Abstract
Introduction
reject all available alternatives to look for new ones. My study takes the same approach and also examines decision refusal. Indecision is an important issue to study, because it can lead to real problems within companies. Indecision might be harmful for companies: there is a point after which investments in the search for more options will no longer pay off and increases the costs of searching (Nijstad & Kaps, 2008), employee motivation may decrease after a long decision-making process, or companies can miss out opportunities in the market if it takes too long to make a decision, so that another company grasps this opportunity. According to Nijstad (2008), bad decision options should be refused, but indecision can lead to real costs when a decision should be made quickly.
decision will often feel sufficiently confident in their judgments, in other words, they have a high actual confidence (AC). However, when group members hold different perspectives rather than the same preferences, this preference heterogeneity may decrease the actual confidence of the group (De Dreu et al., 2008). But having actual confidence in choices is very important, because confidence can have the effect of motivating and directing the behavior of groups. Insufficient confidence is a threat to successful decision-making, as groups with a low confidence level in their decisions are less likely to have the capability to make future group decisions (Sniezek, 1992). Thus, the relationship between group conflict and group indecision could be mediated by group (actual) confidence. Actual confidences varies across people and situations, and should be compared with a point located along a confidence continuum, in other words, a desired level of confidence (Jain & Maheswaran, 2000). Groups want to attain a sufficient degree of confidence to accomplish their goals. Team members cannot be completely certain that their judgments are correct, but they may achieve some sufficient or desired level of confidence, also known as the sufficiency threshold (ST) (Jain & Maheswaran, 2000). It is possible that if a group is not confident enough about a decision (i.e. AC < ST), it will not make the choice but will search for more information or alternatives to make the choice later on. Moreover, with a high sufficiency threshold, groups suspend judgments, engage in more extensive information search and generate multiple interpretations until their AC no longer falls below their ST (De Dreu et al., 2008). Therefore, I argue that a low actual group confidence will lead to a desire for more information if and only if the desired confidence (i.e. ST) is high. When the AC is lower than the ST, decision makers perceive their state of knowledge and information as insufficient to make a decision and thus want to process more information, thereby delaying the decision.
decision are small, team members may not want to expend the time and effort necessary to reach agreement, but instead adopt a simple decision making strategy, such as choosing the option with the most followers (Posavac, Herzenstein & Sanbonmatsu, 2003). Their desired confidence in the decision does not have to be very high. However, high importance stimulates systematic processing, which means a deeper and more elaborate search for information (Chaiken & Maheswaran, 1994; De Dreu, 2007). Systematic processing is likely when there is a gap between the AC and ST, which in turn leads to decision delay (Chen, Chaiken & Duckworth, 1999). Indeed, Nijstad and Oltmanns (2012) proposed that it seems likely that more important decisions often lead to decision refusal. A likely reason is that decisions with significant consequences lead to a higher desired confidence.
Decision importance has been ignored so far in other studies. Thus, with more important decisions and with group conflict, it is likely that the actual confidence of a group falls below the sufficiency threshold. So, I propose that more important decisions more often lead to group indecision, but especially when the actual confidences are low. Figure 1 represents the conceptual model of the relationship between group conflict, group confidence, the importance of the decision and group indecision.
Figure 1.1: Conceptual model
Theory
Group Indecision
Within group decision-making, it is possible that groups do not make a decision at all, also known as group indecision. In many situations, groups can decide not to choose an option or postpone their decision moment (Nijstad & Kaps, 2008). Corbin (1980) has provided a theoretical framework about decisions that might not get made. There are different forms of not deciding in different stages of a decision process: inattention, decision delay or decision refusal. By inattention, the decision process is not even started. Next, when decision makers are uncertain about which alternative to prefer, they may delay their decision. Finally, decision makers may think that none of the alternatives are good enough, which may lead to no choice being made or a refusal of all alternatives (Nijstad, 2008). It is also important to notice that indecision is not the same as choosing the status quo (i.e. choosing not to change anything), because then a decision is made and the decision process is ended. By indecisiveness there is no decision made at all.A few studies have analyzed antecedents that lead to group indecision. For example, Nijstad and Oltmanns (2012) found that higher epistemic motivation makes decision refusal more likely. Another important and much discussed antecedent is group conflict; it is argued that conflict in groups lead to choice deferral (e.g. Dhar & Nowlis, 1999; Tversky & Shafir, 1992; Nijstad & Kaps, 2008). A group discussion about a certain issue can drag on forever when group members have different opinions and ideas. Therefore, no decision is made as there is no consensus, or the decision moment is postponed. During group discussions, members tend to push their own ideas and preferences, make them seem even bigger and more beautiful than they are and ignore information that counters their own perspective and this results in group conflict (De Dreu, Nijstad & Van Knippenberg, 2008).
