It''s Not Just Who You Are, It''s How You Act Influences of Global and Contextualized Personality Traits on Relationship Satisfaction

In the Study 1 we collected global being and contextualized acting self reports of personality and relationship satisfaction from a large, diverse sample of adults in committed romantic

Running head: CONTEXTUALIZED PERSONALITY AND RELATIONSHIPS

It's Not Just Who You Are, It's How You Act:

Influences of Global and Contextualized Personality Traits on Relationship Satisfaction

Richard B Slatcher and Simine VazireThe University of Texas at Austin

Address correspondence to:

Richard B Slatcher

Department of Psychology #A8000

108 E Dean Keaton Street

University of Texas

Austin, TX 78712

Telephone: (512) 471-6852; FAX (512) 471-5935

Email: slatcher@mail.utexas.edu

Abstract How does personality exert its influence on relationship satisfaction? Previous research has shown, for example, that Agreeableness is associated with greater relationship satisfaction, yet little is known about the mechanisms through which personality affects satisfaction We propose

that global personality traits (e.g., being agreeable) exert their influence on relationships through contextualized manifestations of personality (e.g., acting agreeable towards one's partner) In the

Study 1 we collected global (being) and contextualized (acting) self reports of personality and relationship satisfaction from a large, diverse sample of adults in committed romantic

relationships In Study 2 we collected global self reports of personality (being) and relationship satisfaction from undergraduate dating couples We also collected couples’ Instant Messages (IMs) to each other for seven days Independent observers read the IMs and rated each couple member’s personality in the context of their relationship (acting) The results showed that

contextualized personality (acting) predicted relationship satisfaction above and beyond global personality (being), and that acting mediated the relationship between being and relationship satisfaction Our findings point to the importance of examining both global and contextualized personality traits, and demonstrate how personality influences major life outcomes

It's Not Just Who You Are, It's How You Act:

Influences of Global and Contextualized Personality Traits on Relationship SatisfactionAmy and Karen are discussing Amy’s relationship with her boyfriend, David “I don’t understand why you’re with him,” Karen tells Amy, “he’s such a jerk.” “I know what you mean,”replies Amy, “but he’s so different when we’re alone.” Do people really behave differently in romantic relationships than they do in other contexts? If so, what predicts the quality of their relationship—how they are in general, or how they are with their partner?

Researchers have long been interested in the effects of personality on romantic

relationships, with roughly 500 studies dating back to the 1930s published on this topic (Cooper

& Sheldon, 2002) The findings from these studies have demonstrated that certain stable

personality factors are associated with relationship quality (Eysenck & Wakefield, 1981; Karney

& Bradbury, 1995; Terman & Buttenwieser, 1935) However, the associations found between personality and relationship quality often have been modest and inconsistent One potential explanation for the inconsistent findings in these studies is that the measures of personality employed have been too general Researchers largely have ignored relationship-specific

manifestations of personality (i.e., what a person’s personality is like within the context of a particular relationship), which may be important in predicting relationship functioning

While there is no doubt that enduring, stable personality traits influence how people approach and view their relationships, examining the role of contextualized personality is vitally important as well Indeed, a number of scholars in our field have called for a more contextualizedapproach to the study of personality and relationships (e.g., Reis, Capobianco & Tsai, 2002; McAdams, 1995) With this article, we extend previous research on the role of personality dispositions by examining the role of both global and contextualized personality traits in

romantic relationships Is David really a nice guy when he is with Amy, even though he is a jerk

to everyone else? And if so, is the quality of their relationship better predicted by his positive behavior towards her or by his global, negative personality attributes?

