General Results Discussion
This section discusses the results reported in Tables 2–4. Specifically, scale reliability comparisons, mean distance and standard deviation ranges, and changes in dyad distance are covered in this section.
The alpha coefficient comparisons in Table 2 are indicative of the applicability of face-to-face interpersonal scales to a virtual CMC environment. The alpha coefficient comparisons are as follows: low-context communication, alpha =.86 (CMC) compared to .85 (FTF); high-context communication, alpha =.77 (CMC) compared to .94 (FTF); general conversational appropriateness, alpha =.75 (CMC) compared to .80 (FTF); specific conversational appropriateness, alpha =.86 (CMC) compared to .85 (FTF); social attraction, alpha =.79 (CMC) compared to .84 (FTF). In general, the alpha coefficients were all relatively high and were similar to those obtained in face-to-face environments. Interestingly, with the exception of high-context communication, social attraction, and general conversational appropriateness, the reliabilities were all slightly higher in a CMC environment. In fact, only the high context communication alpha was more than .05 different in CMC versus face-to-face environment. The high-context communication scale may be the most suspect in terms of application in a CMC environment. Nevertheless, the reliability of this scale was still acceptable (.77). While these results do not revalidate the scales used (a factor analysis needs to be performed to determine to see if the factor loadings change in a CMC environment), they do indicate that the instruments are reliable in a CMC environment.
From Table 3, the mean distance range was 159.34 pixels at Time 1 and 171.69 pixels at Time 2. Overall mean distance between these two ranges was 165.52 pixels. Given the constraints of this study (i.e., resolution of screen and size of palace window), 1 pixel is equivalent to .0179245288 inches. The mean distances(in pixels) translate to 2.85 inches at Time 1 and 3.07 inches at Time 2. This indicates that there was a slight increase in distance from Time 1 (the distance at the beginning of the conversation) to Time 2 (the distance at the end of the conversation) of 0.22 inches. Overall this is a slight increase in distance. The standard deviations were 58.81 pixels (1.05 inches) at Time 1 and 73.43 pixels (1.31 inches). This indicates that there was variation around the mean distances at each time period and that they moved slightly more (.26 inches) at Time 2. There was movement over time in the dyadic conversations and the changes in dyad distance examines these movements.
From Table 4, the movement of dyads from Time 1 to Time 2 were nearly equally distributed. Fifty-two individuals moved closer together, sixty individuals moved farther apart, and forty-eight individuals did not move during the 7–10 minutes conversation. This indicates that there was movement both closer together and farther apart over time. The mean distance range was thus a combination of individuals who moved closer and farther apart. The mean distances were -55.00 pixels (negative values signify closer together; Time 2-Time 1) for individuals who moved closer together and 80.78 pixels for individuals who moved farther apart. The mean distances (in pixels) translate to .99 inches for the individuals who moved closer together and 1.45 inches for individuals who moved farther apart. Standard deviations of these changes indicates similar results. The standard deviations were 42.47 pixels (.76 inches) for the closer group and 83.03 pixels (1.48 inches) for the farther group. These results indicate, in general, that individuals moved .46 inches farther apart and had a higher range of movement when they moved apart. This finding may be a result of the fact that individuals had more room to move farther apart than closer together, given that the maximum distance for all conversations was 8.44 inches and the fact that they initially started 2.85 inches apart. This could also be explained by possibility of “invasion of personal space.” This explanation would suggest that there is a “intimate” personal zone which restricts movement to close distance ranges. Given that the minimum distance for all conversations was 37 pixels (.67 inches), this indicates that there is a minimum distance that normal conversation can occur (i.e., otherwise avatars would be on top of each other and, hence, undifferentiable). There is some evidence of what happens when avatars get too close together. Figure 2 provides an example of this occurrence.