members holding different views, opinions and ideas. Relationship conflict exists when there are interpersonal inconsistencies among group members, like personal frictions. De Dreu and Weingart (2003) indicate in their meta-analysis that the two types of conflict often co-occur within organizational teams. However, this study will mainly focus on task conflict. Because I expect that relationship conflict does not occur very often, as participating groups discuss for a short period of time. There may be some annoyances, but hard personal frictions will not occur.
The sufficiency principle proposes a continuum of judgmental confidence, along which two critical points lie: a level of actual confidence and a level of desired confidence (Chen, Chaiken & Duckworth, 1999). Jain and Maheswaran (2000) labeled this desired level of confidence as the sufficiency threshold (ST). This is the level of confidence a person, or a group, aspires. People cannot be completely certain that their judgment or decision is correct; therefore they have to achieve some sufficient or desired level of confidence. The level of confidence that is subjectively experienced by the perceiver is known as the actual confidence (AC). It is important that when the actual confidence reaches the sufficiency threshold, people feel adequately confident that a given judgment or decision will satisfy their concerns (Chen, Chaiken & Duckworth, 1999). Groups are less inclined to make a decision and postpone decisions more often when there is a gap between the AC and the ST (i.e. AC < ST).
When decision makers are satisfied and feel that they have sufficient information available to choose, decisions can be made.
that is a high AC. Hence, I propose that group conflict will lower the (actual) confidence of the group about the correctness of their decision.
Hypothesis 2A (H2A): More group conflict results in a lower actual group confidence.
Subsequently, group confidence is also related to group indecision. According to Crawford (1974), conflict instigates a state of uncertainty and the primary method of reducing uncertainty (and increasing actual confidence) is information search. This statement is supported by several other studies. For example, Schulz-Hardt et al. (2002) found that a low actual confidence caused by disagreement indicates that a critical examination is necessary and, thus, facilitates the search for more information. Nijstad and Oltmanns (2012) stated that group members are only willing to adopt a non-preferred alternative when they are fully convinced. So in general, group members want to have a high actual confidence about the decision, otherwise the decision is not accepted. If groups are in conflict and their actual confidence is low, they might think that their current state of knowledge and information about the decision is insufficient. So, they are more motivated to engage in even more processing of decision-relevant information (De Dreu et al., 2008). This all takes time and effort and results in a delay of the decision moment. Hypothesis 2B (H2B): Actual confidence is related to group indecision. Taken together these two hypotheses, imply that more group conflict leads to a lower actual group confidence, which may lead to more information gathering and thereby group indecision.
Hypothesis 2C (H2C): The positive relationship between group conflict and group indecision is mediated by group confidence.
desired level of confidence. The higher the ST, the longer it takes before a decision is made. The sufficiency threshold level can be determined by several variables (e.g. epistemic motivation). But in this study, the ST is affected by one important factor: decision importance.
Importance of decision as moderator
An actual confidence level does not mean much in the absence of some standard, that is, how confident one aspires to be. To determine this desired level of confidence (i.e. sufficiency threshold), I use the importance of a decision as a factor. I argue that the importance of the decision influences the sufficiency threshold: important decisions increase the ST.