The vast majority of relationship-personality studies have examined the association between global personality traits and relationship satisfaction, focusing particularly on the traits

of the Five Factor Model (FFM; McCrae & Costa, 1999)—Extraversion, Agreeableness,

Conscientiousness, Neuroticism, and Openness to Experience In the context of relationships, Neuroticism has been the most extensively researched of these traits (Bouchard, Lussier, & Sabourin, 1999; Caughlin, Huston, & Houts, 2000; Karney & Bradbury, 1995; Terman &

Buttenwieser, 1935) Those who are high in Neuroticism—anxious, irritable, and emotionally unstable—typically report being less satisfied in their romantic relationships than those who are low in Neuroticism Much less is known about how the other four factors of the FFM relate to romantic relationship quality, but preliminary findings indicate that Agreeableness,

Conscientiousness, Extraversion, and Openness to Experience are all positively related to

relationship satisfaction (Botwin, Buss, & Shackelford, 1997; Donnellan, Conger, & Bryant, 2004; Watson, Hubbard, & Wiese, 2000)

Global measures of personality are predictive of satisfaction not only in romantic

relationships but in other domains of life as well For example, the more agreeable people are on

average, the more satisfied they will be across relationships—with family, friends, and so on

(Branje, van Lieshout, & van Aken, 2004; Graziano, Campbell, & Hair, 1996; Campbell, Gleason, Adams, & Malcolm, 2003) Conversely, the more neurotic people are, the less satisfied they likely will be with the various relationships in their lives (Berry, Willingham,

Jensen-& Thayer, 2000; Cheng Jensen-& Furnham, 2002; Eaker Jensen-& Walters, 2002) But while global measures

of personality can tell us a little bit about how a person is in many types of relationships, they fail

to tell us a great deal about how a person is any one particular relationship For example,

Asendorpf and Wilpers (1998) showed that general interpersonal traits—including Extraversion, sociability, and shyness—predict general patterns of social behavior but are only weakly

associated with the qualities of specific relationships Further, trust for a particular partner, but not generalized trust, predicts commitment and well-being in that relationship (Wieselquist, Rusbult, Foster, & Agnew, 1999) By supplementing global measures of personality with

contextualized ones, we may be able to better disentangle the role of personality in relationships

The distinction between global and contextualized measures has been widely used in life satisfaction research For example, Heller, Watson, and Ilies (2004) have shown that global life satisfaction is distinct from (though related to) context-specific satisfaction, such as work or relationship satisfaction It is widely accepted that satisfaction measures should be obtained at the level of analysis of interest to the researcher For example, if researchers are interested in predicting relationship outcomes, they should measure relationship-specific satisfaction The same logic applies to measures of personality In this article, we examine whether contextualized measures of personality in the domain of romantic relationships can predict relationship

functioning better than global measures of personality

Many studies have demonstrated the benefit of contextualized measures of personality For example, in one study (Hunthausen, Truxillo, Bauer, & Hammer, 2003), customer service supervisors at a large U.S airline completed a modified version of the NEO Five-Factor

Inventory (NEO-FFI; Costa & McCrae, 1992) in which they were asked what their personalities were like “at work”; additionally they completed the standard global NEO-FFI Participants’ self-ratings of Extraversion and Openness on the at-work measure predicted job performance, while

the global FFM measure did not In another study (Schmit, Ryan, Stierwalt, & Powell, 1995), college students’ self-ratings of Conscientiousness “at school” predicted students’ GPAs better than global measures of Conscientiousness Similarly, knowing what people’s personalities are like in the context of their romantic relationship should predict relationship quality better than global measures of personality For example, knowing how agreeable a person is with his or her romantic partner should provide unique predictive power about the quality of that person’s relationship above and beyond how agreeable that person is in general

In describing how people are in general and how they are in the context of their romantic

relationships, we borrow the terms being and acting from Fleeson’s density distribution model of

personality (Fleeson, 2001; Fleeson, Malanos, & Achille, 2002) Individual differences in

personality are typically thought of as global traits, such as Agreeableness This generalized

Agreeableness may be conceptualized as being agreeable Fleeson’s model suggests that there are also variations within persons across situations in levels of individual traits In this model, the

extent to which a person’s personality trait is manifested in a particular relationship may be

conceptualized as acting (e.g., acting agreeable) For example, Amy’s boyfriend David may be

disagreeable at the global level—across time, situations and relationships Although this is a good predictor of relationship satisfaction (Watson et al., 2000), we would predict that David’s level of Agreeableness in his romantic relationship (i.e., how agreeable he acts with Amy) is an even better predictor of relationship satisfaction

