Communication

INTERPERSONAL RELATIONS AND GROUP PROCESSES

"I Love You More Today Than Yesterday": Romantic Partners' Perceptions of Changes in Love and Related Affect Over Time

Susan Sprecher Illinois State University

Partners in romantic relationships provided reports on perceived changes in their love, commitment, and satisfaction and completed contemporaneous scales on the same relationship phenomena multiple times over several years. At each wave of the longitudinal study, participants whose relationships had remained intact perceived that their love and related phenomena had increased since they had last participated in the study. However, their scores on contemporaneous scales did not generally increase over time. Analyses indicated that participants' reports of change were related to actual change in love, commit- ment, and satisfaction scores and with future relationship stability. Furthermore, participants who experienced a breakup during the longitudinal study reported an overall decrease in their positive affect in the months prior to the breakup.

Do intimate partners really love each other more with time, as suggested by the title of this article? Do they perceive that their love is increasing? Because close relationships are extended in length, partners' love and other feelings for each other are likely to change over time and be perceived to change. However, the temporal course of love and related relationship phenomena (e.g., satisfaction) has rarely been studied. In this investigation, a sample of romantic couples (all were dating at the beginning of the study) were surveyed five times over a 4-year period about their relation- ship. At each wave of the study, the participants completed con- temporaneous scales of love, commitment, and satisfaction, and they also reported how the same relationship phenomena had changed since they had last participated in the study. As a result, both actual and perceived changes are examined, as well as the association between the two.

Some theoretical perspectives suggest that love, satisfaction, and related phenomena should increase very early in the relation- ship but then level off or even decrease over time. For example, according to principles of learning theory, the partner's behavior or mere presence becomes less rewarding over time because of ha- bituation effects (e.g., Huesmann, 1980). Furthermore, according to the self-expansion model (Aron & Aron, 1986), satisfaction,

This research was supported, in part, by several small grants from Illinois State University. I thank Beverley Fehr, Diane Felmlee, Elaine Hatfield, John Holmes, Kathleen McKinney, and Pamela Regan for com- ments on earlier versions of this article. I also give a special thanks to the participants in this study, who were willing to complete the questionnaires every time I asked.

Correspondence concerning this article should be addressed to Susan Sprecher, Department of Sociology and Anthropology, Illinois State Uni- versity, Campus Box 4660, Normal, Illinois 61790-4660. Electronic mail may be sent to [email protected].

love, and related emotions are likely to decrease over time as relationship partners offer each other fewer opportunities for fur- ther self-expansion. Stage or filter models of relationships (e.g., Murstein, 1987) would argue that certain aspects of premarital relationships (e.g., intimacy, love) should increase as the relation- ship moves from one phase to another, although recent interper- sonal process models (Cate & Lloyd, 1992; Surra, 1990) allow for growth, decline, and stability in properties of the relationship over time. Overall, few theories have focused directly on relationship change, and the predictions about how love and related phenomena should change over time in the course of premarital relationship development are not consistent.

The most accepted method for examining change over time in love and related phenomena is to have relationship partners complete measures of these variables at two or more times and determine whether there is a significant change (e.g., Metts, Sprecher, & Cupach, 1991). Most longitudinal studies on dating relationships have shown little change in love over time (e.g., Lund, 1985; Mathes & Wise, 1983), although love has been found to increase for individuals who advance in relationship stage (e.g., Sprecher & Felmlee, 1993) and for those who are in a very early dating stage at the beginning of the study (Berg & McQuinn, 1986). Some longitudinal research has also demon- strated increases in satisfaction (Rusbult, 1983) and commit- ment (Lund, 1985) in dating relationships. However, a major limitation of prior longitudinal research for examining change in love or related phenomena is the short time period between assessments (often 3 months or less).

Regardless of whether or how love and related phenomena may actually change over time (as reflected in changes in contempora- neous scores), individuals are likely to have theories about how their feelings change over time. In the retrospective developmental design (e.g., Braiker & Kelley, 1979; Cale, Huston, & Nessel-

Journal of Personality and Social Psychology, 1999, Vol. 76, No. 1, 46-53 Copyright 1999 by the American Psychological Association, Inc. 0022-3514/99/S3.00

46

CHANGES JN RELATIONSHIPS 47

roade, 1986), also used to study change over time in relationships, couples who have reached a particular stage of commitment (e.g., early marriage) have been asked to reflect on earlier stages and complete scales (measuring qualities such as love and commit- ment) for these various stages. This research documents more positive change (i.e., increases in love, commitment, and satisfac- tion) than is evidenced in the longitudinal studies conducted with dating couples. People's beliefs about change in relationship phe- nomena are likely to be influenced by their general implicit theo- ries about relationship development. To the degree that people are motivated to view that their relationship is improving or growing over time (e.g., Honeycutt, Cantrill, & Greene, 1989; Knee, 1998; McFarland & Ross, 1987), they are likely to recall possessing a lower level of love at an earlier time than they do at the present time.