In Figure 2, an avatar is partially blocking the top of another avatar. The blocked avatar's response is “get off my forehead.” This indicates that there is some form of minimal distance range and that this may be indicative of a intimate zone infraction analogous to the intimate proxemic range (Hall, 1966). It also indicates a translation of virtual representation of everyday phenomena, as the top of an avatar is translated into “my forehead.”Suler (1996) describes this phenomenon in detail:
Blocking: Members consider it a social faux pas to place your avatar on top of or too close to another person's prop. Unless the person is a friend who's in a mood to be close, it's an invasion of personal space. “Please get off me!” and “You're sitting on me!” are two common complaints. Again, some naive users do this without knowing it is inappropriate, or the person may be lagging and unable to move. But some hostile people deliberately accost others by blocking them.“
Finding 1. Greater spatial distance (over time) signified higher general conversational appropriateness
Finding 1 indicates that individuals find spatial distance as a regulator of whether the other person is, in general, conversationally appropriate. The farther individuals position their avatars apart, the more conversationally appropriate they are perceived. This finding is not surprising and mirrors what we experience in face-to-face conversations. Personal space is considered as territorial and when this personal space is invaded, one's behavior could be considered inappropriate. Sommer (1969) posits that a person who moves closer to another in conversation is perceived as more approachable up to a certain point. It is argued that there is a certain point where it is considered inappropriate (i.e., obnoxious, rude, overbearing, etc.) and moving farther from a person does not encroach upon personal space and would therefore not be considered inappropriate. This finding indicates that it may be considered inappropriate for strangers to move too close in initial conversations, but that moving farther apart would not be considered inappropriate. Interestingly, none of the within-group analyses (i.e., closer, farther, or stable groups) replicated this finding; that greater spatial distance and higher general conversational appropriateness was not due to any one type of movement (i.e., closer or farther). This suggests that other factors play a role in this process.
Finding 1a–for groups that moved closer together, greater distance signified higher uncertainty reduction (i.e., attributional confidence)–provides insight into the Finding 1 (which found no significant results for this group). The closer groups seemed to use greater spatial distance as a form of uncertainty reduction. It should be noted that no other significant results were found relating uncertainty reduction to spatial distance movement. This indicates that uncertainty reduction (i.e., attributional confidence) was found to characterize those who moved closer together, but not those who moved farther apart or remained stable. For those who moved closer, there seemed to be a “too close” range and by moving into this range, an individual's uncertainty may have increased; that greater distance reduced uncertainty for the group that moved closer together. Relating this to Finding 1, it is posited that uncertainty reduction (i.e., attributional confidence for both high- and low-context communicators) can explain higher general appropriateness for the group which moves closer together. Recall that the uncertainty reduction measures (both low- and high-context communication) were based on how people perceives themselves in their interactions with others. This could be attributed to the fact that uncertainty reduction measures one's own perceptions of uncertainty versus the role of the perceived other (as in social attraction or conversational appropriateness).
It should be noted that Finding 1a was observed for those who are both low- and high-context communicators. Further tests indicated that high-and low-communication were indeed different constructs, although they were highly correlated. The correlation between low- and high-context communication in CMC (.64) is similar to that of face-to-face interactions (.68) (Gudykunst & Nishida, 1986). The mean and standard deviations, however, reveal differences from the face-to-face environment. For low-context communication items, attributional confidence in the present study yielded MLC= 29.18 and SDLC=19.84 versus MLC=40.70 and SDLC=38.56 in Gudykunst and Nishida's (1986) face-to-face US strangers. For high-context communication items, attributional confidence yielded MHC=39.10 and SDHC=24.39 in the present study versus MHC=42.59 and SDHC=58.23 in Gudykunst and Nishida's (1986) face-to-face US strangers. The low-context communication mean was much lower than in the face-to-face context. This could be a result of the lack of social cues in a CMC environment or the relatively short time duration (7–10 minutes) spent interacting. Another explanation could be that there may have been no anticipation of future interaction and information exchange between strangers (see Kellermann, 1986). As a result, the amount of uncertainty reduced would be smaller in the CMC environment given the reduced motivation in reducing this uncertainty (Kellermann & Reynolds, 1990). The standard deviations in the CMC environment were also much lower. This could be a result of the relatively low means (i.e., the range of movement would be limited), but also may be indicative of the reduced social cues and limited discussion time. The relatively high standard deviations reported by Gudykunst and Nishida (1986) may also be a product of higher cultural variability among their sample of US students.
Finding 1b, that for the group that moved farther apart over time, greater spatial distance signified higher specific conversational appropriateness, provides another piece of information regarding Finding 1. Given that specific and general conversational appropriateness were found to be different measures, this finding indicates that the farther-apart groups used greater spatial distance as a form of specific conversational appropriateness. It should be noted that no other significant results were found relating specific conversational appropriateness to spatial distance movement. This indicates that specific conversational appropriateness was found to be related to those who moved farther apart, but not for those who moved closer together or remained stable. Recalling that specific conversational appropriateness dealt with items such as whether one said anything embarrassing (reverse coded), this measure would seemingly follow the idea that one cannot be considered inappropriate if they move farther apart, but could be deemed inappropriate if they invaded someone's territory. The mean value for specific conversational appropriateness was rather high (M=6.33 on a 7-point scale), indicating that most individuals were appropriate in their specific conversations. Given that the subjects knew that the conversations were being monitored, it is not surprising that there was very little specifically inappropriate behavior. In a setting where anonymity is more prevalent (i.e., a public Palace), it would seem that specifically inappropriate behavior such as flaming, SHOUTING, or using foul language would occur more frequently.
