SEARCH

SEARCH BY CITATION

Keywords:

  • causal uncertainty;
  • social interactions;
  • uncertainty reduction

Abstract

  1. Top of page
  2. Abstract
  3. STUDY 1
  4. STUDY 2
  5. STUDY 3
  6. GENERAL DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES

Causal uncertainty (CU) refers to persistent doubts people have about their ability to understand causes of social events. Although such confusion about social dynamics should affect social exchanges, previous research has been limited to the realm of social cognition (i.e., computer-based studies exploring perceptions of hypothetical others). In three studies, we explored CU effects during real-time social interactions with unacquainted conversational partners. We found that high CU participants perceived their conversations and conversational partners more negatively than did low CU participants and that these negative social perceptions stemmed from an inability to sufficiently reduce their cognitive uncertainty. Copyright © 2012 John Wiley & Sons, Ltd.

People often need to determine the reasons for people's actions so they can respond appropriately. However, this process can be quite difficult because social events are complex and the specific causes for people's behavior can be ambiguous and difficult to prove (Weary & Edwards, 1996). Interpreting behavior during interactions with strangers, or initial interactions, is further complicated by the fact that we possess little information about our partners, increasing our uncertainty (Berger & Calabrese, 1975).

Berger (1979) proposed that we experience cognitive uncertainty (i.e., uncertainty about others' backgrounds, attitudes, and beliefs) and behavioral uncertainty (i.e., uncertainty about how to act or react in social situations) during these interactions. In turn, this uncertainty produces cognitive and emotional discomfort (Heath & Bryant, 2000). Thus, according to Berger and Calabrese (1975), a primary objective during initial interactions is uncertainty reduction. Specifically, we engage in two forms of uncertainty reduction: proactive uncertainty reduction, so we can predict how our partner will behave in the future, and retroactive uncertainty reduction, so we can explain our partner's behavior and select the appropriate response. This information is then incorporated into behavioral schemas that help people negotiate future interactions (Douglas, 1991).

However, some individuals are unable to reduce their uncertainty as effectively as others, making social interactions more difficult. For example, Douglas (1991) argued that people with more global uncertainty (i.e., confusion about general acquaintanceship processes) are unable to develop coherent schemas to guide their behavior and, therefore, experience less uncertainty reduction as interactions progress. Consequently, they are more likely to avoid initial interactions altogether, resulting in greater loneliness.

Similarly, some individuals experience persistent doubts about their ability to understand causes of social events; these doubts have been labeled causal uncertainty (CU; Weary & Edwards, 1994). Although CU pertains to uncertainty about social situations in general, such as global uncertainty, CU is associated with greater loneliness (Jacobson, Weary, & Chakraborti, 1997; Weary & Jacobson, 1997) and less satisfying interactions (Boucher & Jacobson, 2005). Thus, CU may lead to problems with uncertainty reduction similar to those associated with global uncertainty. Furthermore, CU is moderately and positively correlated with intolerance for and discomfort with ambiguity (Weary & Edwards, 1996), so causally uncertain people may find a lack of uncertainty reduction especially frustrating, exacerbating their negative perceptions of their conversations.

Unfortunately, although causally uncertain people's confusion about social dynamics should have the greatest impact during social interactions (particularly initial interactions), virtually all CU research has been in the realm of social cognition, and little attempt has been made to observe CU effects during real-time social interactions. In most previous research, participants made judgments about hypothetical targets at isolated computer terminals and merely had to process information that was given to them. Consequently, how CU affects the uncertainty reduction process or perceptions of interactions and interaction partners remains unclear. We addressed this limitation in the current research by exploring how CU affects real-time social exchanges between strangers.

Causal Uncertainty

Weary and Edwards (1994, 1996) and Weary, Tobin, and Edwards (2010) argue that everyone can experience CU beliefs and associated feelings of confusion and puzzlement. However, most people tend to be confused by other people's behavior only when it is unusual or horrific. Consider people's reactions to school shootings. Following these events, most people are puzzled about why someone committed such an act, why innocent people died, and what could have been done to prevent the incident (or prevent future incidents). Although this uncertainty is typically temporary, CU beliefs and feelings occur more frequently for some people and eventually become chronically accessible so that more mundane events are sufficient to activate their uncertainty (Edwards, Wichman, & Weary, 2008).

Specifically, CU feelings arise when individuals believe they do not have enough information to confidently identify the cause of a particular event (Weary & Edwards, 1996; Weary et al., 2010). Again, because such uncertainty is aversive (Heath & Bryant, 2000), people tend to pursue various strategies to reduce negative feelings associated with their uncertainty (Weary & Edwards, 1996). One strategy is to reduce the discrepancy between their actual and desired knowledge states (i.e., adopt an accuracy goal), prompting a more extensive search for and processing of social information. For example, causally uncertain people persist longer during social information processing tasks (Jacobson, Weary, & Lin, 2008), rely less on heuristic processing when making social judgments (Weary, Jacobson, Edwards, & Tobin, 2001), and are more likely to scrutinize causal arguments (Tobin & Weary, 2008) and to incorporate important situational information into dispositional attributions (Weary, Vaughn, Stewart, & Edwards, 2006). However, as previously mentioned, these studies have involved only hypothetical situations, and thus, it is unclear how such CU processes apply in real-time social interactions.

Causal Uncertainty in Social Interactions

Arguably, such effortful processing should produce greater uncertainty reduction, yielding more positive social interactions (Berger & Calabrese, 1975). Indeed, Fast, Reimer, and Funder (2008) found that people who pay more attention to and think about social causation (i.e., who are more attributionally complex) are perceived more positively by others and have more satisfying interactions with strangers. However, CU is positively related to loneliness and shyness (Jacobson et al., 1997), suggesting that it may be associated with interpersonal difficulties. In fact, in one of the few studies to explore how causally uncertain people perceive their social interactions, high CU people reported having less intimate, more superficial, and less satisfying interactions than did low CU people (Boucher & Jacobson, 2005).

Why would effortful processing lead to positive outcomes for people high in attributional complexity but negative outcomes for high CU people? Although both of these constructs involve effortful processing of social information, they are negatively correlated (Tobin, 2010), and the effortful processing is motivated by different goals. Whereas attributionally complex people are intrinsically interested in understanding causes of others' behavior (Fletcher, Danilovics, Fernandez, Peterson, & Reefer, 1986), causally uncertain people do not enjoy thinking about causes of events and only do so in an effort to reduce the cognitive and emotional discomfort associated with their heightened uncertainty (Jacobson et al., 2008; Weary & Edwards, 1996). Thus, for attributionally complex people, effortful processing is more proactive, but for causally uncertain people, the same behavior is more reactive.

Furthermore, attributionally complex individuals' effortful processing likely leads to greater uncertainty reduction. However, given their persistent doubts about their ability to identify the causes of social events, information seeking efforts may not successfully reduce causally uncertain people's uncertainty. That is, as Brashers (2001) noted: “I may have a great deal of information about a topic, I may have an amount of information that other people would deem sufficient to make a decision or to predict another's behavior, and I even may have all the information that is currently available, yet I still may feel uncertain” (p. 478). In sum, high CU people may have difficulties trusting their ability to predict and explain others' behavior using information they acquire about them, producing little uncertainty reduction.