Naturally, contextualized personality tendencies are not completely independent from global dispositions Global dispositions are likely to exert an influence on how personality is expressed in any given context As illustrated in Figure 1, we view personality as a hierarchicallyorganized system with global dispositions at the top influencing contextualized personality

tendencies below In our example, David’s Agreeableness when he is with Amy (moderate) is probably influenced in part by his overall level of Agreeableness (low) as well as contextualized factors (e.g., Amy’s kindness toward him) If this hierarchical organization of personality is true,

we would expect contextualized personality to mediate the relationship between global

personality and contextualized outcomes (shown at the bottom of Figure 1) This article will examine the dynamic relationship between global and contextualized levels of personality and relationship satisfaction

Aims of our Research

The primary aim of this article is to examine the role of global and contextualized

personality in the association between the FFM personality traits and romantic relationship satisfaction We explore the extent to which relationship satisfaction is associated with

personality traits in the specific context of a romantic relationship (acting) in comparison with global, decontextualized traits (being) We also investigate whether acting may mediate the association between being and relationship satisfaction, as suggested by a hierarchical

organization of personality traits from global to context-specific

As we have described, previous research has focused almost exclusively on the role of global personality traits in romantic relationships Thus, the major contribution of our work is to examine whether contextualized personality predicts relationship outcomes better than does global personality However, our research also extends previous research in other important aspects Research on the role of personality in romantic relationships has traditionally relied on self reports from small samples of dating college students We improve on this design in

numerous ways First, we examine a large, diverse sample drawn from a non-college-student population (Study 1) Second, we include measures of personality and satisfaction from both

partners in each couple (Study 2) Finally, we conduct a controlled study that allows us to obtain

an objective, naturalistic measure of what people’s personalities are like in the specific context oftheir romantic relationships (Study 2)

Question 1: How strongly is acting in relationships associated with relationship

satisfaction? A long-standing belief in psychology is that a person’s attitudes and behaviors

are a function of both pre-existing attributes and situational context This process, first described by Lewin (1936) in his characterization of behavior as being a function of the person and the environment, and later articulated by contemporary theorists such as

Mischel and Shoda (1995), indicates that personality measures that take into account the context of a person’s behavior (acting) will yield stronger associations with outcome measures than will global measures of personality (being) We thus expected that the established association of personality with self and partner relationship satisfaction

(Donnellan et al 2004; Karney & Bradbury, 1995; Watson et al., 2000) would be

strengthened by taking into account relationship-specific personality traits (acting)

It is possible that contextualized measures will have greater predictive validity than global measures simply because they are narrower measures of personality, and not because they provide any unique insight into the domain of the relationship To rule out this possibility, we will examine whether romantic relationship-context measures (acting in romantic relationships) predict romantic relationship satisfaction better than do other contextualized measures (e.g., acting at work, acting with friends, etc.) If acting in romantic relationships uniquely predicts romantic relationship satisfaction, this would suggest that this contextualized measure of

personality is tapping into how people behave with their relationship partners, and that this behavior affects the quality of the relationship

Previous research has found that, among the FFM personality traits (McCrae & Costa, 1999), Agreeableness and Neuroticism are particularly strong predictors of relationship

satisfaction (Botwin et al., 1997; Donnellan et al., 2004; McCrae, Stone, Fagan, & Costa, 1998; Watson et al., 2000) Based on these findings, we predicted that acting agreeable would be positively associated with self and partner satisfaction and that acting neurotic would be

negatively associated with self and partner satisfaction We further predicted that acting would bemore strongly associated with relationship satisfaction than being for these traits Due to the lack

of conclusive findings in previous research, no specific predictions were made for the other FFMtraits

Question 2: Does acting mediate the relationship between being and relationship

satisfaction? One of the advantages of our approach is that it allows for the examination of

potential mechanisms underlying the relationship between personality traits and relationship satisfaction Specifically, we propose that global personality traits will affect relationship

satisfaction to the extent that these global traits are manifested in relationship-specific traits As with other contextual models (e.g., Bradbury and Fincham, 1988), ours incorporates both

proximal and distal factors into a common framework In our model, acting (proximal factor) is amechanism through which being (distal factor) influences relationship satisfaction We thus predicted that acting would mediate the relationship between being and relationship satisfaction, specifically for Agreeableness and Neuroticism