Overview and Hypotheses

However, even in the retrospective developmental design (e.g., Cate et al., 1986), subjective perceptions about change are not assessed. In this 4-year (five-wave) longitudinal investigation, relationship partners were asked to provide direct reports about how their love and related feelings changed over time at each wave of the study, beginning at the second reporting session. This investigation also included a standard longitudinal design to assess change in relationship phenomena. At each wave, the participants who were in intact relationships completed contemporaneous scales about their love and other feelings.

The first purpose of this study was to examine how individuals in stable relationships perceive positive change in their relation- ship over time. My first hypothesis (Hypothesis 1) was that par- ticipants in intact relationships at each wave of the longitudinal study will perceive that their love, commitment, and satisfaction have increased over time; furthermore, the reported increase will be greater than actual change, as reflected in changes in scores on the contemporaneous self-reports. As people recall how their re- lationships change over time, they are likely to be motivated to perceive increases (rather than no change or decreases) in their love and related affect (Honeycutt et al., 1989; Karney & Coombs, 1998). However, consistent with prior longitudinal research (Lund, 1985), I expected to find very little change in contemporaneous scores on love and related phenomena. Thus, perceived change should be greater than actual change.

The second purpose was to examine perceptions of change in relationship phenomena among couples who break up over the course of the study, in part to provide a comparison with the change perceived by couples while in intact relationships. My second hypothesis (Hypothesis 2) was that couples who break up will perceive and report in assessments obtained after the breakup that their love, commitment, and satisfaction had decreased during the time prior to the breakup. Because their current state is unstable (and presumably unhappy), individuals who have experienced a breakup have probably experienced more negative changes in the previous months than couples who remain together and are prob- ably also invoking different theories of relationship change (a belief that relationship dissolution is associated with a decrease in positive aspects of the relationship; see, e.g., Honeycutt, 1993; Honeycutt, Cantrill, & Allen, 1992).

The third purpose was to examine relationship correlates and

consequences of beliefs about relationship change. Are individuals who perceive that their love and related phenomena were increas- ing over time likely to have higher levels of current love and other affect and more likely to experience increases in such phenomena at a future time? Furthermore, are positive beliefs associated with relationship stability? My third hypothesis (Hypothesis 3) was that the more that relational partners perceive, that their love, commit- ment, and satisfaction increased over time, the greater their expe- rience of love, commitment, and satisfaction at a future time and the more likely they are to remain together. A recent theoretical controversy exists in the literature concerning the relationship consequences of positive illusions and beliefs about a partner and the relationship (e.g., Karney & Bradbury, 1997; Murray, Holmes, & Griffin, 1996). According to what has been called a disillusion- ment hypothesis (Karney & Bradbury, 1997) or a disappointment model (Murray et al., 1996), if individuals have extreme positive beliefs about the attributes of their partner or of the relationship, disappointment or disillusionment may occur, because partners and relationships can never live up to such expectations. However, there appears to be more support for a maintenance hypothesis (Karney & Bradbury, 1997) or a wish-fulfillment model (Murray et al., 1996), which states that positive beliefs have positive effects on the relationship because individuals will be motivated to behave in ways that maintain their relationship and their love. Murray et al. (1996) discuss how positive illusions may buffer a couple from the negative effect of any difficulties that may occur in the rela- tionship. Thus, beliefs that one's love is increasing over time may contribute to a self-fulfilling prophecy and therefore to greater love in the future.

In sum, the present study makes a unique contribution to the study of change in close relationships by considering both subjec- tive perceptions of change over time in love and related phenom- ena (i.e., commitment and satisfaction) and change in contempo- raneous scores for the same relationship phenomena as assessed for the same time period.

Method

The data came from a longitudinal study conducted at a Midwestern university with a sample of romantic heterosexual couples.1 The initial sample consisted of both partners of 101 dating couples who completed a self-administered questionnaire in the fall of 1988 (Time 1). Follow-ups were conducted in the spring of 1989 (Time 2) and the spring-summer of 1990, 1991, and 1992 (Times 3, 4, and 5, respectively). The sample size decreased with each wave, because when couples broke up they completed one last questionnaire (about the breakup) and then did not participate further in the study. By Time 5, 59% (n = 60) of the couples had ended their relationships.