According to communication theories of uncertainty, such a lack of uncertainty reduction should lead to more problematic interactions. For example, Gudykunst's (1993, 1995) Anxiety/Uncertainty Management Theory proposes that communication is effective to the extent that communicators' uncertainty is between their minimum and maximum uncertainty thresholds. If they surpass the maximum threshold, people should withdraw from the conversation because they have no confidence in their ability to predict their partner's behavior. Similarly, when uncertainty falls below the minimum threshold, people become overconfident, also interfering with effective communication. Consistent with this theory, lower levels of uncertainty during social exchanges are associated with greater conversational effectiveness (Gudykunst & Nishida, 2001) and higher levels of communication satisfaction (Neuliep & Grohskpof, 2000).

Thus, from this perspective, if initial interactions are sufficient to activate causally uncertain people's doubts and they are unable to manage their uncertainty, they should perceive their conversations and conversational partners more negatively. We tested this idea in three studies.

STUDY 1

  1. Top of page
  2. Abstract
  3. STUDY 1
  4. STUDY 2
  5. STUDY 3
  6. GENERAL DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES

In Study 1, we examined CU effects during dyadic interactions between strangers. Participants first engaged in a brief unstructured interaction. However, interactions between strangers tend to be relatively superficial, and more intimate aspects of a conversation, such as self-disclosure and emotional support, are critical features to building successful relationships (Rook, Pietromonaco, & Lewis, 1994). Thus, to make interactions more meaningful, participants engaged in two additional conversations: one in which participants discussed a personal problem with their partner and another in which participants listened to their partner's problem. Assessing participants' perceptions of these different conversations provided us with a more comprehensive understanding of how CU affects social exchanges.

Specifically, we examined how CU influences participants' perceptions of conversational effectiveness (i.e., if participants achieved their specific goals during the conversation) and conversational appropriateness (i.e., if participants violated conversational norms) for each conversation. Because these measures do not distinguish between dyad members, participants also rated their partner's overall interpersonal competence after the last conversation to obtain more specific information about how causally uncertain people perceive conversational partners. Given that higher levels of uncertainty are associated with less effective (Gudykunst & Nishida, 2001) and less satisfying (Neuliep & Grohskpof, 2000) communication, we hypothesized that because of their confusion about social dynamics, causally uncertain participants would rate all three conversations as less appropriate and effective and would rate their partners more negatively (i.e., less interpersonally competent).

Furthermore, because Uncertainty Reduction Theory predicts that similarity should lead to greater uncertainty reduction (Berger & Calabrese, 1975) and, in turn, more positive interactions, we also examined if partners' CU would moderate participants' perceptions of the conversation. Because these analyses were largely exploratory, we made no specific predictions for these effects. However, Rosenblatt and Greenberg (1991) have shown that depressed people have more positive interactions with other depressed people and CU is positively correlated with depression (Weary & Edwards, 1994, 1996). Thus, high CU participants may perceive their conversations more positively when they are interacting with another high CU participant.

Method

Participants

Participants were 120 undergraduate students (100 women, 20 men) recruited from introductory psychology classes at Queen's University in Canada and paired to form 60 previously unacquainted same-sex dyads. Participants had a mean age of 18.54 years (SD = 2.33), ranging from 17 to 36. Participants received partial course credit or $CAN20.00 for their participation.

Measures
Causal Uncertainty Scale (Weary & Edwards, 1994)

The Causal Uncertainty Scale (CUS) is a measure of chronic individual differences in CU beliefs, consisting of 14 statements (e.g., “When I see something good happen to others, I often do not know why it happened”) with response options on a 6-point scale from 1 (strongly disagree) to 6 (strongly agree). Responses are summed so that higher scores indicate greater CU. This scale had good internal consistency in the current sample (α = .86), and participants had a mean score of 38.92 (SD = 9.65), ranging from 16 to 72; these scores are consistent with other CU research (e.g., Weary & Jacobson, 1997).

Beck Depression Inventory II (Beck, 1996)

The Beck Depression Inventory II (BDI-II) is an assessment of the severity of depressive symptoms, consisting of 21 items on a 4-point scale, ranging from 0 to 3. Responses are summed with higher scores indicating more severe depressive symptoms. The BDI-II had excellent internal consistency in the current sample (α = .89), and the mean score was 8.19 (SD = 6.16), ranging from 0 to 27. Because CU is related to depression (Weary & Edwards, 1994, 1996),1 which also influences people's social interactions (see Segrin & Abramson, 1994), we controlled for BDI-II scores in all statistical analyses; all reported effects were significant when controlling for depression.

Conversational Appropriateness Scale (Spitzberg & Canary, 1985)

The Conversational Appropriateness Scale (CAS) is a measure of situational perceptions of conversational appropriateness, consisting of 20 items inquiring about awkwardness (e.g., “S/he was a smooth conversationalist”) and embarrassment (e.g., “I was embarrassed at times by her/his remarks”) that are rated on a 5-point scale from 1 (strongly disagree) to 5 (strongly agree). Scores are summed so that higher scores indicate a more appropriate interaction. The CAS had good internal consistency across all three conversations (α = .88 to .91).

Conversational Effectiveness Scale (Spitzberg & Canary, 1985)

The Conversational Effectiveness Scale (CES) is a measure of situational perceptions of conversational effectiveness, consisting of 20 items inquiring about success (e.g., “Our conversation was very beneficial”), control (e.g., “I was in control of the conversation”), and goal achievement (e.g., “I got what I wanted out of the conversation”). Items are rated on a 5-point scale from 1 (strongly disagree) to 5 (strongly agree). Scores are summed so that higher scores indicate a more effective interaction. The CES also had good internal consistency across all three conversations (α = .84 to .86).

Conversational Skills Rating Scale–Coactor Form (Spitzberg, 1995)

The Conversational Skills Rating Scale (CSRS) is a situational measure of perceived competence during social exchanges, and the coactor form (CF) allows participants to rate their partner's competence. Twenty-five items (e.g., speaking rate, use of eye contact) are rated on a 5-point scale from 1 (inadequate) to 5 (excellent), and the remaining five items tap participants' general impressions of their partner's conversational skills (e.g., good versus poor conversationalist). Factor analyses supported a one-factor structure, so we summed all 30 items to obtain a global measure of conversational skill; this global scale had excellent internal consistency (α = .94).

Procedure

Prior to entering the lab, each dyad member was randomly assigned to discussant order (i.e., participant “A” discussed his/her personal problem first and participant “B” discussed his/her problem next). Upon arriving, participants were seated across their partner, while the experimenter monitored interactions from an adjacent room and provided instructions using a one-way intercom.

Dyads were instructed that they would have three conversations and that one conversation would involve discussing a moderately distressing personal problem (adapted from Rook et al., 1994). They first described this problem in writing to ensure that the type of problem chosen was not influenced by participants' interactions with their partner. Then, participants began Conversation 1, a five-minute unstructured interaction in which participants were instructed to get to know each other. After five minutes, the experimenter interrupted the interaction and instructed participants to complete the CAS and CES at isolated computer stations.

Next, participant A was instructed to discuss his or her personal problem with participant B. After eight minutes, the experimenter interrupted the interaction and again instructed participants to complete the CAS and CES. Upon completing these questionnaires, B was instructed to discuss his or her personal problem with A. Again, after eight minutes, the experimenter interrupted the interaction and instructed participants to complete the CAS and CES, as well as the CSRS-CF, CUS, and BDI-II. Finally, participants were fully debriefed.

