With a little help from your friend

Abstract

Transactive memory is a system for encoding, storing, and retrieving information between people, where each person has knowledge of the other’s memory. Through two studies, we assessed whether transactive memory occurs in best friendships (N = 682). Results showed that transactive memory systems (TMSs) do exist in best friendships. Importantly, stronger TMSs are associated with higher friendship quality (satisfaction and commitment), and their strength is related to different friendship characteristics (e.g., trust). A novel method for assessing TMS structure was developed. Mixed-gender friendships were associated with more differentiated structures (different knowledge), and friendships higher in inclusion of other in the self were associated with more integrated structures (similar knowledge). These studies have implications for the quality and operation of friendships.

Keywords

Differentiated memory structure friendship friendship quality integrated memory structure transactive memory

Imagine you invited your boss over for dinner. Suddenly you realize she loves wine and you don’t have any. You want a good wine to pair with dinner, but you don’t know anything about wine. What would you do? You may call a close friend who knows about wine to help you choose the best option. It is even possible that you haven’t learned the fine details about wine because your friend is an expert—and maybe that friend relies on you for your knowledge of beer. If you each know you can trust the other’s information and that information is available to you when needed, it may not be necessary for you both to possess knowledge about wine and beer. These collective knowledge structures in which information is shared across individuals is transactive memory. The purpose of this paper is to demonstrate that transactive memory systems (TMSs) exist in best friendships and are associated with friendship quality.

TMSs are defined as “shared systems for encoding, storing, and retrieving information” (Wegner, Erber, & Raymond, 1991, p. 923) and occur within groups of at least two people. TMSs utilize each individual’s knowledge through meta-memory processes (e.g., knowing what another knows). By using TMSs, individuals can access the other’s knowledge (Lewis, 2003) and access more than the sum of each individual’s knowledge by cueing knowledge recall in others.

TMSs have been examined in romantic couples (Hollingshead, 1998; Wegner et al., 1991), work groups (Lewis, 2004), classrooms (Jackson & Moreland, 2009), and groups created in the lab (Liang, Moreland, & Argote, 1995). Although it has been postulated that TMSs exist in friendships (Wegner, 1986), there is currently no published research on TMSs in these important groups. Furthermore, even though research has demonstrated that TMSs exist in romantic relationships, no research to our knowledge has explored the relationship characteristics that may be associated with these systems.

Examining TMSs in friendships is important for a variety of reasons. First, research on nonromantic relationships—including friendships—has been relatively neglected (Berscheid & Regan, 2005). Thus, conducting research on friends fills in gaps in the close relationships literature. Additionally, TMSs may allow people to gain more benefits from friendships and even aid in activities within the friendship. For example, if a friend has knowledge about raising children, the friend can be an important resource when you start a family. Or if you and your friend are planning a vacation, you may rely on each person’s knowledge in different areas (e.g., organizing the travel arrangements, budgeting, etc.) to create a better vacation than either could have planned alone. Because TMSs are associated with these positive coordination and knowledge benefits, it is valuable to know what types of friendships are likely to provide these benefits (e.g., same-sex friendships, friendships that have existed longer, etc.). Investigating TMSs in friendships also allows us to compare how these systems may be different in friendships versus in romantic relationships or in work groups (e.g., if the amount of time spent together is associated with the strength of TMSs in work groups, is it also associated with the strength of TMSs in friendships?). Finally, and perhaps most importantly, studying TMSs in friendships is crucial because we believe that TMS strength is associated with higher friendship quality, including greater satisfaction and commitment. In sum, exploring TMSs in friendships is of both theoretical and practical interest.

The present studies examine whether TMSs exist in friendships, the association between TMS strength and friendship quality, and friendship characteristics potentially related to the strength of these TMSs. TMSs are likely to develop in friendships, as TMSs develop when people rely on and are aware of each other’s knowledge, and people often rely on and value their friends (Carbery & Buhrmester, 1998; Pulakos, 1989). Because best friendships are more intimate than common friendships, displaying higher levels of trust and interdependence (Yager, 1997), TMSs may be particularly likely to develop in this type of close relationship (Wegner, Giuliano, & Hertel, 1985). Thus, we focus our research questions on exploring best friendships. We are interested in how friendship quality (e.g., commitment and satisfaction) and descriptive friendship characteristics (e.g., friendship length and gender composition) may be associated with TMS strength and structure. Given the benefits of friendships and TMSs, exploring the association between specific friendship characteristics and TMSs can provide insight into what makes friendships most productive and beneficial. We begin by detailing previous work on TMSs, followed by a discussion of the substantial benefits friendships provide and the importance of friendship quality, and we end by describing why various friendship characteristics should be related to TMS strength.

The existence of TMSs in friendships

Initial research on TMSs assessed its operation within romantic relationships. Couples came to the laboratory and either worked together on a memory task or were separated and worked with an opposite-gender partner from another couple (Wegner et al., 1991). Some dyads were assigned structures of who should remember what and others were not. Intact couples outperformed noncouples with unassigned structures, however, the opposite occurred with assigned structures (Wegner et al., 1991). This demonstrated the existence of TMSs in romantic couples, in that they performed better when working together normally but were disrupted by artificial systems.

More recent research has assessed TMS operation within work groups. In initial research examining TMSs in work groups (Liang et al., 1995), TMSs were assessed by observer ratings of observable manifestations, or processes, of TMSs: whether each person remembers specific information, effective knowledge sharing, and trust in each other’s knowledge. Lewis (2003) then created a scale that assesses these aforementioned processes of TMSs. Higher scores on this scale indicate greater TMS strength—that is, greater knowledge of the other’s expertise and greater trust in and coordination of this information. Thus, friends with stronger TMSs should demonstrate these TMS processes to a greater degree than friends with weaker TMSs. Specifically, best friends with stronger TMSs should be more likely to report specializing in particular knowledge, knowing what information their best friend specializes in, trusting their friend’s knowledge, and coordinating the sharing of this knowledge. Looking back at our opening example, the friend who knows less about wine may purposefully not learn more because they can ask their friend. However, they must believe their friend’s knowledge is accurate and have ways to access that knowledge—otherwise, their friend’s knowledge would not benefit them (Lewis, 2003).