Decision importance is in turn related to the sufficiency principle: important decisions increase the sufficiency threshold, because the decision potentially has serious consequences. Therefore, groups want to be sure that they choose a good alternative and investigate the problem in its full complexity. High decision importance leads to a greater interest in and the selection of more information. In contrast, when the consequences of a decision are perceived to be insignificant or meaningless, team members may not want to expend the time and effort necessary, but instead generate a limited number of alternatives (Posavac et al., 2003). Then, the desired level of confidence is not that high, because group members care less about the fact that the judgment or choice is correct. Thus, I argue that an important decision increases the sufficiency threshold of a group.
condition). They received €8 or 4 course credits for participating. Participants were told that they were members of a three-person committee that had to choose among three male candidates who had applied for a workgroup teaching position at the Business Administration department. They had to decide whom to hire, but could also decide not to hire any of the candidates. I used a 2x2 experimental design, in which group conflict (low vs. high) and the importance of the decision (important vs. not important) were manipulated. Conflict was manipulated by giving different members of a group a pre-commitment to competing alternatives. This was done through information distribution. The group as a whole always received all information, but the individuals received different pieces of information. The importance of the decision was manipulated through different instructions. Half of the groups were asked to hire a workgroup teacher for several important courses and for a longer period of time, the other groups were asked to hire a workgroup teacher for a less important course and for a temporary job for one semester.
Materials and manipulations
attributes were (with average important ratings): enthusiasm for teaching (5.28), conscientiousness (5.56), patience (5.31), clarity when explaining (6.46), ability to keep order (5.34) and involved with students (5.05). The average important ratings were very similar to each other, so I distributed the attributes randomly to the candidates. Participants received the candidate profiles and made an individual decision before the discussion, which candidate they would hire. In the high conflict condition, profiles were biased in such a way that group members had different preferences before the group discussion. The participants received different distributions of the pros and cons of each candidate, which should create conflicting opinions. For example, group member 1 received 3 pieces of positive and 1 piece of negative information about candidate A, 3 pieces of negative and 1 piece of positive information about candidate B, and 2 negative and 2 positive pieces of information about candidate C. Then, group member 1 should be in favor of candidate A, indifferent to candidate C and disapprove candidate B. The information distribution was different for group members 2 and 3 in the conflict condition and this distribution is showed in Table 3.1. This created conflicting preferences, but the whole set of information was not in favor of any of the candidates.
Table 3.1: Pre-discussion distribution of information Candidate Complete profile Group member Information distribution Conflict No conflict A (Adam) 1+, 2-, 3-, 4-, 5+, 6+ 1 2 3 1+, 3-, 5+, 6+ 1+, 2-, 4-, 6+ 2-, 3-, 4-, 5+ 1+, 3-, 5+, 6+ 1+, 2-, 5+, 6+ 1+, 4-, 5+, 6+ B (Bryan) 1-, 2+, 3+, 4+, 5-, 6- 1 2 3 1-, 2+, 5-, 6- 1-, 2+, 3+, 4+ 3+, 4+, 5-, 6- 1-, 2+, 5-, 6- 1-, 4+, 5-, 6- 1-, 3+, 5-, 6- C (Chris) 1-, 2-, 3+, 4+, 5-, 6+ 1 2 3 2-, 4+, 5-, 6+ 1-, 2-, 3+, 5- 1-, 3+, 4+, 6+ 1-, 2-, 4+, 5- 1-, 2-, 3+, 5- 1-, 2-, 6+, 5- Note: 1 = enthusiasm; 2 = conscientiousness; 3 = patience; 4 = clarity; 5 = keeping order; 6 = involved. Positive information is indicated with a +, negative with a -. The bold numbers show the expected preference for a candidate in each conflict condition.
To manipulate the importance of the decision, the information the groups received was different. Half of the groups had to make an important decision, the other half a less important decision. In the high importance condition, group members received the following instruction: “this is an important decision, because the workgroup teacher will be hired for several important courses. An extra workgroup teacher is necessary and since this year required by the faculty. This is a position for a longer period of time”. In the low importance condition, the instruction was as follows: “this is a less important decision, because the workgroup teacher will be hired for a less important course. An extra workgroup teacher is not so much necessary, but is required by the faculty. This is a position for one semester, thereafter the old workgroup teacher will come back”.
important courses and that it is a position for a longer period of time”. The instruction in the low importance condition was: “please note that the workgroup teacher will be hired for a less important course and that it is a temporary position”. The manipulation was checked in the pre- and post-discussion questionnaire.