Design of the Studies

We examined these two research questions using a multi-method approach in two studies.The purpose of Study 1 was to test our research questions in a large sample of American adults incommitted dating relationships Participants from all over the U.S were directed to a website

where they completed a traditional FFM measure of personality (being), a modified measure of the FFM (acting) in which participants indicated how they act in the context of their romantic relationships, and a measure of satisfaction in their romantic relationships We also asked

participants to report how they act in various other contexts (with coworkers, friends, and family)

to rule out the possibility that simply any contextualized measure would predict romantic

relationship satisfaction better than a global measure

The purpose of Study 2 was to provide a more direct test of our research questions in a controlled setting Drawing on a sample of undergraduate dating couples, we obtained self ratings of being using a global FFM measure, and both self and partner ratings of relationship satisfaction, allowing us to examine the effects of personality on both one’s own and one’s partner’s level of satisfaction We then obtained objective measures of acting by directly

observing how people act with their relationship partners To do this, we recruited couples who use Instant Messaging (IM) as a daily form of communication With their consent, we recorded

all of their IM conversations over seven days An important aspect of IM is that it allows

researchers to subtly and unobtrusively study close relationships in their natural settings This new technology complements existing naturalistic methods such as daily diaries (Bolger, Davis,

& Rafaeli, 2003; Drigotas, Whitney, & Rusbult, 1995; Reis, 1994) in which couples’ interaction patterns may be studied on a day-to-day basis IM conversations can serve as windows into real-world dyadic interactions and allow researchers to examine links between behavioral

manifestations of personality and relationship functioning

What is the best way to obtain an objective measure of acting in the context of a romanticrelationship? One way is to expose independent observers to couple members’ behaviors

exclusively in the context of their relationship, and ask them to rate how the couple members act

towards one another We did this by showing the IM conversations to a team of trained observerswho completed an acting measure of the FFM for each couple member This technique allowed

us to extract objective information about how people act towards their romantic partners,

independent of their global self-views

Study 1: Online Questionnaire Study

Method

Participants

Using the online classified webpage Craig’s List, 708 participants (522 females, 186

males) were recruited from 10 major U S cities to take part in the study Participants were recruited on the condition that they were at least 18 years old and were currently involved in a committed heterosexual dating relationship of at least 3 months in duration Relationship lengths

ranged from 3 months to 15 years (M = 1.93 years; SD = 1.77 years) Participants were drawn

from a diverse non-university sample (4.7% African American; 9.7% Asian; 72.2% Caucasian;

6.2% Latino; 7.2% other) and ranged in age from 18 to 60 (M = 27.06; SD = 6.56) They were

unpaid but were given basic computer-generated feedback about their personality The feedback told them whether they scored below average, about average, or above average on the five dimensions of the FFM compared to others who had previously completed the same measure

Measures

Personality The Ten-Item Personality Inventory (TIPI; Gosling, Rentfrow, & Swann, 2003) was used in this study The standard version of the TIPI constituted our being measure of

personality The TIPI contains two items for each of the FFM dimensions, with each item rated

on a 7-point Likert-type scale ranging from 1 (disagree strongly) to 7 (agree strongly) The TIPI shows high convergent validity with other widely used FFM scales in self and observer reports

The scale was constructed to emphasize content validity considerations, such that internal

consistency estimates (alphas) are inappropriate; however, the scale has demonstrated very good

test-retest reliability (mean r = 72 across traits; Gosling et al., 2003)

We also measured contextualized personality (acting) in five different contexts: around

romantic partners, coworkers, friends, parents, and siblings For the five acting measures, the TIPI was altered to reflect the different relationship contexts in which participants’ acting would

be assessed At the top of each of the respective acting measures, participants were instructed to indicate how they typically act around particular people (e.g., coworkers, romantic partners) For example, the top of the romantic partner acting measure read, “How You Act Around Your Romantic Partner.” The instructions for each of the acting measures were modified from the

standard TIPI instructions The original TIPI instructions (which were used in our being

measure) read, “Below are a number of personality traits that may or may not apply to you You should rate the extent to which the pair of traits applies to you, even if one characteristic applies more strongly than the other,” whereas the instructions for the acting measure in the romantic relations condition read, “Here are a number of descriptions that may or may not apply to you with regard to how you act around your romantic partner Please indicate the extent to which youagree or disagree with each description below You should rate the extent to which each pair of words applies to you, even if one characteristic applies more strongly than the other.” The