Although research demonstrates that TMSs exist in romantic relationships and work groups, it has not been explored in friendships. Thus, we first wanted to demonstrate that TMSs exist in friendships. TMSs develop when people rely on each other’s knowledge (Wegner, 1986). As people often rely on their friends and find them important (Carbery & Buhrmester, 1998; Pulakos, 1989), TMSs are likely to develop in friendships. As mentioned above, we focus this initial study of TMSs in friendships on exploring best friendships as these relationships are particularly intimate. Developing and maintaining a close friendship with another should lead to knowledge of what the other knows and the development of trust in and access to their knowledge. Thus, strong TMSs should exist in best friendships (Hypothesis 1).

We sought to demonstrate the existence of TMSs in best friendships using two different types of assessments. First, we adapted the Lewis (2003) scale to friendships and measured TMSs in this way. The Lewis (2003) TMS scale has been used to measure TMS strength in over 20 published manuscripts in many types of groups, including classroom groups (Jackson & Moreland, 2009) and MBA consulting teams (Lewis, 2004). However, to our knowledge, Lewis’s (2003) scale has never been used to assess TMSs in personal relationships, making this a novel application of the scale. Thus, we used a second type of assessment of TMSs in friendships. To ensure the scale adequately assesses TMSs in this context, we related the scale to measures of behaviors representative of TMSs (e.g., knowing what information the other has, relying on the other’s knowledge, etc.). Thus, reports of behaviors related to TMSs should be positively associated with TMS strength (Hypothesis 2).

Friendship quality and TMS strength

It is necessary to study TMSs in friendships because friendships are important (Fehr, 2008), contributing to both physical and psychological health. Siebert, Mutran, and Reitzes (1999) found that identity as a friend had the strongest relation to life satisfaction compared to other variables (e.g., income and marital status). Social support from friends is also important to psychological well-being (Winefield, Winefield, & Tiggemann, 1992). For elderly people, contact with friends reduces mortality risk (Sabin, 1993). Thus, evidence demonstrates friends play an important role in psychological and physical health. Due to the importance of friendships, it is crucial to understand what factors facilitate friendship maintenance (i.e., commitment), and TMSs may be one such factor.

Although having friends provides psychological benefits, friendship quality is also important. Waldrip, Malcolm, and Jensen-Campbell (2008) found that children with lower friendship quality and fewer friends had greater maladjustment, and friendship quality may buffer against maladjustment resulting from fewer friends. In other research, quality of best friendships was an important predictor of happiness (Demir, Ozdemir, & Weitekamp, 2007). Because friendship quality is essential, it is advantageous to understand factors that are related to, and could potentially influence, friendship quality.

We believe TMSs are related to friendship quality (i.e., commitment and satisfaction) and thus may promote friendship maintenance and the positive outcomes associated with friendship quality. Friendship satisfaction indicates that the friendship meets a person’s needs. Early research suggested that the more romantic partners agree on who knows what, the more satisfying the relationship will be (Wegner et al., 1985). Commitment, another indicator of relationship quality, is thought of as an intention to persist, psychological attachment, and long-term orientation (Rusbult, Martz, & Agnew, 1998). Relying on a friend and gaining knowledge of what they know may be associated with increased commitment. Thus, TMS strength should be positively associated with satisfaction and commitment (i.e., friendship quality; Hypothesis 3).

TMS strength and friendship characteristics

Because we believe TMS strength is associated with friendship quality, a critical factor in people’s physical and psychological health, we are also interested in the characteristics of best friendships that are related to TMS strength. Based on previous research and theory, we believe the length of the friendship, the amount of time spent together, the gender composition of the friendship, the interdependence of the friends, and the trust the friends have in each other should predict the strength of their TMS.

Regarding friendship length, being friends with someone longer provides more opportunities to learn what they know. Indeed, as teams worked together longer, their TMS strength increased (Lewis, 2004). Thus, friendship length will be positively associated with TMS strength (Hypothesis 4).

Friends who spend more time together may also have stronger TMSs as they may encounter more situations where they learn the other’s knowledge. Previous research on classroom groups (Jackson & Moreland, 2009) and romantic couples (Hollingshead, 1998) suggests that face-to-face interaction is important to TMS development and implementation. Thus, time spent together will be positively associated with TMS strength (Hypothesis 5).

Gender composition of a best friendship may also produce different dynamics, potentially leading to differences in TMS strength. Women feel more at ease with and find it easier to feel close to other women (Reisman, 1990). Same-gender friendships were also seen as more supportive than mixed-gender friendships (Hand & Furman, 2009). Because same-gender friends are more supportive and at ease with each other, they should have greater knowledge of and accessibility to the other friend’s knowledge. Additionally, same-gender friends are likely to be more similar, potentially contributing to greater coordination and communication of each other’s knowledge. Thus, same-gender best friendships will report stronger TMSs than mixed-gender best friendships (Hypothesis 6).

Another characteristic that may be related to the TMS strength in a best friendship is interdependence, represented by the inclusion of the other in the self (IOS; Aron & Aron, 1986). IOS is a concept derived from self-expansion theory (Aron & Aron, 1986), which suggests that people are motivated to expand the self. One way to do this is to include another’s resources, perspectives, and identities in the self-concept, which is especially likely when two people are more interdependent (Aron, Ketay, Riela, & Aron, 2007). Including the other in the self and being more interdependent should elicit knowledge of what the other knows and allow others to rely on this knowledge. Additionally, including the other in the self-concept could lead to gaining the other’s perspective, providing insight into when they may need help recalling information (e.g., cued recall). Thus, higher IOS will be positively associated with TMS strength (Hypothesis 7).