Data Analysis

The data were analyzed using Kashy and Kenny's (2000) Actor–Partner Interdependence Model (APIM), which examined the degree to which participants' CU affected their perceptions of the conversation (actor effect) and the degree to which their partner's CU affected these perceptions (partner effect). In other words, a significant actor effect indicates that the participant's CU predicted his or her perceptions above and beyond the effects of the partner's CU, whereas a significant partner effect indicates that the partner's CU predicted the participant's perceptions, above and beyond the effects of his or her own CU.

For the following analyses, dyads were treated as the unit of analysis, and discussant order (A versus B) was treated as a repeated measure within the dyad. Furthermore, with the exception of analyses on CSRS-CF scores, which were obtained only once, conversation was treated as another repeated-measure variable within the dyad.

To examine the actor and partner effects for CU, participants' continuous CUS scores and their partner's continuous CUS scores were included as covariates in the model. Therefore, the actor effect was measured by the main effect of participant CUS, and the partner effect was measured by the main effect of partner CUS. Furthermore, to examine the possible moderating effect of matching participants on CU, we also included the two-way interaction between participant and partner CUS.

To examine the effect of discussant order and conversation, we also included an effects-coded order variable (A = +1; B = −1) and two orthogonal contrast-coded conversation variables. The first variable, Conversation Type (i.e., unstructured versus problem-oriented), compared perceptions of Conversation 1 (+.67) to those of Conversations 2 and 3 (−.33). The second variable, Problem Type (i.e., A's problem versus B's problem), compared perceptions of Conversation 2 (+.5) to those of Conversation 3 (−.5). Finally, all corresponding interaction terms were included in the model.2 To reduce multicollinearity between the main effect and interaction terms, all continuous predictor and outcome variables were standardized.

Results and Discussion

As predicted, the actor effect was significant for CAS, CES, and CSRS-CF scores. Consistent with the Anxiety/Uncertainty Management Theory (Gudykunst, 1993, 1995), relative to low CU participants, high CU participants rated all three conversations as less appropriate, β = −.22, t(83.10) = −3.09, p = .003, and less effective, β = −.23, t(99.68) = −3.90, p < .001, and rated their partners as less conversationally skilled, β = −.30, t(101.61) = −3.16, p = .002. Actor effects on CAS and CES scores were not moderated by conversation type (β = .08, p = .42, and β = −.01, p = .93, respectively) or problem type (β = .10, p = .39, and β = −.08, p = .46, respectively), suggesting that causally uncertain participants' negative perceptions of conversational appropriateness and effectiveness were consistent across all three conversations.

In contrast, partner CU did not affect participants' perceptions of conversational appropriateness (β = .03, p = .70) and conversational effectiveness (β = .09, p = .11) or of their partner's conversational skill (β = .05, p = .59). These preliminary findings suggest that although high CU participants generally perceived their interactions negatively, their partners did not share these negative perceptions. Furthermore, unlike matching effects observed with depression (Rosenblatt & Greenberg, 1991), the interaction between participants and partner CU was not significant for conversational appropriateness (β = .02, p = .83), conversational effectiveness (β = −.04, p = .59), or the partner's conversational skill (β = .11, p = .25). That is, high CU participants' perceptions did not improve when interacting with another high CU participant, suggesting that high CU individuals may not prefer interacting with other high CU individuals. Although these effects should be replicated before drawing conclusions from such null results, Tobin and Osika (2011) also found that people reported liking high CU others less than low CU others, regardless of their own CU level.

These data demonstrate that high CU participants perceive their conversations as less appropriate and effective, replicating previous uncertainty research (e.g., Gudykunst & Nishida, 2001; Neuliep & Grohskpof, 2000). According to Uncertainty Reduction Theory, high levels of uncertainty also should lead to lower levels of intimacy and liking (Berger & Calabrese, 1975), which has been corroborated in several studies (Clatterbuck, 1979; Douglas, 1990; Douglas, 1994; Kellerman & Reynolds, 1990; also see Lee, 2001, for an uncertainty reduction explanation for mere exposure effects). Although these findings suggest that causally uncertain people's negative perceptions may extend to liking, Tobin and Osika (2011) found that the only situation in which high CU participants liked others less than low CU participants was when targets were extremely low CU. However, Tobin and Osika examined liking for hypothetical targets after participants viewed the target's completed CUS; thus, we might observe different effects with real conversational partners, when a partner's CU is not as clear. In fact, in the current study, causally uncertain participants rated their partners as less conversationally skilled, which could result in reduced liking.

In addition, although we examined participants' perceptions across a series of conversations, all conversations were with the same conversational partner. Thus, whether these negative perceptions extend beyond an initial interaction with one partner to subsequent interactions with new partners remains unclear. Conceivably, an initial meeting with a stranger may be sufficient to activate participants' CU beliefs, but when given the opportunity to engage in similar interactions with new partners, the practice from the first interaction could reduce their behavioral uncertainty (i.e., uncertainty about how to act or react) and consequently lead to more positive perceptions of these subsequent interactions (and conversational partners). We explored this idea in Study 2.

STUDY 2

  1. Top of page
  2. Abstract
  3. STUDY 1
  4. STUDY 2
  5. STUDY 3
  6. GENERAL DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES

The purpose of Study 2 was to further investigate causally uncertain people's negative social perceptions to determine if they would report liking conversational partners less following a brief initial interaction. In addition, we examined if these negative perceptions would improve in subsequent interactions with new partners when behavioral uncertainty should be lower.

We assigned participants to four-person groups, and using a round-robin design, participants interacted with each group member for five minutes. Given the lack of significant interactions between CU and conversation type in Study 1, participants engaged in just one conversation with each group member. Thus, each participant engaged in three separate unstructured dyadic conversations, and following each conversation, participants rated how much they liked their partner.

One advantage of using this design is that it yields information about perceiver (i.e., how a participant tends to rate all of his or her conversational partners), target (i.e., how a participants tends to be viewed by all conversational partners), and relationship effects (i.e., unique, dyadic effect measuring how a participant views a particular partner; see Kenny, 1994). By assessing these different effects, our aim was to demonstrate that causally uncertain people rate all of their conversational partners negatively, not just the first.

Engaging in one initial interaction should provide participants with a behavioral script for interactions of this nature, making them more confident about how to act or react during similar interactions with new partners, thereby lowering behavioral uncertainty. However, cognitive uncertainty (i.e., uncertainty about one's ability to predict others' attitudes, beliefs, and feelings) should remain high across all three conversations and, consequently, should reduce liking and intimacy between conversational partners (Berger & Calabrese, 1975). Thus, we expected a significant perceiver effect such that high CU participants would report liking all three partners less than low CU participants would. Although we made no specific predictions for target effects given the lack of significant partner effects in Study 1, Tobin and Osika (2011) found that people reported liking a hypothetical target who has high CU less than one who has low CU, suggesting that participants may report liking high CU partners less than low CU partners.

Finally, we also wanted to explore how CU might be related to other personality variables that might influence people's social perceptions. Specifically, we included a measure of social anxiety because although uncertainty about how to make a desired impression is considered to increase social anxiety (see Schlenker & Leary, 1982), no studies have reported the relationship between these two constructs. We predicted that CU and social anxiety would be positively correlated, but that social anxiety would not account for the hypothesized CU effects.

Method

Participants

Participants were 126 female undergraduate students recruited from Queen's University in Canada and assigned to four-person groups.3 Participants had a mean age of 18.53 years (SD = 0.89), ranging from 17 to 23. Participants received partial course credit or $CAN15.00 for their participation.