content of the actual items themselves remained exactly the same across the being and acting

measures

Relationship satisfaction Self reports of romantic relationship satisfaction were measured

using the Relationship Assessment Scale (RAS; Hendrick, 1988) The RAS is a validated

measure of relationship satisfaction that correlates strongly with measures of love, commitment,

investment and dyadic adjustment The RAS consists of 7 items on a 7-point Likert-type scale such as, “In general, how satisfied are you with your relationship?” The alpha reliability for the current sample was 89, which is typical of reliability estimates reported for this measure.

Procedure

Participants were directed from advertisements posted on Craig’s List to a

password-protected website at the University of Texas at Austin Previous research has demonstrated that web-based questionnaires provide valid, reliable data, and are not adversely affected by non-serious responders (Gosling, Vazire, Srivastava, & John, 2004) After completing an online consent form, participants provided demographic information and basic information about their

romantic relationship They then completed the being TIPI measure and the five acting TIPI

measures assessing how they typically act around their romantic partners, coworkers, friends,

parents, and siblings The order of presentation of the being and acting measures was

counterbalanced to prevent any potential order effects between being and acting and also within acting In some cases, participants were unable to complete a particular measure For example,

participants without siblings could not complete the siblings acting measure In such cases,

participants were instructed to go on to the next questionnaire After completing the being and

acting measures, participants completed the RAS with respect to their romantic relationship and

then were given computer-generated feedback about their personality

Results and Discussion

Question 1: How strongly is acting in relationships associated with relationship satisfaction?

We predicted that the association between personality factors and romantic relationship satisfaction would be strengthened by taking into account the context of the relationship As shown in Table 1, being was correlated with romantic relationship satisfaction for Agreeableness,

Conscientiousness, Neuroticism, and Openness to Experience Across traits, the absolute

averages of the simple correlations with relationship satisfaction for being, acting around

partners, acting around coworkers, acting around friends, acting around parents, and acting around siblings were 14, 35, 05, 04, 10, and 08, respectively Acting around partners was—for every trait—the strongest predictor of romantic relationship satisfaction in comparison with being and acting in other contexts1

We next sought to determine the uniqueness of the effects of each being and context acting trait on romantic relationship satisfaction as well as the overall predictive power

partner-of being vs acting on satisfaction To test this, we first entered all five being traits together into astepwise multiple regression analysis, with relationship satisfaction as the dependent variable Only Neuroticism and Conscientiousness significantly predicted satisfaction, with standardized beta weights of -.16 and 14, respectively Together, these two being traits accounted for 6% of the variance in satisfaction As a comparison, we entered the five partner-context acting traits together into a stepwise multiple regression analysis All five traits independently predicted satisfaction, with beta weights of 24, 21, -.19, 13 and 08, respectively, for Agreeableness,

Extraversion, Neuroticism, Openness, and Conscientiousness (all p’s < 05) Overall,

partner-context acting explained more than 31% of the variance in relationship satisfaction Further, acting in other relationship contexts was not a significant predictor of romantic relationship satisfaction when controlling for partner-context acting

Question 2: Does acting mediate the relationship between being and relationship satisfaction?

We next tested the possibility that the effects of being on relationship satisfaction were mediated by acting A variable is considered a mediator to the extent that it carries the influence

of a given independent variable to a given dependent variable Based on the guidelines of Barron

and Kenny (1986), mediation is inferred when: (1) the IV (being) significantly affects the

mediator (acting); (2) the IV (being) significantly affects the DV (relationship satisfaction) in the absence of the mediator; (3) the mediator (acting) has a significant unique effect on the DV (relationship satisfaction); and (4) the effect of the IV (being) on the DV (relationship

satisfaction) shrinks or is reduced to zero upon the addition of the mediator (acting) to the model

If these four steps are met, a formal test of mediation can be conducted using a Sobel z-test (MacKinnon, Warsi, & Dwyer, 1995; Preacher & Leonardelli, 2001; Sobel, 1982)