The last characteristic of friendships that we believe should predict TMS strength is trust. Interpersonal trust, expecting someone can be relied upon to respond to one’s needs even with conflicting self-interests (Rempel, Holmes, & Zanna, 1985), allows friends to rely on each other to remember specific things. In a work setting, trust in teammates predicted transactive memory (Robertson, Gockel, & Brauner, 2013). Thus, friendships characterized by more trust should have stronger TMSs (Hypothesis 8).

Present studies

We designed two studies examining the association between friendship quality and characteristics and TMS strength. Study 1 examined the relationship between TMSs and friendship quality and characteristics. Study 2 replicates and expands on Study 1 using a newly developed measure that explores TMS structure, a dimension of TMSs that will be discussed in detail below. Thus, these studies make multiple contributions to the literature by identifying the existence of TMSs in best friendships, determining whether TMS strength is related to friendship quality in best friends, assessing characteristics of best friendships that are associated with TMS strength and structure, adapting a previously developed scale in a new way, and creating a new method of measuring TMS structure.

Study 1

Study 1’s goals were to demonstrate that TMSs exist in friends, that behaviors representative of TMSs are related to TMS strength, that TMS strength is associated with friendship quality, and that different friendship characteristics are associated with TMSs.

Method

Participants

Volunteer participants (n = 161) were recruited via the Internet and from introductory psychology classes at a large Midwestern university in the United States (n = 64) in exchange for partial course credit.¹ Participants below 18 years old (n = 5) were removed from analyses. Additionally, to reduce the possibility that one person completed the survey multiple times, we excluded data from later instances of IP addresses (n = 4; Gosling, Vazire, Srivastava, & John, 2004).

This left 216 participants (102 females). Participants’ age ranged from 18 to 68 (median = 20, interquartile range [IQR] = 4). The majority of participants were Caucasian (60.6%) and others were African American (7.9%), Hispanic/Latino (6.5%), Asian/Pacific Islander (20.4%), and other (4.7%). Participants were mostly from the United States (79.6%), although 3.7% were from Europe and 16.2% from other places (e.g., Canada and Australia). Friendship length varied from under a year to 43 years² (median = 6.00, IQR = 7). Thirty-three (15%) friendships were mixed-gender.

Procedure

Participants completed an Internet survey via links posted on socialpsychology.org, http://psych.hanover.edu/research/exponnet.html, and www.onlinepsychresearch.co.uk. Other participants were recruited through snowball sampling—members of the research team recruited acquaintances to complete the survey and these people were then asked to send the survey to others. Students signed up and completed the study online. In all cases, the study was listed as concerning “interpersonal memory processes.”

First, participants answered basic demographic questions. Then, participants were told to think of a best friend who was not a romantic partner while answering the questions.

TMS strength and behaviors

To assess TMS strength, participants answered questions adapted from Lewis’s TMS Scale (Lewis, 2003). Lewis and Herndon (2011) recommend using the Lewis (2003) scale when exploring the existence or strength of TMSs. We adapted the original questions from work groups to friendships (e.g., “I know whether me or my best friend has knowledge in specific areas,” “I trust that my best friend has credible knowledge”). The measure consists of 15 questions assessed on 5-point scales (1 = strongly disagree; 5 = strongly agree), with higher scores indicating greater TMS strength (α = .86).

In addition, participants answered 10 questions regarding TMS behaviors in their friendship (e.g., “My best friend and I can remind each other of things we know”). See Appendix 1 for these questions. We developed these to compare Lewis’s (2003) scale with behaviors representing transactive memory processes. As the scale has not been used in a close relationships context before, we wanted to ensure the scale appropriately captured TMSs in friendships. These questions were developed from descriptions of cued recall and other transactive memory processes, such as knowledge of and relying on the other’s knowledge (Hollingshead, 1998). These questions were assessed on 5-point scales (1 = strongly disagree; 5 = strongly agree) and formed a reliable measure (α = .86).

Friendship quality

Participants answered questions regarding friendship quality: shortened versions of the satisfaction and commitment components of the investment model scale (Vanderdrift, Lehmiller, & Kelly, 2012) adapted to relate to friendships. Two questions for satisfaction (e.g., “I feel satisfied with our friendship”) and two questions for commitment (e.g., “I am committed to maintaining our friendship”) were answered on 9-point scales (0 = do not agree at all; 8 = agree completely) and were combined to form a reliable scale of friendship quality (α = .84).

Friendship characteristics

Participants completed objective questions about their friendship, including how long they had been friends (in years), how much time they spent together per week (in hours), and their best friend’s gender.

Participants also answered subjective friendship questions. They completed an IOS scale (Aron, Aron, & Smollan, 1992) consisting of seven pictures of two circles ranging from not overlapping to almost completely overlapping. Higher numbers represent higher IOS and thus greater interdependence.

Lastly, participants answered questions about how much they trust their best friend. We selected six questions from Rempel and colleagues’ (1985) trust scale³ for romantic partners and adapted them to friends (e.g., “I can rely on my best friend to keep the promises he/she makes to me”). They were answered on 7-point scales (1 = strongly disagree; 7 = strongly agree) and formed a reliable measure (α = .76).

Results

TMS existence and strength

When examining the TMS scale (Lewis, 2003), we found that best friends have strong TMSs (M = 4.01 out of 5), supporting Hypothesis 1 (see Table 1 for correlations, means, and standard deviations for all variables). This was significantly greater than the midpoint of the scale, t(215) = 26.16, p < .001,⁴ and greater than means from the original Lewis (2003) studies, ts > 7.73, ps < .001. Thus, TMSs exist in best friendships and seem particularly strong.