Measures

As in Study 1, participants completed the CUS (M = 42.14, SD = 0.75, ranging from 16 to 84) and the BDI-II (M = 7.78, SD = 7.73, ranging from 0 to 48). Although CUS scores were slightly higher in Study 2, the mean was still consistent with other CU research (e.g., Edwards & Weary, 1998).

To assess social anxiety, participants completed the Interaction Anxiousness Scale (IAS; Leary, 1983), which is a self-report measure assessing the degree to which participants experience subjective feelings of social anxiety. It consists of 15 items (e.g., “I often feel nervous even in casual get-togethers”) rated on a 5-point scale from 1 (not at all) to 5 (extremely characteristic of me). Responses are summed so that higher scores indicate greater social anxiety. The IAS had excellent internal consistency in the current sample (α = .91), and participants had a mean score of 45.94 (SD = 12.05), ranging from 17 to 73; these scores are consistent with previous studies using this scale with college samples (see Leary & Kowalski, 1993).

Finally, to assess liking, participants completed McCroskey and McCain's (1974) Social Attraction Subscale (SAS). The SAS is a measure of liking among strangers consisting of five items (e.g., “I think s/he could be a friend of mine”) rated on a 7-point scale from 1 (strongly disagree) to 7 (strongly agree). Responses are summed so that higher scores indicate greater liking. In the current study, the SAS had good internal consistency for all three conversations (α = .81 to .90).

Procedure

Following Kenny's (1994) Social Relations Model (SRM), participants were assigned to a group with three unacquainted women and randomly assigned to the role of “A”, “B”, “C”, or “D.” For logistical reasons, participants in Study 2 interacted with each other via web-based cameras and microphones using MSN Messenger©; however, we re-created a face-to-face conversation by maintaining the vocal and visual cues present in these interactions. Upon arriving, participants were led to individual computer stations linked to a computer network; experimenters monitored conversations from another computer that controlled all participant computers.

The experimenter first briefly demonstrated the use of the computer interface and web-based cameras to participants. Then, participants engaged in a five-minute unstructured interaction with another group member. The instructions for this task were the same as those used for the first conversation in Study 1. After five minutes, the experimenter interrupted the interaction and instructed participants to complete the SAS. This process was repeated for the second and third conversations, so that participants interacted with each group member.

After participants finished all three conversations, they completed the CUS, BDI-II, and IAS. Finally, participants were fully debriefed.

Social Relations Model Analyses

This analysis calculates a perceiver effect, estimating every participant's tendency to view the other group members; a target effect, estimating how consistent other group members were in rating a particular participant, and a relationship effect, estimating a particular dyad's tendency to rate each other (see Kenny, Kashy, & Cook, 2006, for a description of this model). Thus, the perceiver and target effects in this model are analogous to the actor and partner effects described in Study 1.

These analyses were conducted using structural equation modeling (SEM) in Amos 17.0 (Amos Development Corporation, Meadville, PA, USA), and the group was treated as the unit of analysis. Four latent variables represented each participant's perceiver effect, and four latent variables represented the target effects. Therefore, A's ratings of B, C, and D were used to estimate the perceiver effect for A, whereas B's, C's, and D's ratings of A were used to estimate the target effect for A. The relationship effect was represented by the error terms associated with each rating. Corresponding perceiver and target effects were correlated (i.e., A's perceiver and target effects), as were error terms for corresponding ratings (i.e., error terms associated with A's rating for B and vice versa).

To determine if CU was associated with less liking overall (actor effect), we estimated the direct effect of each participant's CU on the corresponding perceiver effect. Similarly, to determine if participants' CU influenced if their partners liked them (partner effect), we also estimated the direct effect of each participant's CU on the corresponding target effect.

Because group members were interchangeable, we imposed a series of equality constraints as recommended by Kenny and West (2006). Because the labels assigned to participants were arbitrary (i.e., the A, B, C, and D labels were not meaningful), we did not expect systematic differences across these roles in terms of liking. Thus, we imposed equality constraints on the four perceiver variances (A, B, C, and D), the four target variances, the 12 relationship variances, the four perceiver–target covariances, and the six error covariances. We also did not expect systematic differences across the participant labels in terms of CU level, so we placed additional equality constraints on the four CUS means and variances, the four CUS and perceiver effect covariances, and the four CUS and target effect covariances (see Kenny & West, 2006, for more information on conducting SRM analyses with interchangeable group members).

However, to determine whether liking changed across conversations, we placed equality constraints across the four levels of participant for the same conversation instead of constraining all 12 liking scores to be equal. Specifically, we constrained the four liking ratings for conversation 1 to be the same, the four liking ratings for conversation 2 to be the same, and the four liking ratings for conversation 3 to be the same. Consequently, the participants' individual means could differ across the three conversations, allowing us to test for CU effects on liking across conversations, but not across the four participant roles as there was no reason to expect systematic differences across these arbitrary roles. Similarly, relationship variances and error covariances were constrained within conversations but could differ across conversations. To determine if these ratings should be treated as static, we tested an alternative nested model in which paths from the observed variables to the latent variables representing the actor and partner effects were set to one.

Results and Discussion

As predicted, CU was positively and significantly correlated with social anxiety, r(126) = .28, p < .001. Thus, consistent with the Anxiety/Uncertainty Management Theory (Gudykunst, 1993, 1995), people with greater CU also experienced higher levels of social anxiety. Given this significant relationship, all SRM analyses reported below were replicated controlling for social anxiety as well as depression (depression-CU: r(126) = .30, p < .001); all reported effects were significant or marginal when controlling for these variables.

Liking

Model comparison for the liking model indicated that constraining the paths from the observed variables to the latent variables to be equal to one did not significantly affect model fit (p = .06).4 In other words, participants' liking ratings did not differ substantially from one conversation to the other. Therefore, the estimates reported below are from this alternative model rather than the full SRM model. As predicted, the covariance between CU and the perceiver effect was significant, cov = −5.75 (SE = 2.85), p = .04, indicating that high CU participants reported liking all three of their conversational partners less than did low CU participants. In contrast, the covariance between CU and the target effect was not significant, cov = 0.98 (SE = 2.62), p = .71, indicating that participants' CU did not determine how much their partner liked them.

These data demonstrate that the negative social perceptions associated with CU we observed in Study 1 do extend beyond perceptions of the conversation and their conversational partner's competence to actual (dis)liking for their partner. Consistent with the Uncertainty Reduction Theory (Berger & Calabrese, 1975) and other studies of uncertainty and liking (e.g., Clatterbuck, 1979), causally uncertain participants reported liking their conversational partners less following a brief unstructured interaction. Furthermore, as in Study 1, people interacting with causally uncertain partners did not appear to share these negative perceptions. Thus, contrary to Tobin and Osika's (2011) findings with hypothetical targets, participants did not report liking high CU partners less than low CU partners.

Importantly, the current data also demonstrate that these negative social perceptions extend beyond a single initial interaction. Even though the practice from having had a conversation with a stranger already should ease some of their confusion about how to act in similar interactions with new partners, high CU participants seemed to perceive these later partners as negatively as their first conversational partner. Hence, behavioral uncertainty may not be responsible for causally uncertain people's negative social perceptions.