Our mediation analyses were limited to the four traits for which both being and acting were significantly correlated with relationship satisfaction: Agreeableness, Conscientiousness, Neuroticism, and Openness Results of these analyses are presented below

Agreeableness As shown in Panel A of Figure 2, acting agreeable fully mediated the effect of being agreeable on relationship satisfaction (z = 9.77, p < 001); being agreeable

predicted relationship satisfaction only to the extent that it was related to acting agreeable in the context of one’s relationship

Conscientiousness As shown in Panel B of Figure 2, acting conscientious fully mediated the effect of being conscientious on relationship satisfaction (z = 5.20, p < 001); being

conscientious predicted relationship satisfaction only to the extent that it was related to acting conscientious in the context of one’s relationship

Neuroticism As illustrated in Panel C of Figure 2, acting neurotic fully mediated the effect of being neurotic on relationship satisfaction (z = 9.37, p < 001); being neurotic predicted

relationship satisfaction only to the extent that it was related to acting neurotic in the context of one’s relationship

Openness As shown in Panel D of Figure 2, acting open fully mediated the effect of being open on relationship satisfaction (z = 7.71, p < 001); being open predicted relationship

satisfaction only to the extent that it was related to acting open in the context of one’s

relationship

Summary of Study 1 Findings

In Study 1, we set out to answer two central questions Question 1 asked how strongly acting in relationships is associated with romantic relationship satisfaction We found that

relationship-specific personality (acting) was strongly associated with relationship satisfaction across the FFM traits for romantic relationships, while acting in the domains of other

relationships was not associated with romantic relationship satisfaction when controlling for acting in romantic relationships Further, all five relationship-specific personality traits uniquely predicted satisfaction when controlling for the effects of the other acting traits, accounting for over 30% of the variance in relationship satisfaction Question 2 asked whether acting mediates the relationship between being and relationship satisfaction We found that acting fully mediated the relationship between being and satisfaction for Agreeableness, Conscientiousness,

Neuroticism, and Openness In other words, global personality traits predicted relationship satisfaction only to the extent that they were manifested specifically in the context of one’s relationship

Study 2: Acting in Couples’ Everyday InteractionsThe results of Study 1 were promising but were limited to self-report measures Althoughself-reported acting was strongly associated with relationship satisfaction across traits, one could argue that our measure of acting was biased by people’s self perceptions and reflected not how

people act around their partners but rather how they think they act Would we get the same

effects if we measured actual behavior in the context of relationships? Additionally, Study 1 focused solely on how acting and being were related to participants’ own levels of relationship satisfaction; partners’ satisfaction levels were not considered To what extent might acting and being be related to partners’ satisfaction?

Thus, in Study 2 our goals were: 1) to take a more naturalistic approach and directly examine how people actually act in their everyday interactions with their romantic partners; and 2) to explore how being and acting might be related to partners’ satisfaction levels

MethodStudy 2 involved two phases of data collection In the first phase, both members of datingcouples completed self-report measures of personality and relationship satisfaction and submitted

7 days of IMs In the second phase, we obtained objective ratings of personality in the context of participants’ relationships (acting) from observers’ ratings of the couples’ IMs

Participants

Undergraduate couples at the University of Texas at Austin were recruited through an online computer sign-up system on the basis that they: 1) were in a committed heterosexual romantic relationship, and 2) IMed with each other every day Sixty-eight couples (136

participants: 68 women, 68 men; mean age 19.04, SD = 1.39) participated in the study in

exchange for course credit Couples had been dating an average of 1.44 years (SD = 1.25) Measures

Personality Our being measure of personality was again a self report on the standard

version of the TIPI (Gosling et al., 2003) As mentioned previously, the TIPI was constructed to emphasize content validity considerations, such that internal consistency estimates (alphas) are

inappropriate; however, the scale has demonstrated very good test-retest reliability (mean r = 72

across traits; Gosling et al., 2003)

Because Study 1 found only weak associations between acting and relationship

satisfaction in non-romantic relationship contexts, we included only a measure of romantic relationship-context acting in Study 2 To create an objective measure of acting in Study 2, 12 independent observers used the TIPI to rate participants’ personalities based on examination of participants’ IMs with their romantic partners