Finding 1a provided insight into the group which moved closer together and Finding 1b provided insight into the group which moved farther apart. These specific within-group comparisons are evidence toward the argument that closer groups had higher reduced uncertainty at greater distance intervals and that farther groups had higher perceived specific conversational appropriateness at greater distance intervals. Finding 1c–for the groups that moved over time (i.e., closer and farther groups), greater spatial distance signified higher general conversational appropriateness–lends insight into the relationship between Finding 1a and Finding 1b. General conversational appropriateness can be seen as a mediating variable between closer and farther groups; it cannot explain within-group differences, but can explain between-group differences. Figure 3 depicts a model incorporating the findings in this section establishes the relationship between variable leading to greater spatial distance over time:
Figure 3 provides a model which allows for testing the findings discussed in this section. The model depicts the relationship of greater distance movement as a combination of higher Uncertainty Reduction for those that move closer together and higher Specific Conversational Appropriateness for those that move farther apart. The resultant interaction is that of higher General Conversational Appropriateness. If one wants to reduce uncertainty in a virtual environment, the model suggests moving closer to another avatar over time, but not too close. If one wants to be specifically appropriate, then the model suggests moving farther from an avatar over time. The result could incorporate different phases in an online relationship and also suggests that spatial movement can be utilized in different ways depending on the purpose of the relationship.
Utilizing the model in a practical setting provides some interesting applications. One example of applying the model would be to the case where another person misinterprets a message to be rude or offensive (a common misunderstanding because non-verbal cues are missing). In this case, the model suggests moving farther away from this person while continuing the conversation. At this point one could “back off,” using the distance between avatars, and try to clarify your meaning. Moving closer to the other person would not be suggested in this case. As another example, suppose you would like to get to know a person better or would like to learn more about them (i.e., reduce uncertainty). The model suggests that moving closer would be an acceptable spatial solution, but that one should be careful about moving too close–that there should still be some distance between yourself and the other person (the actual distances of these movements will be discussed in the last part of this section). Incorporating both of these examples, the model predicts that one would be considered as generally appropriate by using spatial distance as a means to establish relationships in a virtual CMC environment. In the case where one wants to be perceived as generally appropriate while conversing in a graphical chat environment with a boss or respected figure (i.e., someone by whom one would want to be perceived as generally appropriate), one could adopt the above examples as a strategic form of spatial communication. By utilizing Uncertainty Reduction Theory and Specific Conversational Appropriateness, one adds a new non-verbal communication channel, that of spatial distance, to their online repertoire.
Finding 2. The relation between spatial distance and social attraction (i.e., liking) is a positive parabolic function
Finding 2 relates the absolute distance to perceived social attraction. This finding should be differentiated from Finding 1 which linked the change in distance over time to general conversational appropriateness. This finding further specifies the distance ranges in virtual computer-mediated communication. This spatial distance range was determined to be between 37 and 471 pixels (0.66 inches and 8.44 inches, given the conditions in this study). This indicates that there is a minimum and maximum distance by which communication occurred in this study.
The mean value of social attraction (7-point scale) was found to be 4.96 and the standard deviation was 1.13. This result shows that individuals perceived the other person as somewhat attractive (i.e., likeable). There was some variation around this value, but individuals did not generally perceive the other as socially unattractive. This changed when looking at the distance at Time 2 compared to social attraction. The quadratic regression indicated that there was a positive parabolic function which described the relation between social attraction and absolute spatial distance at Time 2. The equation relating this relationship was:
Equation 2. Social Attraction (y) as a Function of Pixel Distance (Unstandardized)
where: y= perceived social attraction; 2.55 × 10−5 is the (unstandardized) slope of the parabola; 219 is the vortex (in pixels), and 4.76 is the unstandardized coefficient (constant). It should be noted that this equation is unstandardized and should be used for comparison purposes only (i.e., the values have no absolute meaning). Equation 2 provided a means to predict social attraction given the absolute distance (in pixels) of avatars and was illustrated in Figure 1. This equation states that as the absolute distance between avatars approaches the value of 219 pixels (approximately 4 inches, in this case), the perceived social attraction is decreased.
The results indicate that there are three distance ranges which occur in 2-D virtual communication. The first range is termed as the “Danger Zone,” or middle-range distance. This is the range you do not want to be in if you want to be perceived as socially attractive: The Danger Zone is 175–274 pixels or 3.12–4.88 inches in the context of this study. The second range is termed the “Close-Range Zone.” This distance yields higher social attraction: The Close-Range Zone is 37–174 pixels or 0.67–3.10 inches in the context of this study. The third range is termed “Far-Range Zone.” This distance also yields higher social attraction. The Far-Range Zone is 275–471 pixels or 4.90–8.44 inches in the context of this study.