Indeed, Berger and Gudykunst (1991) suggest that individuals can conduct successful interactions with strangers despite high levels of behavioral uncertainty because they can carry out these interactions using behavioral scripts. Instead, reducing cognitive uncertainty (i.e., uncertainty about one's ability to predict others' thoughts, feelings, and attitudes) may be more important in these contexts. To test this idea, we examined the role of behavioral and cognitive uncertainty in the CU–liking relationship in Study 3.

STUDY 3

  1. Top of page
  2. Abstract
  3. STUDY 1
  4. STUDY 2
  5. STUDY 3
  6. GENERAL DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES

The purpose of Study 3 was to determine if behavioral and/or cognitive uncertainty mediate the CU–liking relationship observed in Study 2. As in Study 1, participants were paired into unacquainted, same-sex dyads and engaged in a brief unstructured interaction. Following this interaction, participants rated how much they liked their partner as well as their level of behavioral and cognitive uncertainty.

Higher levels of uncertainty have been linked to less liking and intimacy (e.g., Clatterbuck, 1979), which we replicated with CU in Study 2. Therefore, we predicted that CU also would be negatively related to liking in Study 3. Furthermore, given their persistent doubts about their ability to understand social dynamics, causally uncertain people should feel less confident in predicting and explaining their partner's behavior, attitudes, and feelings. Consequently, high CU people should maintain relatively high levels of behavioral and cognitive uncertainty. These heightened levels of uncertainty should, in turn, result in less intimacy and liking (Berger & Calabrese, 1975). Therefore, we also predicted that, relative to low CU participants, high CU participants would report higher levels of behavioral and cognitive uncertainty and that this uncertainty would mediate the CU–liking relationship.

Finally, although we demonstrated in earlier studies that our CU effects could not be attributed to depression or social anxiety, these effects could be the result of other individual differences in uncertainty, such as global uncertainty (i.e., confusion about general acquaintanceship processes), which also has been linked to greater cognitive and behavioral uncertainty following initial interactions and to less liking for conversational partners (Douglas, 1994). Therefore, we examined the relationship between CU and global uncertainty and predicted that CU would be positively related to global uncertainty, although global uncertainty would not account for any of the hypothesized CU effects.

Method

Participants

Participants were 114 undergraduate students (88 women, 26 men) recruited from introductory psychology courses at Queen's University in Canada and paired to form 57 same-sex dyads. Participants had a mean age of 18.87 years (SD = 3.59), ranging from 16 to 44. Participants received partial course credit or $CAN10.00 for their participation.

Measures

As in previous studies, participants completed the CUS (α = .89; M = 37.44, SD = 10.94, ranging from 14 to 65), BDI-II (α = .89; M = 6.67, SD = 6.00, ranging from 0 to 26), and IAS (α = .91; M = 43.42, SD = 12.00, ranging from 18 to 69). Again, these scores were consistent with previous studies using these scales with college samples.

To assess global uncertainty, participants completed the Global Uncertainty Scale (GUS; Douglas, 1991) as part of in-class prescreening sessions; however, because prescreening was not mandatory, 10 participants did not complete the GUS. The GUS consists of seven items assessing uncertainty about strangers during initial interactions (e.g., “How confident are you of your general ability to predict how strangers will behave?”). In this study, the items were rated on a 6-point scale from 1 (not at all confident) to 6 (extremely confident). All ratings were reverse-coded so that higher scores indicated greater uncertainty and then averaged. The GUS achieved adequate internal consistency in the current sample (α = .78), and participants had an average GUS score of 3.48 (SD = 0.75), ranging from 1.86 to 5.29, which was consistent with previous research (Douglas, 1991, 1994).

Participants also completed the SAS to measure the extent to which they liked their partner (α = .77), as well as four uncertainty reduction questionnaires (Douglas, 1990) to assess their uncertainty about (i) their own future behavior (UOB; e.g., “If I meet my partner again, I will know what to say”), (ii) their partner's future behavior (UPB; e.g., “If I meet my partner again, I know what s/he will say”), (iii) their feelings toward their partner (UOF; e.g., “I know what my attitude toward my partner is”), and (4) their partner's feelings toward them (UPF; e.g., “I know what my partner's attitude toward me is”). Thus, two questionnaires assessed participants' behavioral uncertainty (UOB and UPB), and two assessed their cognitive uncertainty (UOF and UPF). Each item was rated on a 9-point scale from 1 (not at all confident) to 9 (extremely confident), and responses were summed so that higher scores indicated less uncertainty.

Procedure

Upon arriving, participants were seated across each other while the experimenter monitored the interaction from an adjacent room and provided instructions using a one-way intercom. To avoid unusual behavior that could arise when people have no expectations for future interaction (see Berger, 1979), dyads were informed that they would engage in three conversations but actually engaged in only one. Dyads engaged in a five-minute unstructured conversation (instructions for this task were identical to those used in Studies 1 and 2). Following this conversation, participants completed the SAS, uncertainty reduction questionnaires, CUS, BDI-II, and IAS at isolated computer stations. Finally, participants were fully debriefed.

Results and Discussion

As predicted, CU was significantly and positively correlated with global uncertainty, r(104) = .24, p < .05, suggesting that high CU people also tend to experience greater uncertainty about general acquaintanceship processes. Furthermore, consistent with Study 2, CU was significantly and positively correlated with social anxiety, r(114) = .26, p < .001, and depression, r(114) = .30, p = .001. All of the analyses reported below also were conducted controlling for participants' GUS, IAS, and BDI-II scores.

Dyadic Analyses

As in Study 1, the data were analyzed using the APIM, including participants' continuous CUS scores, partners' continuous CUS scores, and the corresponding two-way interaction as predictors in the model. Consistent with Study 2, high CU participants reported liking their partners less compared with low CU participants, β = −.21, t(99.93) = −2.18, p = .03. Also replicating the effects from Study 2, the corresponding partner effect, β = −.002, t(99.25) = 0.03, p = .98, and the two-way participant × partner CU interaction, β = −.06, t(52.32) = −0.45, p = .66, were not significant.

High CU participants also reported higher levels of uncertainty about their feelings toward their partner (UOF), β = −0.21, t(102.30) = −2.25, p = .03, as well as their partner's feelings toward them (UPF), β = −.33, t(100.24) = −3.69, p < .001. However, CU was not significantly related to participants' uncertainty about their own future behavior (UOB), β = −.13, t(100.23) = −1.43, p = .16, but was related to uncertainty about their partner's future behavior (UPB), β = −.18, t(96.45) = −1.98, p = .05 (although the latter was not significant when controlling for participants' GUS, IAS, or BDI-II scores). Therefore, as suggested in Study 2, CU appears to be more strongly related to cognitive uncertainty than behavioral uncertainty.5 Again, none of the corresponding partner effects (β = −.09 to .02; p > .31) and two-way interactions (β = −.11 to .07; p > .35) were significant.

Mediation Analyses

The purpose of Study 3 was to determine if a lack of uncertainty reduction during initial interactions could explain the CU–liking relationship. The previous analyses tested Step 1 (i.e., establishing a significant relationship between the predictor and outcome variables) and Step 2 (establishing a significant effect between the predictor and mediator variables) of Baron and Kenny's (1986) mediation testing procedures. Therefore, we conducted additional analyses to test Step 3 (i.e., demonstrating that the relationship between the predictor and outcome variables is not significant when including the mediator, whereas the relationship between the mediator and outcome variables is significant) for the uncertainty constructs that were significantly related to CU (i.e., UOF and UPF scores).