Table 1.
Correlation matrix for variables in Study 1.

Variables 1 2 3 4 5 6 7 8

1. TMS strength

2. Behaviors .35

3. Friendship quality .49 .48*

4. Friendship length .15* −.02 .17*

5. Time spent −.11 −.06 −.02 −.20

6. Gender comp .03 .07 .01 .13^† .06

7. IOS .12^† .31 .45** .04 .14* −.02

8. Trust .59 .50 .55 .12^† −.11 .12^† .28

Mean 4.01 3.92 7.40 7.63 10.52 .85 4.66 5.60

SD .57 .65 1.36 6.28 15.42 .36 1.50 .99

Note. Gender comp is gender composition (coded such that 0 = mixed-gender friendships and 1 = same-gender friendships); IOS indicates inclusion of the other in the self; SD: standard deviation; TMS: transactive memory systems.

^† p < .10. *p < .05. **p < .01.

In addition, we predicted that reported TMS behaviors would be correlated with the Lewis (2003) measure of TMS strength, and we tested this hypothesis with a bivariate correlation between TMS strength and TMS behaviors. Reported TMS behaviors were correlated with TMS strength, supporting Hypothesis 2 (see Table 1 for results).

Friendship quality and TMSs

We predicted that TMS strength would be associated with friendship quality and tested this hypothesis with a bivariate correlation between TMS strength and friendship quality. People who reported stronger TMSs reported higher friendship quality, supporting Hypothesis 3 (see Table 1 for results).⁵

Friendship characteristics and TMS strength

The results of Hypotheses 4–8 are summarized in Table 1. We predicted that various friendship characteristics would be related to TMS strength. To test these hypotheses, we conducted bivariate correlation tests with each friendship characteristic (friendship length, time spent together, gender composition, IOS, and trust) and with TMS strength. Consistent with Hypothesis 4, the longer people were friends, the stronger their TMS was. Neither spending more time together nor being same-gender best friends were significantly associated with TMS strength, providing no support for Hypothesis 5 or 6. We found support for Hypotheses 7 and 8, showing that TMS strength was positively associated with IOS (marginally) and trust (significantly).⁶

Discussion

Study 1 demonstrated that TMSs exist in best friendships. Behaviors associated with TMSs were associated with self-reports of TMS strength, assessed by the adapted Lewis (2003) scale, suggesting this is an appropriate measure of TMSs in friendships. Results also showed that TMS strength is positively associated with friendship quality, friendship length, IOS, and trust. This replicates previous findings from work groups regarding length and trust, although it does not replicate previous findings regarding time spent together and TMS strength. IOS was only marginally related to TMS strength, but we continue to explore this variable in Study 2 with a larger sample.

Given previous research suggesting that same-gender friendships are closer and more supportive than opposite-gender friendships (Hand & Furman, 2009; Reisman, 1990), characteristics that should lead to stronger TMSs, one surprising finding is that gender composition was not associated with TMS strength. The number of mixed-gender best friendships was much lower than the number of same-gender best friendships (33 vs. 181), reducing our power to detect differences. However, this accurately reflects the reality that most friendships are between people of the same gender (Reeder, 2003). The next study further explores the role of gender composition in friendship TMSs by also looking at the TMS structure.⁷

Study 2

Study 2 extends Study 1 by investigating not only TMS strength but also TMS structure. Wegner and colleagues (1985) distinguish between differentiated and integrated TMS structures. Differentiated TMSs occur when people hold different knowledge (e.g., one friend knows a lot about wine and the other about beer), whereas integrated TMSs occur when people hold similar knowledge (e.g., both friends know a lot about wine and beer). Integrated TMSs are better than differentiated systems for error correcting, helping, and collaboration (Gupta & Hollingshead, 2010). However, differentiated TMSs offer the benefit of more collective knowledge.

Previous laboratory research has investigated TMS structure (Gupta & Hollingshead, 2010), but to our knowledge, these structures have not been explored in close relationships. Thus, exploring factors associated with different TMS structures in close relationships is a novel contribution of our research. Knowing what types of friendships are likely to be associated with different types of TMS structures gives an initial understanding of the potential benefits of TMSs (e.g., sharing diverse information vs. correcting mistakes) specific friendships may afford. Because these structures have not been examined in existing relationships before, we developed a new measure to assess TMS structures. We predict that differentiated and integrated TMS structures may be associated with different friendship characteristics.

Gender composition of friendships may be related to TMS structures. Stereotypically, mixed-gender friendships may have more differentiated TMSs than same-gender friendships. When participants anticipated working with someone of the opposite gender rather than the same gender on a recall task, they were more likely to learn gender-consistent information (Hollingshead & Fraidin, 2003). Additionally, women and men are socialized to participate in different activities and occupy different roles (Eagly & Steffen, 1984), potentially leading to knowledge differences. Thus, we predicted that participants in mixed-gender best friendships will report more differentiated TMS structures compared to participants in same-gender best friendships (Hypothesis 9).

IOS may also be associated with TMS structure. The more people include another in their self-concept, the more they may accept and acquire the other’s knowledge as their own. Thus, people who have a higher IOS with their friend should report more integrated TMS structures (Hypothesis 10).

A final goal of Study 2 was to supplement the previous study using a particularly large sample, ensuring adequate power to test our hypotheses. Although we reduced the number of measures to keep the survey a reasonable length, we tested hypotheses from Study 1 where we could (Hypotheses 1 and 3–7).

Method

Participants

Participants (n = 382) were recruited online and through snowball sampling via the same methods as Study 1. They received no compensation. Participants below 18 years old were removed (n = 12). We also excluded data from later instances of IP addresses (n = 30; Gosling et al., 2004).