Procedure
All instructions and questionnaires were provided on paper. Three participants entered the video room and were first seated individually at a table and completed a personality test, measuring two traits (agreeableness and extraversion) of the big 5-personality dimensions test (this was part of another study). Thereafter, participants read a general instruction and the candidate profiles. The instruction told them to imagine that they were members of a 3-person selection committee that had to fill a vacancy. They were instructed to decide, together with the other 2 participants, which candidate they would hire, or hire none of them.
Measures
Pre-discussion questionnaire
Each participant was asked to choose the favorite decision option (candidate A, B, C or none of them) and rated the job suitability (Nijstad & Kaps, 2008) of each candidate on a 7-point scale (1= totally disagree, 7= totally agree). These 3 items per candidate (e.g. “candidate A/B/C would be a good workgroup teacher”) were averaged to create job suitability (Cronbach’s α: 0.94, 0.95 and 0.95 for candidate A, B and C respectively). Participants also mentioned how certain they were about their choice on a 7-point scale (1= totally uncertain, 7= totally certain) (Nijstad & Kaps, 2008). Lastly, the importance of the decision was measured with 5 items on a 7-point scale (1= totally disagree, 7= totally agree). These questions were based on ‘the importance of the task conflict’ items from Jehn, Greer, Levine & Szulanski (2008) and Rispens (2012). For example, “this discussion is about an important problem”. The scale proved reliable, Cronbach’s α was 0.95.
Post-discussion questionnaire
Table 4.1: Ratings of candidate A, B and C (Mean and Standard Deviation) Condition Conflict No conflict A B C A B C Participant 1 2 3 5.86 (0.61) 3.88 (1.32) 2.45 (0.90) 2.17 (0.96) 5.45 (0.50) 4.07 (0.92) 4.24 (0.89) 2.62 (0.86) 5.34 (0.55) 5.48 (0.69) 5.29 (0.76) 3.74 (1.40) 2.19 (0.71) 3.50 (1.31) 2.70 (1.04) 2.76 (1.03) 2.36 (1.02) 2.29 (0.85)
Note: candidate ratings were measured on a 7-point scale. Standard deviation in parentheses. The bold numbers represent the expectations about the candidate ratings in each condition.
Second, a 2x2 univariate ANOVA was conducted on the post-discussion task conflicts (1&2), with the manipulations as fixed factors (conflict & importance). The results showed a main effect of the conflict manipulation on the second task conflict measures, F (1, 24) = 17.18, p < 0.01. The effects of importance, F (1, 24) = 0.01, p = 0.94 and the interaction, F (1, 24) = 0.03, p = 0.86 were not significant. Task conflict was perceived to be higher in the conflict condition (M = 4.39, SD = 1.05) than in the no-conflict condition (M = 2.46, SD = 1.31). However, the first measures on task conflict showed no significant effect of the manipulation, F (1, 24) = 2.32, p = 0.14. The effects of importance, F (1,24) = 0.22, p = 0.64 and the interaction, F (1, 24) = 0.00, p = 0.99 were not significant. Task conflict was still higher in the conflict condition (M = 3.27, SD = 1.01) than in the no conflict condition (M = 2.58, SD = 1.28). Although the last check was not significant, I assume that the conflict manipulation worked, based on the other tests.
high importance condition (M = 5.67; SD = 0.44) than in the low importance condition (M = 3.68; SD = 0.75). Second, the effect of the importance manipulation on post-discussion importance was also significant, F (1, 24) = 48.99, p < 0.01. The main effects on conflict, F (1, 24) = 0.67, p = 0.42 and the interaction, F (1, 24) = 0.09, p = 0.77, were not significant. The perceived importance of the groups was higher in the high importance condition (M = 5.77, SD = 0.38) than in the low importance condition (M = 3.73, SD = 0.99). Therefore, the manipulation of the importance of the decision was successful. Taken together, the results showed that the manipulations were successful.
