer some support for a "self reinforcement" hypothesis. Another explanation might be found in O'Keefe's classifications of message design logics (O'Keefe, 1992). Compulsive talkers may represent the most extreme cases of the "expressive" communicators. By definition, they could not be "rhetorical" or "conventional," since rhetoricals are receiver-oriented and conventionals are rules-oriented. Talkers violate both of these expectations. An interesting investigation might involve the content of the speech of compulsive talkers to see if their primary content conforms to that which mig;ht be expected of expressives. We might also explore whether talkers are "narcissistic" (Vangelisti, Knapp, & Daly, 1990). Both O'Keefe (1992) and Vangelisti et al. (1990) discuss the negative social outcomes associated with compulsive talkers.

Compulsive talking is probably not the most important problem confronting communication researchers today, but talkers and their families find the phenomenon unpleasant and would welcome amelioration. Future research ought to examine the content of talkers' messages, and strategies that coworkers and families take to adapt to this behavior.
 


References


 


Arntson, P. H., Mortenson, C. D., & Lustig, M. W. (1980). Predispositions toward verbal behavior
in task-oriented interaction. Human Communication Research, 60, 239-252.

Ayabe, H. J. (1979). The curvilinear relationship between reflection impulsivity and locus of
control. Journal of Genetic Psychology, 135, 309-310.

Bostrom, R. (1962). Classroom criticism and speech attitudes. Central States Speech Journal, 14,
27-34.

Bostrom, R. (1970). Cognitive, affective, and behavioral dimensions of communicative attitudes.
Journal of Communication, 20, 359-366.

Bostrom, R. (1981, May). Communication attitudes and communicative behavior. Paper presented
at the annual meeting of the International Communication Association, Boston, MA.

Bostrom, R. (1990). Listening behavior: Measurement and application. New York: Guilford.

Bostrom, R., Grant, N., Davis, W., & Einerson, M. (1990, May). Characteristics of compulsive
talkers: A preliminary investigation . Paper presented at the annual meeting of the International Communication Association, Dublin, Ireland.

Bostrom, R., Prather, M., & Harrington, N. G. (1998, July). Birth order and self-reports of communicative behavior. Paper presented at the annual meeting of the International Communication Association, Jerusalem, Israel.

Daly, J. A. (1978). The assessment of social-communicative anxiety via self-report: A comparison
of measures. Communication Monographs, 45, 204-218.

Daly, J. A., & McCroskey, J. C. (1984). Avoiding communication. Beverly Hills, CA: Sage.

Daly, J. A., McCroskey, J. C., & Richmond, V. P. (1976). Judgments of quality, listening, and
understanding based on vocal activity. Southern Speech Communication Journal, 41, 189-197.

Hamachek, D. (1982). Encounters with others. New York: Holt, Rinehart, & Wilson.

Hays, W. L. (1973). Statistics for the social sciences. New York: Holt, Rhinehart, & Winston.

Infante, D. & Wigley, C. (1986). Verbal aggressiveness: An interpersonal model and measure.
Communication Monographs, 53, 61-69.

Lefcourt, H. H. (1982). Locus of control: Current trends in theory and research. Hillsdale, NJ:
Lawrence Erlbaum.

McCroskey, J. C. (1970). Measures of communication-bound anxiety. Speech Monographs, 45,
192-203.

McCroskey, J. C. (1977). Oral communication apprehension: A summary of recent theory and
research. Human Communication Research, 4, 88-112.

McCroskey, J. C. (1978). Validity of the PRCA as an index of oral communication apprehension.
Communication Monographs, 45, 192-203.

McCroskey, J. C. & Richmond, V. P. (1987). Willingness to communicate. In J. C. McCroskey
& J. A. Daly (Eds.), Personality and interpersonal communication.(pp. 129-156). Newbury Park, CA: Sage.

McCroskey, J. C., & Richmond, V. P. (1995). Correlates of compulsive communication: Quantitative
and qualitative characteristics. Communication Quarterly, 43, 39-52.

Mortensen, C., Arntson, P., & Lustig, M. (1977). The measurement of verbal predispositions: scale
development and application. Human Communication Research, 3, 146-158.

O'Keefe, B. J. (1992). Developing and testing rational models of message design. Human Communication Research 18, 637-649.

Ollendick, D. G. (1979). Parental locus of control and the assessment of children's personality
characteristics. Journal of Personality Assessment, 43, 401-405.

Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University
Press.

Rotter, J. B. (1966). Generalized expectancies for internal versus external control of reinforcement.
Psychological Monographs, 8 (1), No. 609.

Spitzberg, B., & Cupach, W. (1984). Interpersonal communication competence. Beverly Hills, CA:
Sage.

Surlin, S. H. (1976). Fatalism and authoritarianism: Predictors of professional attitudes in
journalism. Journalism Quarterly, 53, 68-73.

Vangelisti, A. L., Knapp, M. L. & Daly, J. A. (1990). Conversational narcissisim. Communication Monographs, 57, 251-274.

Verplanck, W. (1955). The control of content of conversation. Journal of Abnormal and Social
Psychology, 51, 668-676.

Wiemann, J., & Knapp, M. (1975). Turn-taking in conversations. Journal of Communication, 25 (2), 75-92.
 