This resulted in 340 participants (233 females). Participants’ ages ranged from 18 to 64 (median = 21, IQR = 8). Participants were mostly Caucasian (57.4%), others were African American (8.2%), Hispanic/Latino (8.2%), Asian/Pacific Islander (4.7%), other (7.0%), and 13.5% omitted. Participants were mostly from the United States (74.7%), although 5.3% were from Europe, 6.4% from other places (e.g., Canada and Australia), and 13.6% omitted. Friendship length varied from under a year to 49 years (median = 7.00, IQR = 10). Sixty-seven friendships were mixed-gender (20%).

Procedure

Participants completed an Internet survey. Participants were told to think about a best friend who was not a romantic partner to answer the questions.

Participants responded to the same demographics, TMS strength (α = .85), friendship quality (α = .84), friendship length, time spent together, and IOS items as in Study 1.

We also implemented our new TMS structure measure by asking participants to imagine they were entering a trivia contest with their best friend. They were shown different areas of expertise and asked to indicate whether they, their best friend, both of them, or neither of them would specialize in these areas. These questions assessed specific content areas of expertise (history, geography, politics, music, movies and television, literature, sports, food and drinks, trends and celebrities, and science) developed based on categories likely to appear in typical trivia games.⁸ Specifically, they were asked an item such as “Would you or your best friend specialize in information about music?” and could select “me and my best friend,” “me,” “my best friend,” or “neither of us.”

Responses were aggregated to create a measure of differentiated versus integrated TMSs. Indicating that one friend would be responsible for a specific area of expertise was considered a differentiated TMS. Indicating that both friends would be responsible for a specific area of expertise was considered an integrated TMS. The number of responses of each were recorded, and a difference score was calculated with the sum of differentiated items being subtracted from the sum of integrated items. Thus, the more negative the number, the more differentiated the TMS. Indicating that neither would be responsible for a certain area of expertise was not considered part of either system, as a lack of knowledge is not representative of a memory system.

Results

Replication of Study 1 hypotheses

We found further support for Hypothesis 1 (see Table 2 for correlations, means, and standard deviations of all variables). As in Study 1, the TMS strength mean (M = 4.07 out of 5) was significantly greater than the midpoint of the scale, t(340) = 42.65, p < .001. The results of Hypotheses 3–7 are summarized in Table 2. To examine Hypotheses 3–7, we ran bivariate correlations. Again, TMS strength was positively associated with friendship quality (Hypothesis 3).⁹ Looking at friendship characteristics, friendship length was positively associated with TMS strength (Hypothesis 4) and the amount of time spent together was not significantly associated with TMS strength (Hypothesis 5), replicating Study 1. As in Study 1, gender composition was not significantly related to TMS strength (Hypothesis 6) and IOS was positively associated with TMS strength (Hypothesis 7).¹⁰

Table 2.
Correlation matrix for variables in Study 2.

Variables 1 2 3 4 5 6 7

1. TMS strength

2. Friendship quality .48

3. Friendship length .15 .08

4. Time spent −.06 .12* −.19**

5. Gender comp .09 −.01 .13* −.09

6. IOS .20 .45 −.09 .22** −.09

7. Structure .09 .06 −.12* .07 .11* .13*

Mean 4.07 7.56 8.97 13.00 .80 4.62 −3.65

SD .46 1.39 7.81 19.38 .40 1.48 4.17

Note. Gender comp is gender composition (coded such that 0 = mixed-gender friendships and 1 = same-gender friendships); IOS indicates inclusion of the other in the self. SD: standard deviation; TMS: transactive memory systems.

^† p < .10. *p < .05. **p < .01.

Friendship characteristics and TMS structure

Hypotheses 9 and 10 concerned the integrated versus differentiated TMS structures. To test Hypothesis 9, we conducted a t-test with gender composition as a categorical predictor variable and TMS structure as the outcome variable. Participants with mixed-gender friendships reported more differentiated TMS structures (M = −4.59, SD = 4.35) than same-gender friendships (M = −3.40, SD = 4.11), t(301) = −2.00, p = .047, d = .28, providing support for Hypothesis 9. To test Hypothesis 10, we conducted a bivariate correlation with IOS and TMS structure. Higher IOS was associated with more integrated TMS structure, supporting Hypothesis 10 (see Table 2 for results).

Discussion

Study 2 expanded on Study 1 by investigating the additional component of TMS structure using a newly developed measure. We found that mixed-gender friendships have more differentiated TMS structures than same-gender friendships. This is particularly interesting considering that gender composition was not associated with TMS strength in either study. That is, both same- and mixed-gender friendships have equally strong TMSs, but mixed-gender friendships hold more diverse knowledge. In addition, friendships with higher IOS have more integrated structures, suggesting that interdependence may lead to similar knowledge stores.

Additionally, Study 2 results replicated Study 1 results. TMS strength was again related to friendship quality, friendship length, and IOS, consistent with Study 1. Although IOS was only marginally significantly related to TMS strength in Study 1, it was a significant association in Study 2. This study again replicates findings from work groups, showing an association between TMS strength and length (Lewis, 2004) but not findings showing an association between time spent together and TMS strength (Jackson & Moreland, 2009).

General discussion

Overall, the current work provides an initial investigation into TMSs in best friendships. Although it has been assumed that TMSs exist in friendships (Wegner, 1986), this research is the first to our knowledge to actually test and support this assumption. Friendships are critical to well-being and life satisfaction (Siebert et al., 1999; Winefield et al., 1992), and specifically, friendship quality is important to physical and psychological health (Demir et al., 2007; Sabin, 1993; Waldrip et al., 2008). Thus, it is crucial to examine the factors related to friendship maintenance and quality. This work provides insight into one such factor: TMSs.