Group decision (hypothesis 1)
Table 4.2 reports the main dependent variables of this study by condition. For hypothesis 1, group decision was coded dichotomously as either decision (choose candidate A, B, or C) or indecision (hire none of the candidates). It was predicted (H1) that conflict in groups leads to more group indecision. A chi-square test revealed that there was no direct effect from group conflict on group indecision, χ2 (1, N = 28) = 0.00, p = 1.00. In the conflict condition, 28.6% of the groups made no decision, in the no conflict condition this was also 28.6% (both 4 out of 14). Thus, hypothesis 1 is not confirmed in this study.Group confidence (hypothesis 2)
Group confidence was measured with the actual confidence of individuals; this variable was aggregated to the group level by taking the group mean. First, it was predicted that conflict results in a lower actual group confidence (H2A), but a 2x2 univariate ANOVA did not yield a significant effect of conflict on actual confidence, F (1, 24) = 1.98, p = 0.17. There was no significant difference between actual confidences in the conflict condition (M = 7.22; SD = 1.05) and the no conflict condition (M = 7.75; SD = 0.93)1.
Table 4.2: All other variables by condition.
Conditions
High importance Low importance
Conflict No conflict Conflict No conflict
Group Indecision (%*) Actual confidence Sufficiency threshold 57.1% 7.54 (1.32) 7.98 (0.55) 42.9% 7.86 (0.98) 8.38 (0.23) 0% 6.90 (0.62) 7.60 (0.69) 14.3% 7.65 (0.95) 7.55 (0.74)
Note. *: % groups with indecision. Standard deviations in parentheses. Actual confidence and sufficiency threshold were measured on a 10-point scale. Next, in hypothesis 2B was predicted that actual group confidence is related to group indecision. In a logistic regression, actual confidence had no significant relation with indecision, B = 0.79, SE = 0.47, Wald = 2.83, p = 0.09. Moreover, the positive effect suggests that a higher confidence was associated with more rather than less indecision. So actual confidence increased indecision, while I predicted that a lower confidence level would result in more indecision. Finally, hypothesis 2C predicted that actual confidence in a group would mediate the effect of conflict on group indecision. However, given that conflict had no effect on the mediating variable (group confidence), there is no evidence for mediation, and the hypothesis should be rejected.
In addition to the hypotheses, in the theory section I assumed that the moderator, decision importance, is related to the desired level of confidence (i.e., the sufficiency threshold). This assumption was confirmed with a 2x2 univariate ANOVA, F (1, 24) = 7.45, p = 0.01, and with important decisions (M = 8.18, SD = 0.46) the ST was higher than with less important decision (M = 7.57, SD = 0.69) 1. The main effects of conflict, F (1, 24) = 0.64, p = 0.43, and the interaction, F (1, 24)
Importance of the decision (hypothesis 3)
The last hypothesis in my moderated-mediation model was tested with multiple regressions, using the SPSS Process Macro (Hayes, 2015). It was expected that conflict in groups decreased the actual confidence of groups, and that this would lead to indecision, especially for important decisions (H3). However, this hypothesis received no support. The analysis on the moderated-mediation model showed no significant interaction between group confidence and importance, B = -1.1, SE = 1.48, z = -0.75, p = 0.46, showing that the relation between confidence and indecision is not moderated by decision importance. Furthermore, the conditional indirect effects (mediation) were not significant (see Table 4.3). Consequently, this indicates that the overall model is not significant in this study. Table 4.3: Summary of the moderated-mediation analysis. Conflict Importance Actual confidence Interaction (AC*imp) Dependent variables Actual confidence Indecision 0.53 (SE = 0.37) - - - -0.17 (SE = 0.61) 5.80 (SE = 10.76) 0.87 (SE = 0.61) -1.10 (SE = 1.48) Conditional indirect effects
Importance Effect Boot SE Boot LLCI Boot ULCI
AC AC 0 (important) 1 (not important) 0.46 -1.23 3.30 1.76 -1.15 -6.28 12.92 0.53
Additional results
indecision) was tested, I found that the indirect effect from importance of the decision on group indecision was mediated by the sufficiency threshold, B = -1.44, SE = 2.38, LLCI = -5.95 and ULCI = -0.03. The zero was excluded, so the result was significant. This demonstrates that the ST was higher when the decision was more important, and that a higher ST subsequently led to more group indecision. Lastly, a chi-square test also confirmed that decision importance had a direct effect on group indecision, χ2 (1, N = 28) = 6.3, p = 0.01, and important decisions (50%) were postponed more
often than unimportant ones (7.14%).