We conducted these mediation analyses using Amos 17.0 to estimate the APIM effects. To examine the actor effect, we estimated the direct effect of each dyad member's CU on their own SAS scores. To examine the partner effect, we estimated the direct effect for each dyad member's CU on his or her partner's SAS scores. Because we found no interactions between participant and partner CU in earlier analyses, we did not include these interactions in these models. To test for mediation, we included either participants' UOF or UPF scores (i.e., we conducted separate analyses for each potential mediator) as additional predictors in this model. That is, we tested the mediating path from participants' CUS scores to their SAS scores via each of the uncertainty reduction scores to determine if uncertainty reduction played a role in the CU–liking relationship.

Although separate effects were calculated for each dyad member, members actually were interchangeable. Therefore, we imposed a series of equality constraints on the actor and partner effects, the predictor means and variances, outcome intercepts, and residual variances (see Olsen & Kenny, 2006, for a description of APIM in SEM with interchangeable dyads).

As predicted, participants' uncertainty about their feelings toward their partner as well as their uncertainty about their partner's feelings toward them mediated the CU–liking relationship. Specifically, the actor effect for CU on liking was no longer significant once participants' UOF, β = −.05, p = .14 (bootstrap CI: −.24 to −.001), or UPF, β = −.03, p = .31 (bootstrap CI: −.23 to .04), scores were included in the model. However, both UOF and UPF scores were significantly related to liking in these models, β = −.09 (bootstrap CI: .29 to .57), p = .03, and β = −.14 (bootstrap CI: .26 to .57), p < .001, respectively. (See Figures 1 and 2 for depictions of these mediation models including all standardized parameter estimates.) Thus, one viable explanation for causally uncertain people's negative social perceptions is that they are unable to reduce their cognitive uncertainty sufficiently during brief initial interactions.

image

Figure 1. Participant and partner causal uncertainty (CU) effects on liking (Social Attraction Subscale (SAS)) as mediated by uncertainty about participants' feelings toward their partner (UOF)

Download figure to PowerPoint

image

Figure 2. Participant and partner causal uncertainty (CU) effects on liking (Social Attraction Subscale (SAS)) as mediated by participants' uncertainty about partner's feelings toward them (UPF)

Download figure to PowerPoint

One limitation of Study 3 is that we did not measure uncertainty reduction directly. Rather, we measured uncertainty only after the interaction, so we could not calculate a difference score assessing actual uncertainty reduction. Therefore, high CU participants may have reduced their uncertainty while maintaining higher levels of cognitive uncertainty relative to low CU participants. Indeed, Douglas (1994) found that people with higher levels of global uncertainty actually experienced more uncertainty reduction during initial interactions but that their levels of uncertainty remained higher relative to participants with less global uncertainty. Although future research should measure uncertainty reduction more directly to determine the extent to which high CU people experience any uncertainty reduction, the level of uncertainty remaining after their conversation ended is more important. That is, even if high CU people are able to reduce their uncertainty during the conversation, they should still experience more negative interpersonal consequences if they still maintain higher levels of uncertainty than do low CU people.

GENERAL DISCUSSION

  1. Top of page
  2. Abstract
  3. STUDY 1
  4. STUDY 2
  5. STUDY 3
  6. GENERAL DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES

Causally uncertain individuals experience lingering doubts about their ability to determine the causes of social events (Weary & Edwards, 1994). When their doubts are activated, these individuals tend to adopt a subjective accuracy goal that prompts a deliberate search for and effortful processing of social information (Weary & Edwards, 1996). Although such effortful processing should lead to greater uncertainty reduction and, consequently, more positive and effective social interactions (Berger & Calabrese, 1975; Gudykunst, 1993, 1995), the current research demonstrates that high CU people perceive their conversations and conversational partners more negatively than do low CU people.

Consistent with Berger and Calabrese's (1975) Uncertainty Reduction Theory and Gudykunst's (1993, 1995) Anxiety/Uncertainty Management Theory, these negative perceptions appear to stem, at least in part, from causally uncertain people's difficulties with uncertainty reduction in the brief unstructured interactions we observed. More specifically, in Study 3, causally uncertain participants reported more uncertainty about how they felt about their partner and about how their partner felt about them. Replicating other findings in the uncertainty reduction literature (e.g., Clatterbuck, 1979), these heightened levels of cognitive uncertainty led to less liking following a brief conversation. Importantly, these effects could not be attributed to similar forms of uncertainty, such as global uncertainty, or to other individual difference variables associated with problematic interactions, such as social anxiety (Christensen, Stein, & Means-Christensen, 2003) and depression (Segrin & Abramson, 1994).

Causal Uncertainty and Uncertainty Reduction

Previous research demonstrates that causally uncertain people engage in more information seeking behaviors (Weary & Jacobson, 1997), presumably leading to greater uncertainty reduction. So why do causally uncertain people maintain higher levels of cognitive uncertainty? One plausible explanation is that uncertainty reduction further requires them to use the information they acquire to predict and explain their partner's attitudes and feelings, and their lack of confidence in tasks requiring causal analysis may make them more reluctant to make these social judgments (see Edwards, 1998).

Alternatively, causally uncertain people may not engage in greater information seeking during actual social exchanges, or they may engage in inefficient information seeking strategies. For example, Douglas (1991) found that although globally uncertain individuals were more likely to pursue information seeking during initial interactions, they were more likely to run out of things to say and used less efficient information seeking strategies (i.e., greater disclosure but fewer questions).

Currently, all of the research demonstrating that causally uncertain people engage in more effortful processing has been limited to the realm of social cognition (e.g., Jacobson et al., 2008; Tobin & Weary, 2008; Weary et al., 2001; Weary et al., 2006). In these studies, participants are typically in an isolated room and instructed to process information that is presented to them on a computer screen. This laboratory context may be particularly conducive to processing information thoroughly because distractions are few and participants do not have to seek out the information themselves but merely have to process what has already been provided to them.

In contrast, face-to-face communication has many elements that would make it more complicated to process information thoroughly than at an isolated computer terminal (see Lane, 2008). Specifically, people must simultaneously encode and decode messages, which can vary in terms of complexity. People also must attend to both verbal and nonverbal behaviors for feedback from conversational partners and to monitor their own behavior throughout the conversation. To make matters worse, external noise (e.g., loud sounds) or internal noise (e.g., fatigue or unrelated preoccupations) can interfere with the communication process. Given their confusion about social dynamics, high CU people likely experience more cognitive load in this environment than do low CU people, which should interfere with their effortful processing (see Weary et al., 2006).

Weary and Edwards (1996) proposed that when the expectancy for success in pursuing an accuracy goal (and thereby reducing the discrepancy between one's current and desired knowledge states) is low, causally uncertain people are unlikely to engage in effortful processing and should disengage from this goal instead. Such disengagement can occur, for example, by withdrawing from the situation that activated their CU beliefs, by focusing their attention on another goal or by re-evaluating the importance of their goal to reduce uncertainty. Similarly, Gudykunst (1993) argues that when people surpass a maximum uncertainty threshold, they have no confidence in their ability to predict or explain other people's behaviors, attitudes, or feelings. Consequently, they are not motivated to reduce their uncertainty and disengage from the conversation. Therefore, given the additional demands of negotiating social exchanges, causally uncertain people may abandon their accuracy goal and find other ways to cope with their uncertainty.Future research should explore possible explanations for causally uncertain people's difficulties with uncertainty reduction. Specifically, we should aim to determine if causally uncertain people maintain higher levels of cognitive uncertainty because they were unable to use the information they acquired about others to predict and explain people's attitudes and feelings or if they simply disengaged from their uncertainty reduction attempts. Gaining a better understanding of why causally uncertain people maintain their cognitive uncertainty will be particularly important in identifying ways to improve their uncertainty reduction during social interactions and, in turn, their perceptions of these exchanges.