In addition to being related to friendship quality, TMSs may result in multiple advantages for friendships, such as facilitating coordination within the friendship and providing more knowledge benefits to each individual. Thus, determining the friendship characteristics associated with stronger TMSs provides information regarding which types of friendships likely afford higher friendship quality and these additional advantages (e.g., friendships higher in trust, friendships that have existed longer, etc.). We also gain insight into the factors associated with the development of TMSs within friendships and possibly other interpersonal relationships. We can also compare TMSs in friendships versus work groups, based on previous findings.

Importantly, behaviors representative of TMS processes correlated with TMS strength, as measured by our adapted Lewis (2003) scale, suggesting it is a good measure of TMSs in best friendships. Because we used both student and broader Internet-based samples, with same-gender and mixed-gender friendships, the scale seems to be a good measure of TMSs in best friendships for different types of friendships and people. Thus, the Lewis (2003) scale can be adapted to measure TMSs in close relationships in a less resource-intensive and more flexible manner. Being able to utilize this scale to assess TMSs in close relationships creates opportunities to further study TMSs in the relationships field.

Stronger TMSs were related to higher friendship quality. High friendship quality is key to well-being, happiness, and physical health (Demir et al., 2007; Sabin, 1993; Waldrip et al., 2008), making factors associated with friendship quality, like TMSs, essential to identify. As strong TMSs are positively associated with higher friendship quality, these systems could be a critical aspect of friendships. Indeed, TMS strength predicted friendship quality above and beyond the effects of all the other variables we studied, making TMS strength a strong predictor of friendship quality.

Our results demonstrated that participants who were friends longer, had higher IOS (interdependence) with their best friend, and had more trust in their friendship, had stronger TMSs. The findings regarding friendship length and trust supported previous research and theory from TMSs in work groups (Lewis, 2004; Robertson et al., 2013). However, there was no relationship between how much time friends spend together and their TMS strength. This was surprising as past research found that face-to-face communication leads to stronger TMSs in classroom work groups (Jackson & Moreland, 2009). This suggests that different processes might operate in the development of TMSs in best friendships, such as interdependence or friendship length. In work groups where people typically interact in a limited context, time spent together might be more important to developing strong TMSs. On the other hand, in best friendships, people interact in a variety of contexts and communicate frequently using various methods (e.g., texting and e-mail), potentially making the amount of time best friends spend together a less crucial predictor of TMS strength.

TMS structure also differed based on friendship characteristics. Although TMS strength did not differ by gender composition, TMS structure did. Mixed-gender best friends had more differentiated TMS structures than same-gender best friends. This suggests that mixed-gender friendships may contain more total knowledge than same-gender friendships, although people in both types of friendships may rely on and know each other’s knowledge. Previous research suggests that same-gender friendships should allow for more disclosure and closeness (Hand & Furman, 2009), which we hypothesized should lead to stronger TMSs. However, there may be fewer differences on these factors when the mixed-gender friendship is a best friendship. In fact, in our studies, we find no significant differences between mixed-gender and same-gender best friends in IOS or friendship quality, ps > .10. Thus, at least in our sample, there may be few differences between mixed-gender and same-gender best friends, leading to equally strong TMSs. These findings suggest that having a best friend of the opposite gender may have benefits when it comes to collective knowledge, as these friendships are composed of similar levels of TMS strength and more differentiated TMS structures. Additionally, friendships with higher IOS had more integrated TMS structures. This suggests that the more interdependent the friends are, the more similar knowledge they have. Similar knowledge stores could develop over time in highly interdependent people who include the other in their self-concept.

In sum, these studies are not only the first to demonstrate the existence of TMSs in friendships but also contribute to our knowledge of close relationships and TMSs more broadly by providing insight into factors associated with TMS strength and structure, including friendship quality.

Strengths and limitations

The present research has multiple strengths. It extends our knowledge of TMSs to a domain where it has not been explored and answers a call for focusing on nonromantic close relationships (Berscheid & Regan, 2005). Additionally, as the reported TMS behaviors were positively associated with TMS strength, these studies show that the TMS scale (Lewis, 2003) can be successfully adapted to study TMSs in interpersonal relationships.

Few studies have looked at differentiated versus integrated structures, and no research, to our knowledge, has examined TMS structures in close relationships. Our trivia game measure of TMS structure is a novel way of measuring whether dyads possess differentiated versus integrated structures. This measure could be used to look at other close relationships and could even be adapted to workplaces, making it easier to assess TMS structure in the field and facilitating future research.

Finally, although research has demonstrated the existence of TMSs in romantic relationships, the associations between specific close relationship characteristics and TMSs have not been explored. Thus, the current work provides novel evidence of factors related to strong TMSs in close relationships. Research should further explore other characteristics such as the amount of non-face-to-face communication that friends engage in, especially given the increase in communication via technology.

Our method also has its weaknesses. Due to the correlational nature of our studies, we are unable to determine the direction of our effects. For example, it is impossible to tell whether TMS strength leads to increased friendship quality or whether high friendship quality leads to the development of stronger TMSs. Having a strong TMS in a best friendship should make those friendships more satisfying—these friends would be able to access more information, share information more easily, and remember more themselves when their friend cues them. Additionally, taking the time to learn what your friend knows and potentially developing unique knowledge from them would be a large investment that should lead to more commitment, according to Rusbult and colleagues’ (1998) investment model. However, it could also be that friends must be highly committed to, and/or highly satisfied with, their friendship to take the time to develop strong TMSs. It is likely a cyclical process wherein friendship quality may lead to stronger TMSs which in turn lead to higher friendship quality. Conducting a longitudinal study or studying friendships in the initial stages versus people who are already best friends would shed light on directionality.

We also cannot rule out potential third variables that might lead to increases in TMS strength and friendship quality, such as trust. As we demonstrated, greater trust is associated with stronger TMSs, and trust is related to friendship quality as well. However, results from supplementary, exploratory multiple regressions⁵ ^, ⁹ suggest TMS strength influences friendship quality above and beyond the other variables we measured, including trust.