Relationship satisfaction Relationship satisfaction was measured using the RAS

(Hendrick, 1988) In the present sample, the alpha reliability was 79

Procedure

Phase I During an introductory session with the experimenter, couples were instructed

to forward IMs to a secure email address during two monitoring periods—three days of IMs during the 1st monitoring period and 4 days during the 2nd The monitoring periods were

separated by three days in between2 Considerable effort was taken by the experimenter during the introductory session to ensure that participants and their partners felt at ease about

forwarding their IMs and to encourage them to contact the experimenter if they had any

concerns Upon receipt by the experimenter, all IMs were saved as text files in a secure location accessible only to the experimenter and all personally identifiable information was removed The

mean length of couples’ IM conversations over the seven days was 4,813 words (SD = 4,854; Mdn = 3,355; Min = 272; Max = 23,221).

Online self-report questionnaires were completed by couples after the introductory

session with the experimenter on day 1 of the study Participants completed the being measure ofpersonality (TIPI) and the measure of relationship satisfaction The importance of completing

these questionnaires privately and confidentially was emphasized by the experimenter during the introductory session.

Phase II Twelve independent observers rated the participants’ acting based on

examination of participants’ IMs with their romantic partners The observers were undergraduate students working on the project as research assistants They were unacquainted with the

participants and were instructed not to discuss their ratings with one another or with others outside of the project The order in which the observers rated the IMs was randomly generated for each observer Observers completed TIPI ratings of each couple member after reading the complete transcripts of IMs for each couple Alpha reliabilities for the composites of the ratings were 85, 70, 90, 93, and 91 for Extraversion, Agreeableness, Conscientiousness, Neuroticism,and Openness to Experience, respectively

Results and Discussion

Question 1: How strongly is acting in relationships associated with relationship satisfaction?

We predicted that observers’ ratings of acting would be strongly associated with self and partner ratings of relationship satisfaction Separate correlational analyses were conducted to assess: 1) the association between being and self reports of satisfaction; 2) the association

between acting and self reports of satisfaction; 3) the association between being and partner reports of satisfaction; and 4) the association between acting and partner reports of satisfaction The results of these analyses are presented in Table 2

Correlations with self-reported satisfaction Consistent with the findings in the literature

(Botwin et al., 1997; McCrae et al., 1998), being agreeable was positively correlated with

satisfaction and being neurotic was negatively correlated with satisfaction In addition, acting agreeable was positively correlated with satisfaction and acting neurotic was negatively

correlated with satisfaction Separate multiple regression analyses indicated that being agreeable and being neurotic together accounted for 15% of the variance in self-reported satisfaction, whereas acting agreeable and acting neurotic together accounted for 19% of the variance.

Correlations with partner-reported satisfaction We next tested how strongly being and

acting were associated with partners’ relationship satisfaction Being was correlated with

partners’ relationship satisfaction only for Agreeableness Acting agreeable, acting conscientious,and acting neurotic all were associated with partners’ satisfaction Being accounted for 3% of thevariance in partners’ relationship satisfaction, whereas acting accounted for 9%

Question 2: Does acting mediate the relationship between being and relationship satisfaction?

As with Study 1, we followed the guidelines of Barron and Kenny (1986) for assessing mediation effects Sobel z-tests were used as formal tests of mediation (MacKinnon, Warsi, & Dwyer, 1995; Preacher & Leonardelli, 2001; Sobel, 1982) Our analyses were limited only to the traits for which both being and acting were associated with satisfaction: Agreeableness and Neuroticism for self-reported satisfaction and Agreeableness for partner-reported satisfaction

Agreeableness and self-reported satisfaction As shown in Panel A of Figure 3, acting

agreeable partially mediated the association between being agreeable and self-reported

relationship satisfaction (z = 3.16, p < 002) These results suggest that part of the explanation for

the relationship between being agreeable and relationship satisfaction is that agreeable people tend to act more agreeable in their relationships, which is in turn likely to result in greater

relationship satisfaction

Neuroticism and self-reported satisfaction As illustrated in Panel B of Figure 3, there

was a trend of acting neurotic partially mediating the effect of being neurotic on self-reported

satisfaction (z = 1.77, p < 08) These results are consistent with the idea that part of the