These results are interesting because they provide a response to RQ2, which asked about the demarcations in range. The significant results indicate that there are three separate distance ranges in relation to perceived social attraction. Further tests indicated that the Danger Zone was significantly different than both the Close-Range and Far-Range Zone. Interestingly, the Close-Range and Far-Range Zone were not significantly different in relation to social attraction. These results suggests that the Close-Range and Far-Range Zones reflect higher social attraction, whereas the Danger Zone (Middle-Range) reflects lower social attraction (Results 2a and 2b, respectively).
In light of past research on proxemics, Hall (1966) specifies four distance ranges of Americans: (1) intimate (0–18 inches); (2) personal (18 inches-4 feet); (3) social (4 feet-12 feet); and (4) public (12 feet-25 feet). These four categories are based on observations that both animals and man exhibit territorial behavior. The intimate range from 0–18 inches is largely characterized by non-verbal communication cues which may be absent in a Palace-type computer-mediated environment. Although it cannot be determined given the results in this study, perhaps the three ranges found in this represent personal, social, and public distances.
Equilibrium Theory (Argyle & Dean, 1965), posits that intimacy and distance vary together–the closer the distance, the greater the intimacy; and, conversely, the greater the distance, the lower the intimacy. The findings in this study indicate that spatial distance versus attraction (i.e., liking) reflects a parabolic function; that higher perceived social attraction occurs in Low-Range and High-Range individuals and that lower perceived social attraction occurs in middle-distant individuals. This is puzzling. It differs from what we experience in face-to-face contexts and (at least initially) is difficult to believe. What is happening here?
First of all, it should be noted that this relation is in terms of perceived social attraction. This measure was found by examining questions such as “I think he (she) could be a friend of mine.” It has been found that social attraction is highly correlated to general conversational appropriateness (r=.57) and specific conversational appropriateness (r=.52) (Canary & Spitzberg, 1987). This study found social attraction to be significantly correlated with general conversational appropriateness, r=.48 (p≤.01) and specific general conversational appropriateness, r=.47 (p≤.01) (see Table 5). Table 5 also provides the significant correlations between social attraction and low-context communication (r=.25; p≤.01) and high-context communication (r=.32; p≤.01). These results suggests that social attraction has some of the same underlying causal mechanisms as that of general conversational appropriateness. Recall that we found that greater spatial distance over time results in higher specific conversational appropriateness. Correlating specific conversational appropriateness with the social attraction range (i.e., parabola) reflects the “A” relationships depicted in Figure 4.
Logically, there are three ways to move farther apart (left to right in Figure 4) and end up out of the Danger Zone: (1) Staying within the Close-Range Zone (1a); (2) Moving into the Far-Range Zone (1b); or (3) Staying within the Far-Range Zone (1c). However, the parabola indicates that (1a), staying in the Close-Range Zone, results in a decrease in social attraction. Because a significant decrease in specific conversational appropriateness was not found Figure 4 does not include (1a). Correlating the uncertainty reduction results with the social attraction range (i.e., parabola) reflects the “1” relationships in Figure 4. Similarly, there are three ways to move closer together (right to left in Figure 4) and end up out of the Danger Zone: (1) Staying in the Close-Range Zone (2a); (2) Moving into the Close-Range Zone (2b); or (3) Staying in the Far-Range Zone (2c). However, the parabola indicates that (2c), staying in the Far-Range Zone, results in a decrease in social attraction. Because a significant decrease in uncertainty reduction was not found, Figure 4 does not include (2c). In Figure 4, the distance ranges for this study are provided (37–471 pixels). The three zones are also provided: 37–274 pixels is the Close-Range Zone; 174 to 275 pixels is the Danger Zone; and 275–471 pixels is the Far-Range Zone.
As an example, consider first the case where people move farther apart (left to right) from the Danger Zone into the Far-Range Zone (arrow 2a). In this case, the distance indicates that an individual is moving into a more socially attractive zone via conversational appropriateness. This can be explained as a strategy of increased specific conversational appropriateness. As another example, consider the case where people move closer together (right to left) into the Close-Range Zone (arrow 2b). In this case, the distance indicates that an individual is moving into a more socially attractive zone through conversational appropriateness. This can be explained as a strategy of increased uncertainty reduction (i.e., increased attributional confidence). Figure 4 thus represents a culmination of the study as it correlates the Social Attraction parabolic function illustrated in Figure 1 with the elements of the model specified in Figure 3–the general conversational appropriateness results specify the direction and change in distance, while the social attraction results specify the actual range (or zones) to which these individuals moved.