Temporary Versus Lasting Relationships

Although we examined CU effects during initial interactions, our findings suggest that causally uncertain people may experience difficulties in close relationships. Berger and Gudykunst (1991) argue that reducing cognitive uncertainty becomes more important as relationships progress because people must acquire individuating information about each other to make accurate interpersonal judgments and successfully negotiate the more intimate encounters.

Furthermore, causally uncertain people may experience other forms of uncertainty in these closer relationships. Knobloch and Solomon (1999) argue that people can experience relational uncertainty within intimate relationships. Presumably, if causally uncertain people maintain higher levels of uncertainty in their relationships with strangers, they also might experience more relational uncertainty in their intimate relationships, which could have detrimental effects. For example, heightened levels of relational uncertainty can make conversations between romantic partners more difficult by interfering with message processing (Knobloch & Solomon, 2005) or by increasing avoidant or indirect communication (see Knobloch, 2007).

Therefore, future research should explore the impact of CU within close relationships. Specifically, we should determine if causally uncertain people maintain higher levels of uncertainty even in these more intimate relationships and how CU affects such relationships over time.

Conclusions

In summary, we demonstrated across three studies that high CU people perceive their social interactions and interaction partners more negatively than do low CU people. Interestingly, unlike matching effects observed with variables associated with CU, such as depression (Rosenblatt & Greenberg, 1991) and social anxiety (Kashdan & Wenzel, 2005), these negative perceptions did not improve when high CU people interacted with another high CU partner. Furthermore, although these effects need to be replicated, people interacting with high CU partners did not appear to share these negative social perceptions, suggesting that the negative perceptions may not reflect truly problematic conversations.

Rather, causally uncertain people appear to possess negative social perceptions because of their inability to sufficiently reduce their cognitive uncertainty during brief initial interactions. Although future research is required to further explain the relationship between CU and uncertainty reduction, the current set of studies expands our understanding of CU beyond its social cognitive effects on information seeking and processing to actual, real-time social interactions and to perceptions of real conversational partners. We believe this area of research is important, particularly given the social basis of CU, and warrants further attention by researchers, including explorations of how CU affects other types of social exchanges.

ACKNOWLEDGEMENTS

  1. Top of page
  2. Abstract
  3. STUDY 1
  4. STUDY 2
  5. STUDY 3
  6. GENERAL DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES

This manuscript is based in part on a master's thesis and dissertation by the first author, and some of the data were presented at annual meetings of the Society for Personality and Social Psychology. This research was supported in part by a Queen's University Advisory Research Committee grant and a Canadian Foundation for Innovation grant awarded to Jill A. Jacobson. We would like to thank Shannon Currie, Kyla Bondy, Chiara Papile, Andrew Gris, Alex Wilson, Jennifer Passey, Carly Ainlay, and Kimberly Thompson for their assistance with the data collection. We also would like to thank Dr. Jennifer Rockett and Dr. Jorden Cummings for their comments on earlier drafts of this manuscript.

  • 1

    Although CU and depression generally are found to be positively correlated (Weary & Edwards, 1994, 1996), CUS scores were not significantly correlated with BDI-II scores in Study 1 (r = .16, p = .07).

  • 2

    Considering the nature of the two problem interactions, we expected conversation type × order and problem type × order interactions. However, these interactions are not of interest here because the roles resulting from the order of problem presentation were experimentally induced, and thus, the nature of these interactions is somewhat artificial. Therefore, any effects of problem presentation order, conversation type, or problem type that did not interact with CU are not relevant to the central purpose of this study and are not discussed.

  • 3

    Only female participants were recruited because recruiting four same-sex participants was more feasible with women given the greater number of women enrolled in psychology classes at Queen's University. However, no gender differences have been observed in previous CU research, so we had no reason to expect that limiting our sample to women would affect the relationship between CU and liking.

  • 4

    Nine other alternative models also were tested based on Kenny's (personal communication, April 18, 2007) recommendations. In the first of these, paths from the observed to the latent variables representing the actor and partner effects were set to zero. In the second, the group variance was set to zero. In the third, the relationship and the actor–partner covariances were set to zero. The six remaining models were combinations of the other four models. However, using any of these other models would have reduced fit for the liking model (p < .04).

  • 5

    We also conducted these analyses on a total uncertainty reduction score because the four scores were positively correlated (r > .58, p < .001). High CU participants reported higher levels of overall uncertainty than did low CU participants, β = −.26, t(97.49) = −2.80, p = .006. Neither the partner effect, β = −.03, t(96.69) = −0.36, p = .72, nor the two-way interaction, t(52.50) = −0.31, p = .76, were significant.