Table 1

Scales for the Various Factors in the PVB Test

_________________________________________________________________
 

FACTOR LOADINGS
 

I              II              III
                                                   I. Frequency.

.844       .439          -.257     3. When I am with other people I generally talk often.
.793       .501         -.384      7. In most social situations I generally speak quite frequently.
.695       .405         -.502    14. I generally find that I express myself quite freely.
.690       .289          .270     23. I prefer to keep my comments brief (negative).
.785       .315         -.303    24. I probably speak for shorter periods of time than the average person
(negative).

                                                  II. Dominance.

.465       .654        -.055     2. I have a tendency to dominate informal conversations with other people.
.533       .678        -.213     4. In most social situations I tend to direct the course of conversation.
.298       .718        -.149    10. In most social situations I tend to come on strong.
.345       .610        -.108    15. I try to take charge of things when I am with people.
.452       .765        -.215    17. I would describe myself as dominant in social situations.
.218       .616         .082    18. When I am with others I am inclined to talk forcefully.
.397       .648       -.317     20. In one-to-one conversations I tend to talk more than half the time.
.397       .648       -.399     21. In most social situations I tend to speak for long periods of time.
 

                                                  III. Inhibition

-.257     -.409       .542       1. I am inclined to let other people start conversations.
-.688     -.548        .602       5. When I am with others it generally takes me quite a while to warm up.
-.541     -.377        .741       6. I generally rely on others to keep conversations going.
-.301     -.055        .723       8. I tend to hesitate when I speak.
-.033      .098        .645      11. I find myself pausing often when I speak.
-.470    -.208        .719       13. I tend to feel inhibited when I talk to others.



Table 2

Means and Standard Deviations of Compulsive Talkers and Normals
on Varying Communicative Characteristics

________________________________________________________________________

                                   Talkers         Normals        t       df       p
 

Dominance                .79               -.41            5.618    262   .093
                                 (.90)             (1.00)

Frequency              1.75                .72              5.184   262    .114
                                (.82)              (.92)

Inhibition                1.05               .55               2.357  263      .023
                               (1.02)              (.91)

Attitude               26.44            23.88               2.963  263       .033
                              (4.02)             (4.91)

Argumentativeness 7.00         2.01                 1.033 256
                           (14.16)           (11.9)

Apprehension    27.00           33.63                 3.644  251    .047
                            (8.02)           (7.83)

Self Esteem         9.04              8.56                0.480    255
                           (5.06)            (4.87)

Locus of Control 10.63        10.71                 0.271    229
                              (3.13)        (3.97)

______________________________________________________________________________________
 

* p < .02, **p <.01, ***p < .001
 

1. McCroskey and Richmond (1995) use the term "talkaholic" to describe the compulsive talker.

2. Previously (Bostrom, Grant, Davis, & Einerson, 1990) we had established that: 1. When talkers are involved, there are no periods of silence. The talker fills space whenever possible. 2. Talkers are poor turn takers, ignoring other interactants generally. 3. Those who live with talkers simply interrupt when they wish to speak, and most talkers are used to this mode and are apparently not offended. 4. No particular subject matter seems to characterize talkers. They talk about baseball, their grandchildren, or nuclear disarmament with equal fervor.

3. Possible alternatives include the "Willingness to Communicate" (WTC) Scale (McCroskey & Richmond, 1987), its opposite, an "unwillingness to communicate" scale (Burgoon, 1976) and varying other attitudinal measures related to approach (Bostrom, 1990). The PVB, however, is clearly a self-report of actual communicative behavior. Items are phrased this way "When I am with other persons, I generally talk often" and "I generally find that I tend to talk quite freely." The distinction between a behavioral intention and a description of actual behavior may seem to be a narrow one, but is quite important. McCroskey and Richmond (1987) dismiss the PVB on the grounds of its supposed isomorphism with apprehension. This is based on Mortensen's report of a .67 correlation of PVB with apprehension. Other studies (Bostrom, Grant, Davis, & Einerson, 1990) report lower figures (.61), but it is clear that apprehension and behavioral reports are closely related. Being closely related is certainly not the same as being the "same," however. The WTC is also related to apprehension (r =.51; McCroskey & Richmond, 1990).

4. We have no problems with the term "dominance," which is relatively easy to designate, and "frequency," which we originally labeled "time in the interaction." Mortenson et al.'s term "engagement" factor seems to be more an "inhibition" index than anything else so we have made that alteration.

5. This index consisted of five Likert-type items with 5-point intervals. They were designed to determine the extent to which individuals "enjoy" various aspects of talking. The five items were: 1) I like to talk to people I haven't met before, 2) I'd rather be talking about anything than not talk at all, 3) I like to tell people about myself, 4) I like to get in the last word, and 5) I really enjoy talking. The Attitude toward Talking index was highly reliable (Cronbach's alpha = .915).

6. It was clear tht these persons were indeed compulsive talkers. At every point in their interactions they generated a stream of nonstop verbalizations. In the initial telephone calls, the "talkers" were reluctant to hang up. The researcher that called them did not want to offend them by say "Goodbye, now," and some of the calls lasted 30 minutes or more. When respondents finished filling out the questionnaire booklet, they were reluctant to leave the site. They continued talking out into the hall and down the stairs. While the sample may not have been randomly drawn, their behavior showed that they fit the definition of "compulsive" very well.