Discussion
time, group members felt confident enough to make a decision: the overall actual confidence levels were very high, even in conflict situations (M = 7.22, on a 10-point scale), and conflict did not affect the actual confidence levels. Thus, group confidence did not mediate the relationship between conflict and indecision. Finally, the overall moderated-mediation did not work: there was no evidence that decision importance moderated the relation between actual confidence and group indecision, and the indirect effect of conflict through actual confidence on group indecision was neither found for unimportant nor for important decisions. One possible reason for this is that the number of participating groups was too low.
However, I did find support for a relation between decision importance and group indecision. I found that there was more indecision in the high (50%) rather than the low (7.14%) importance condition. It was also found that important decisions increased the sufficiency threshold, and that the ST mediated between decision importance and group indecision. To explain this pattern of findings I refer to Table 4.2 in the Results section. It is perhaps not surprising that almost no groups were indecisive in the low importance condition. Because, when the consequences of a decision are small, people may not be willing to expend the time and effort necessary, and in stead adopt a simple decision-making strategy and choose the first option that comes to mind (Posavac et al., 2003).
Implications
I believe that the group decision-making literature neglected two related issues that were important in this study: the sufficiency principle in combination with decision importance. This new insight contributes to existing literature. First, the moderated-mediation analysis did not work and although I could not confirm that there is more indecision when the AC falls below the ST, I can confirm another part of this sufficiency principle theory. Actual confidence was generally high and almost at the same level as the sufficiency thresholds. All the confidence levels were on average between the 70% and 80% in every condition (see Table 4.2). This confirmed that when decision makers feel they already have the information they need to make a decision (i.e. AC ≥ ST), examining the information again or searching for new information is no longer worth the effort and the decision can be made (De Dreu, Nijstad & Van Knippenberg, 2008). Such a decision was made in 71.6% (20 out of 28) of the groups. However, also 28.4% (8 out of 28) of the groups chose to hire none of the candidates, and 7 out of these 8 indecision groups had to make an important decision. So, high importance probably influenced the confidence in the correctness of the decision, and then groups chose to postpone the decision. This is further explained below.
(1999) found that a high task importance stimulates systematic processing and thus involves an in-depth treatment of the information, and Chaiken and Maheswaran (1994) stated that a high importance should result in more elaboration of relevant information. Which may lead to a postponement of the decision. However, decision importance is never related with epistemic motivation (as part of the motivated information processing in groups (MIP-G) model). This could be further addressed in future research. The expectation is that high importance can be associated with high epistemic motivation, because high importance stimulates systematic processing and this processing is likely with high epistemic motivation (De Dreu, Nijstad & Van Knippenberg, 2008). A high epistemic motivation is in turn good for group decision-making, because groups expend effort to achieve a thorough and accurate understanding of the decision problem and this results in better decisions (De Dreu, 2007).
In addition, more research on decision importance is recommended, because this issue is also important from a practical point of view. That important decisions lead to more indecision is a potential downside for organizations, because it is not good if especially important decisions are often postponed or refused. In the end, one has to make the decision, precisely because it is an important one. A positive point is that more effort is invested in more information search for existing options or in finding good alternatives and that the team is motivated to make a good choice. Indeed, being accurate has its advantages and disadvantages, as ascribed above. But for managers, it is important to consider all possibilities and find the right balance between accuracy and the moment of decision. To what extent should one postpone decisions and obtain new information, and when should decisions be made? For example, some opportunities (in the market) are unique and organizations do not want that their competitors grasp these opportunities. This might be a consequence when decision-making takes too long.