In both studies, we only have responses of one member of the friendship dyad. However, supplemental evidence⁷ suggests that best friends report similar levels of friendship characteristics (friendship length, amount of time spent together, IOS, and trust), TMS strength, and friendship quality. Thus, we can assume that these reports would be similar in these studies.

It is also crucial to note that we only conducted analyses based on a priori predictions in this initial investigation of transactive memory processes in friendships. Many other interesting effects involving interactions and mediations would be a fruitful avenue for future research.

Future directions

There are many potential future directions that could explore TMSs in friendships and close relationships more generally. Future research should extend the findings of the current research to explore the development of TMSs and structures. For example, how long does it take for people to develop TMSs? Do people purposefully develop more differentiated TMS structures over time in their close relationships? What factors might lead people to develop more differentiated TMS structures in their close relationships?

Certain individual differences or relationship characteristics may relate to how quickly TMSs develop, how strong they are, and what type of TMS structure is likely to develop. Attachment style might be one such variable. People with secure attachment styles may follow a normal trajectory of developing a TMS and develop typically strong TMSs. However, those who are higher in avoidance may delay developing a TMS with close others, due to a lack of trust, and even over time the TMSs they do develop may be weaker. Additionally, those higher in anxiety may hasten their TMS development, potentially in attempts to make the relationship more dependent. Thus, their TMSs may be equally strong as those with secure attachment styles. Similarly, in looking at TMS structure, those who are more avoidant may prefer to maintain more integrated TMS structures, not wanting to rely on others for information, but those who are more anxious may prefer to maintain more differentiated TMS structures to potentially increase dependence. Looking at relationship-specific characteristics, similarity may play an important role in the development of TMSs and structures. For example, people who are more similar should have an easier time coordinating with each other, potentially leading to quicker development of TMSs, and stronger TMSs. People who are similar may also be more likely to develop integrated TMSs and thus may have to make more of an effort to differentiate their TMSs to hold more collective knowledge.

Future research may also investigate different types of memory. For example, episodic memory is memory for autobiographical events, and semantic memory is memory for general knowledge (Tulving, 1972). Although we focused on semantic memory (e.g., trivia game categories), future research may benefit from investigating episodic memory (e.g., knowledge of how you became friends). Episodic memories may be particularly important for friendship quality, but semantic memories may be particularly important for performance benefits or developing more collective knowledge (i.e., more differentiated structures).

Conclusions

In conclusion, the present studies provide initial evidence that TMSs exist in best friendships. Additionally, the strength of these systems is related to friendship quality and friendship characteristics (e.g., friendship length). Thus, having strong TMSs may contribute to friendship maintenance and friendship quality, an important factor in both psychological and physical health. There is also evidence that two different structures may develop depending on certain friendship characteristics (e.g., gender composition). Our measure of structure is novel and facilitates the measurement of TMS structures outside the lab. Exploring TMSs and structures advances both the transactive memory and the friendship literatures, especially as TMS strength is positively associated with friendship quality.

Variables	1	2	3	4	5	6	7	8
1. TMS strength
2. Behaviors	.35**
3. Friendship quality	.49**	.48*
4. Friendship length	.15*	−.02	.17*
5. Time spent	−.11	−.06	−.02	−.20**
6. Gender comp	.03	.07	.01	.13^†	.06
7. IOS	.12^†	.31**	.45**	.04	.14*	−.02
8. Trust	.59**	.50**	.55**	.12^†	−.11	.12^†	.28**
Mean	4.01	3.92	7.40	7.63	10.52	.85	4.66	5.60
SD	.57	.65	1.36	6.28	15.42	.36	1.50	.99

Variables	1	2	3	4	5	6	7
1. TMS strength
2. Friendship quality	.48**
3. Friendship length	.15**	.08
4. Time spent	−.06	.12*	−.19**
5. Gender comp	.09	−.01	.13*	−.09
6. IOS	.20**	.45**	−.09	.22**	−.09
7. Structure	.09	.06	−.12*	.07	.11*	.13*
Mean	4.07	7.56	8.97	13.00	.80	4.62	−3.65
SD	.46	1.39	7.81	19.38	.40	1.48	4.17

Footnotes

Appendix 1 Authors’ Note

Portions of this research were presented at the 2014 meeting of the Midwestern Psychological Association.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Aron

E. N.

(1986). Love as the expansion of self: Understanding attraction and satisfaction. New York, NY: Hemisphere.

Aron

E. N.

Smollan

(1992). Inclusion of other in the self scale and the structure of interpersonal closeness. Journal of Personality and Social Psychology, 63, 596–612. doi:10.1037/0022-3514.63.4.596

Aron

Ketay

Riela

Aron

E. N.

(2007). How close others construct and reconstruct who we are and how we feel about ourselves. In Wood

Tesser

Holmes

(Eds.), The self and social relationships (pp. 209–229). Philadelphia, PA: Psychology Press.

Berscheid

Regan

(2005). The psychology of interpersonal relationships. Upper Saddle River, NJ: Pearson Education Inc.

Carbery

Buhrmester

(1998). Friendship and need fulfillment during three phases of young adulthood. Journal of Social and Personal Relationships, 15, 393–409. doi:10.1177/0265407598153005

Cohen

West

S. G.

Aiken

L. S.

(2003). Applied multiple regression/correlation analysis in the behavioral sciences (3rd ed.). Mahwah, NJ: Erlbaum.

Demir

Ozdemir

Weitekamp

L. A.

(2007). Looking to happy tomorrows with friends: Best and close friendships as they predict happiness. Journal of Happiness Studies, 8, 243–271. doi:10.1007/s10902-006-9025-2

Eagly

A. H.

Steffen

V. J.

(1984). Gender stereotypes stem from the distribution of women and men into social roles. Journal of Personality and Social Psychology, 46, 735–754. doi:10.1037/0022-3514.46.4.735

Fehr

(2008). Friendship formation. In Sprecher

(Ed.), Handbook of relationship initiation (pp. 29–54). New York, NY: Psychology Press.