REFERENCES

  1. Top of page
  2. Abstract
  3. STUDY 1
  4. STUDY 2
  5. STUDY 3
  6. GENERAL DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. REFERENCES
  • Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 11731182.
  • Beck, A. T. (1996). Beck Depression Inventory II. San Antonio, TX: Psychological Corporation.
  • Berger, C. R. (1979). Beyond initial interaction: Uncertainty, understanding, and the development of interpersonal relationships. In H. Giles, & R. St. Clair (Eds.), Language and social psychology (pp. 122144). Oxford: Basil Blackwell.
  • Berger, C. R., & Calabrese, R. J. (1975). Some explorations in initial interaction and beyond: Toward a developmental theory of interpersonal communication. Human Communication Research, 1, 99112.
  • Berger, C. R., & Gudykunst, W. B. (1991). Uncertainty and communication. In B. Dervin, & M. J. Voigt (Eds.), Progress in communication sciences (vol. 10, pp. 2166). Norwood, NJ: Ablex Publishing.
  • Boucher, E. M., & Jacobson, J. A. (2005). Shyness isn't everything: Causal uncertainty as a predictor of computer-mediated communication use. Poster presented at the Conference on Group Processes in Computer-Supported Interaction: Technological and Social Determinism, Oxford, OH.
  • Brashers, D. E. (2001). Communication and uncertainty management. Journal of Communication, 51, 477497.
  • Christensen, P. N., Stein, M. B., & Means-Christensen, A. (2003). Social anxiety and interpersonal perception: A social relations model analysis. Behaviour Research and Therapy, 41, 13551371.
  • Clatterbuck, G. W. (1979). Attributional confidence and uncertainty in initial interaction. Human Communication Research, 5, 147157.
  • Douglas, W. (1990). Uncertainty, information seeking, and liking during initial interactions. Western Journal of Speech Communication, 54, 6681.
  • Douglas, W. (1991). Expectations about initial interaction: An examination of the effects of global uncertainty. Human Communication Research, 17, 355384.
  • Douglas, W. (1994). The acquaintanceship process: An examination of uncertainty, information-seeking, and social attraction during initial conversation. Communication Research, 21, 154176.
  • Edwards, J. A. (1998). Effects of causal uncertainty on the dispositional attribution process. Journal of Experimental Social Psychology, 34, 109135.
  • Edwards, J. A., & Weary, G. (1998). Antecedents of causal uncertainty and perceived control: A prospective study. European Journal of Personality, 12, 135148.
  • Edwards, J. A., Wichman, A. L., & Weary, G. (2008). Causal uncertainty as a chronically accessible construct. European Journal of Social Psychology, 39, 694706.
  • Fast, L. A., Reimer, H. M., & Funder, D. C. (2008). The social behavior and reputation of the attributionally complex. Journal of Research in Personality, 42, 208222.
  • Fletcher, G. J. O., Danilovics, P., Fernandez, G., Peterson, D., & Reefer, G. D. (1986). Attributional complexity: An individual difference measure. Journal of Personality and Social Psychology, 51, 875884.
  • Gudykunst, W. B. (1993). Toward a theory of effective interpersonal and intergroup communication: An anxiety/uncertainty management perspective. In R. L. Wiseman, & J. Koester (Eds.), Intercultural communication competence (pp. 3371). Newbury Park, CA: Sage.
  • Gudykunst, W. B. (1995). Anxiety/uncertainty management theory: Current status. In R. L. Wiseman (Ed.), Intercultural communication theory (pp. 858). Thousand Oaks, CA: Sage Publications.
  • Gudykunst, W. B., & Nishida, T. (2001). Anxiety, uncertainty, and perceived effectiveness of communication across relationships and cultures. International Journal of Intercultural Relations, 25, 5571.
  • Heath, R. L., & Bryant, J. (2000). Human communication theory and research: Concepts, contexts, and challenges (2nd Edition). Mahwah, NJ: Lawrence Erlbaum.
  • Jacobson, J. A., Weary, G., & Chakraborti, I. (1997). Uncertainty about the causes of one's own outcomes: Implications for loneliness and shyness. Poster presented at the annual meeting of the Midwestern Psychological Association, Chicago, IL.
  • Jacobson, J. A., Weary, G., & Lin, Y. S. (2008). Causal uncertainty and metacognitive inferences about goal attainment. Cognition & Emotion, 22, 12761305.
  • Kashdan, T. B., & Wenzel, A. (2005). A transactional approach to social anxiety and the genesis of interpersonal closeness: Self, partner, and social context. Behavior Therapy, 36, 335346.
  • Kashy, D. A., & Kenny, D. A. (2000). The analysis of data from dyads and groups. In H. T. Reis, & C. M. Judd (Eds.), Handbook of research methods in social and personality psychology (pp. 451477). New York: Cambridge University Press.
  • Kellerman, K., & Reynolds, R. (1990). When ignorance is bliss: The role of motivation to reduce uncertainty in uncertainty reduction theory. Human Communication Research, 17, 575.
  • Kenny, D. A. (1994). Interpersonal perception: A social relations analysis. New York: Guilford Press.
  • Kenny, D. A., & West, T. (2006). Estimation of the SRM not using specialized software. Unpublished manuscript, University of Connecticut.
  • Kenny, D. A., Kashy, D. A., & Cook, W. L. (2006). Dyadic data analysis. New York: Guilford Press.
  • Knobloch, L. K. (2007). The dark side of relational uncertainty: Obstacle or opportunity. In B. H. Spitzberg, & W. R. Cupach (Eds.), The dark side of interpersonal communication, 2nd edition (pp. 3160). Mahwah, NJ: Lawrence Erlbaum.
  • Knobloch, L. K., & Solomon, D. H. (1999). Measuring the sources and content of relational uncertainty. Communication Studies, 50, 261278.
  • Knobloch, L. K., & Solomon, D. H. (2005). Relational uncertainty and relational information processing: Questions without answers? Communication Research, 32, 349388.
  • Lane, S. D. (2008). Interpersonal communication: Competence and contexts. Boston, MA: Pearson.
  • Leary, M. R. (1983). Social anxiousness: The construct and its measurement. Journal of Personality Assessment, 47, 6675.
  • Leary, M. R., & Kowalski, R. M. (1993). The interaction anxiousness scale: Construct and criterion-related validity. Journal of Personality Assessment, 61, 136146.
  • Lee, A. Y. (2001). The mere exposure effect: An uncertainty reduction explanation revisited. Personality and Social Psychology Bulletin, 27, 12551266.
  • McCroskey, J. C., & McCain, T. A. (1974). The measurement of interpersonal attraction. Speech Monographs, 41, 261266.
  • Neuliep, J. W., & Grohskpof, E. L. (2000). Uncertainty reduction and satisfaction during initial interaction: An initial test and replication of a new axiom. Communication Reports, 13, 6777.
  • Olsen, J. A., & Kenny, D. A. (2006). Structural equation modeling with interchangeable dyads. Psychological Methods, 11, 127141.
  • Rook, K. S., Pietromonaco, P. R., & Lewis, M. A. (1994). When are dysphoric individuals distressing to others and vice versa? Effects of friendship, similarity, and interaction task. Journal of Personality and Social Psychology, 67, 548559.
  • Rosenblatt, A., & Greenberg, J. (1991). Examining the world of the depressed: Do depressed people prefer others who are depressed? Journal of Personality and Social Psychology, 60, 620629.
  • Schlenker, B. R., & Leary, M. R. (1982). Social anxiety and self-presentation: A conceptualization model. Psychological Bulletin, 92, 641669.
  • Segrin, C., & Abramson, L. Y. (1994). Negative reactions to depressive behaviors: A communication theories analysis. Journal of Abnormal Psychology, 104, 655668.
  • Spitzberg, B. H. (1995). Instructional assessment of interpersonal competence: The conversational skills rating scale. Annandale, VA: Speech Communication Association.
  • Spitzberg, B. H., & Canary, D. J. (1985). Loneliness and relationally competent communication. Journal of Social and Personal Relationships, 2, 387402.
  • Tobin, S. J. (2010). Causal uncertainty and attributional complexity. Unpublished data, University of Houston.
  • Tobin, S. J., & Osika, M. M. (2011). Interpersonal attraction as a function of self and other causal uncertainty. Manuscript under review.
  • Tobin, S. J., & Weary, G. (2008). The effects of causal uncertainty, causal importance, and initial attitude on attention to causal persuasive arguments. Social Cognition, 26, 4465.
  • Weary, G., & Edwards, J. A. (1994). Individual differences in causal uncertainty. Journal of Personality and Social Psychology, 67, 308318.
  • Weary, G., & Edwards, J. A. (1996). Causal uncertainty beliefs and related goal structures. In R. Sorrentino, & E. T. Higgins (Eds.), Handbook of motivation and cognition: The interpersonal context, vol. 3 (pp. 148181). New York: Guilford Press.
  • Weary, G., & Jacobson, J. A. (1997). Causal uncertainty beliefs and diagnostic information seeking. Journal of Personality and Social Psychology, 73, 839848.
  • Weary, G., Jacobson, J. A., Edwards, J. A., & Tobin, S. J. (2001). Chronic and temporarily activated causal uncertainty beliefs and stereotype usage. Journal of Personality and Social Psychology, 2, 206219.
  • Weary, G., Tobin, S. J., & Edwards, J. A. (2010). The causal uncertainty model revisited. In R. M. Arkin, K. C. Oleson, & P. J. Carroll (Eds.), Handbook of the uncertain self (pp. 78100). New York: Psychology Press.
  • Weary, G., Vaughn, L. A., Steward, B. D., & Edwards, J. A. (2006). Adjusting for the correspondence bias: Effects of causal uncertainty, cognitive busyness, and causal strength of situational information. Journal of Experimental Social Psychology, 42, 8794.