it is said that a higher ST leads to higher epistemic motivation and more systematic processing. It is also said that discrepancies between the AC and ST trigger epistemic motivation (De Dreu et al., 2008). However, in my study, there are almost no (big) discrepancies between the AC and ST, both confidences are generally high. Then, I expect less epistemic motivation (i.e. expend effort to have an accurate understanding in the decision problem), because confidence in the correctness of the decision is already high. But the ST level is also high, which should lead to a high epistemic motivation. These statements contradict each other. In my opinion, there is missing information in the literature about what happens with epistemic motivation if both the AC and ST are high.
Limitations and future research
In addition, because participants felt no real consequences, there may be a lack of process accountability (that is, the need to obtain a rich and accurate understanding of the issue). When this happens, individuals are more likely to jump to conclusions on the basis of insufficient evidence, and do not engage in deep thinking. Groups may forego extensive processing of information and instead form a final judgment by aggregating members’ preferences (Scholten, van Knippenberg, Nijstad & De Dreu, 2007). For this reason, fieldwork is also important. Then, groups feel that they are accountable for the decision-making process and there might be different outcomes. Groups would be more motivated to engage in deep information processing (De Dreu, Nijstad & Van Knippenberg, 2008). And, as mentioned before, decision makers have to find the right balance between accuracy and the decision moment.
Another limitation may be that some groups consisted of a few friends or sometimes even three friends. This group composition may have influenced the decision making process: with strangers, someone with a certain personality can take the leader position or dominate the discussion, but with friends, these roles are often fixed. The decision-making process may differ by the composition of the group. For example, Decrop, Pecheux and Bauvin (2007) stated that groups of friends are more efficient than groups of acquaintances, but are also more vulnerable to symptoms of groupthink (negative influence on the quality of the decision making process). Friendship can hinder productivity by creating a relaxed atmosphere that distracts the group from the objective; decisions are delayed and made late. So, there are enough reasons to take into account the influence of friendship in decision-making in future (field) research. Although groups of friends are not common within organizations, it is possible that groups of colleagues know each other for a longer period of time and made friendly relationships, and that this friendship affects the decision-making process as described above.
2003). However, I only focused on task conflict. In this study, it became clear that there was no relationship conflict. Participants felt, on average (on a 7-point scale), no friction or tension within the decision-making group (M = 1.89), and participants found each other nice and sympathetic (M = 5.92). For future (field) research, it would also be interesting to look at relationship conflict. Within organizational teams, there are probably more interpersonal incompatibilities (i.e. relationship conflict), as team members know each other for a longer period of time. So in field research, relationship conflict might also be used as a measure of intergroup conflict.
Finally, in this study, the no choice option was the option to look for new alternatives. Corbin (1980) provided a framework for forms of indecision, but I distinguished different forms for the no-choice option, such as postponing a decision in order to search for more information about available alternatives (rather than looking for new ones) (Nijstad, 2008). There were questions in the pre-discussion test about whether participants found that they had sufficient information about the candidates to give a good opinion. The average score was 3.57 (on a 7-point scale), which is a neutral score and could indicate that participants would like to have more information about the available job candidates before making a decision. Therefore, it could be interesting to investigate different forms of indecision, such as: looking for new alternatives, more information about available alternatives, or refuse to not decide at all. Sometimes, it might even be possible to shift responsibilities for a decision to someone else, in cases where groups cannot come to a good conclusion (Nijstad & Kaps, 2008). These are all interesting possibilities for future research. As an example: search for more information might be preferable when available alternatives are close in attractiveness, or look for new alternatives might be preferable when the alternatives are not attractive at all.
Conclusion
Summarized, although in this study no relation was found between conflict and indecision, I did find that there is a relationship between the desired level of confidence (ST) and the importance of the decision. The ST mediates the relationship between important decisions and group indecision. This indicates that important decisions more often lead to group indecision and that groups more often want to decide later on when decisions are important. Thus, although important decisions are often made more thoroughly than unimportant ones, they have a potential downside: they more easily lead to indecision.