10.

Gosling

S. D.

Vazire

Srivastava

John

O. P.

(2004). Should we trust Web-based studies? A comparative analysis of six preconceptions about Internet questionnaires. American Psychologist, 59, 93–104. doi:10.1037/0003-066X.59.2.93

11.

Gupta

Hollingshead

A. B.

(2010). Differentiated versus integrated transactive memory effectiveness: It depends on the task. Group Dynamics: Theory, Research, and Practice, 14, 384–398. doi:10.1037/a0019992

12.

Hand

L. S.

Furman

(2009). Rewards and costs in adolescent other-sex friendships: Comparisons to same-sex friendships and romantic relationships. Social Development, 18, 270–287. doi:10.1111/j.1467-9507.2008.00499.x

13.

Hollingshead

A. B.

(1998). Retrieval processes in transactive memory. Journal of Personality and Social Psychology, 74, 659–671. doi:10.1037/0022-3514.74.3.659

14.

Hollingshead

A. B.

Fraidin

S. N.

(2003). Gender stereotypes and assumptions about expertise in transactive memory. Journal of Experimental Social Psychology, 39, 355–363. doi:10.1016/S0022-1031(02)00549-8

15.

Jackson

Moreland

R. L.

(2009). Transactive memory in the classroom. Small Group Research, 40, 508–534. doi:10.1177/1046496409340703

16.

Lewis

(2003). Measuring transactive memory systems in the field: Scale development and validation. Journal of Applied Psychology, 88, 587–604. doi:10.1037/0021-9010.88.4.587

17.

Lewis

(2004). Knowledge and performance in knowledge-worker teams: A longitudinal study of transactive memory systems. Management Science, 50, 1519–1533. doi:10.1287/mnsc.1040.0257

18.

Lewis

Herndon

(2011). Transactive memory systems: Current issues and future research directions. Organization Science, 22, 1254–1265. doi:10.1287/orsc.1110.0647

19.

Liang

D. W.

Moreland

R. L.

Argote

(1995). Group versus individual training and group performance: The mediating role of transactive memory. Personality and Social Psychology Bulletin, 21, 384–393. doi:10.1177/0146167295214009

20.

Pulakos

(1989). Young adult relationships: Siblings and friends. Journal of Psychology, 123, 237–244. doi:10.1080/00223980.1989.10542978

21.

Reeder

H. M.

(2003). The effect of gender role orientation on same- and cross-sex friendship formation. Sex Roles, 49, 143–152. doi:10.1023/A:1024408913880

22.

Reisman

J. M.

(1990). Intimacy in same-sex friendships. Sex Roles, 23, 65–82. doi:10.1007/BF00289880

23.

Rempel

J. K.

Holmes

J. G.

Zanna

M. P.

(1985). Trust in close relationships. Journal of Personality and Social Psychology, 49, 95–112. doi:10.1037/0022-3514.49.1.95

24.

Robertson

Gockel

Brauner

(2013). Trust your teammates or bosses? Differential effects of trust on transactive memory, job satisfaction, and performance. Employee Relations, 35, 222–242. doi:10.1108/01425451311287880

25.

Rusbult

C. E.

Martz

J. M.

Agnew

C. R.

(1998). The investment model scale: Measuring commitment level, satisfaction level, quality of alternatives, and investment size. Personal Relationships, 5, 357–391. doi:10.1111/j.1475-6811.1998.tb00177.x

26.

Sabin

E. P.

(1993). Social relationships and mortality among the elderly. Journal of Applied Gerontology, 12, 44–60. doi:10.1177/073346489301200105

27.

Siebert

D. C.

Mutran

E. J.

Reitzes

D. C.

(1999). Friendship and social support: The importance of role identity to aging adults. Social Work, 44, 522–532. doi:10.1093/sw/44.6.522

28.

Tulving

(1972). Episodic and semantic memory. In Tulving

Donaldson

(Eds.), Origins of Memory (pp. 381–403). San Diego, CA: Academic.

29.

Vanderdrift

L. E.

Lehmiller

J. J.

Kelly

J. R.

(2012). Commitment in friends with benefits relationships: Implications for relational and safer sex outcomes. Personal Relationships, 19, 1–13. doi:10.1080/00224491003721694

30.

Waldrip

A. H.

Malcolm

K. T.

Jensen-Campbell

L. A.

(2008). With a little help from your friends: The importance of peer relationships on adolescent adjustment. Social Development, 17, 832–852. doi:10.1111/j.1467-9507.2008.00476.x

31.

Wegner

D. W.

(1986). Transactive memory: A contemporary analysis of the group mind. In Mullen

Goethals

G. R.

(Eds.), Theories of group behavior (pp. 185–208). New York, NY: Springer-Verlag.

32.

Wegner

D. M.

Erber

Raymond

(1991). Transactive memory in close relationships. Journal of Personality and Social Psychology, 61, 923–929. doi:10.1037/0022-3514.61.6.923

33.

Wegner

D. M.

Giuliano

Hertel

(1985). Cognitive interdependence in close relationships. In Ickes

W. J.

(Ed.), Compatible and incompatible relationships (pp. 253–276). New York, NY: Springer-Verlag.

34.

Winefield

H. R.

Winefield

H. R.

Tiggemann

(1992). Social support and psychological well-being in young adults: The multi-dimensional support scale. Journal of Personality Assessment, 58, 198–210. doi:10.1207/s15327752jpa5801_17

35.

Yager

(1997). Friendshifts: The power of friendship and how it shapes our lives. Stamford, CT: Hannacroix Creek Books.