Participants

Dating couples were recruited through announcements in classes, adver- tisements in the student newspaper, and posters placed around campus. The mean age of the participants at Time 1 was 20 years. Most participants in

1 Other papers (Sprecher, 1994; Sprecher & Felmlee, 1992, 1997) have been written from aspects of this longitudinal data set but do not focus on the issues examined in this article, which are perceptions of change in love and related phenomena and their association with scores on contempora- neous scales.

48 SPRECHER

the sample were Caucasian (98%) and of the middle or upper-middle class (87%). The mean number of months the couples had been dating when the study began was 18.7; the range was from 1 month to 55 months. Of the 41 couples who were still together at Time 5, 29 (71%) had married.

Procedure

At Time 1, couples were scheduled to complete a self-administered questionnaire (separately) at a campus location. In the follow-ups, partic- ipants who were still attending classes at the university came to the campus to complete a questionnaire (partners from couples who broke up came at different times). Participants who had moved away were mailed the ques- tionnaire and a stamped, self-addressed return envelope. Among the cou- ples who remained together over the study (n = 41), there was very little nonresponse.2 Thirty-eight of the women and 36 of the men participated in all five waves of the study. A higher rate of nonresponse occurred in the final contact (the breakup questionnaire) for the subsample of couples who broke up (n = 60), although the response rate was still very high i

Measurement

Beginning at Time 2, the questionnaire for the intact couples contained a set of questions that measured their subjective perceptions of changes in their relationships for the period of time since the last data collection (which was 6 months at Time 2 and approximately 1 year at Times 3-5). Participants were asked, "How do you think each of the following has changed—if at all—in the past year?" Couples, upon breaking up, were asked to respond for the period of time prior to the breakup. Although several relationship phenomena were listed, the focus of this study was on change in love and related affect. The particular items analyzed were listed in the following ways: "love and affection for this person," "commitment to the relationship," and "satisfaction in the relationship." Each item was followed by a 7-point response scale ranging from 1 = decreased to 4 = same (i.e., no change) to 7 = increased. Because the three items were highly intercorrelated (mean r = .81 for men and .72 for women at Time 2), an index of perceived change, represented by the mean response to the three items, was created. The higher the score on this index, the more the participant perceived his or her positive feelings to increase over time. The alpha coefficient for this index was .93 for men and .84 for women, on the basis of the intact sample at Time 2.3

At each wave of the study, participants who were in intact relationships also completed a lengthy questionnaire about those relationships, which included scales that measured their current love, commitment, and satis- faction. Love was assessed by the 10-item Braiker and Kelley (1979) love scale (each item was followed by a 7-point response scale; Time 1 alpha was .85 for men and .81 for women). Commitment was measured by 4 items from the Lund (1985) commitment scale and one additional global item of commitment (each item was followed by a 7-point response scale; Time 1 alpha was .89 for men and .78 for women). Satisfaction was assessed by the 7-item Hendrick (1988) Relationship Assessment Scale (each item was followed by a 5-point response scale; Time 1 alpha was .81 for men and .75 for women). For each scale, a higher score indicated greater love, commitment, and satisfaction, respectively.

In addition, an index of contemporaneous feelings comparable to the three-item index of perceived change was created by combining, from each scale, one item that referred most directly to the specific affect. These items were "To what extent do you love at this stage"; "How committed are you to your partner"; and "In general, how satisfied are you with your relationship?" The latter item, which had a 5-point response scale, was first multiplied by a factor of 1.4 to make it a 7-point response scale similar to the other two items. Time 1 alpha for this index of contemporaneous feelings was .79 for men and .77 for women.

Results

Perceptions of Change in the Stable Couples

The participants in intact relationships at Time 2 (n = 84 couples) reported there were increases in their love and related affect for their partner since they had completed the first question- naire (6 months earlier), in support of Hypothesis 1. The mean response to the index of perceived change was 5.22 for men and 5.62 for women. These mean scores were significantly above the midpoint, no change, as indicated by t tests for a single sample. The score for women was significantly higher than that for men, paired f(79) = -3.05,;? < .01.

The subsamples of intact couples surveyed at Times 3, 4, and 5 also perceived increases in love and related affect (see means in Table 1). In each case, the mean score to the index of perceived change was significantly higher than the midpoint (which repre- sented no change). Women's scores were significantly higher (p < .05) than men's at Times 3 and 5 (as well as at Time 1, as reported above). Partners perceived similar changes in relationship phe- nomena only at some of the waves. The interpartner correlation for the index of perceived change was .63 (p < .001) at Time 2, .36 (p < .01) at Time 3, .26 (ns) at Time 4, and .17 (ns) at Time 5. Table 1 also presents the mean scores for each retrospective item (i.e., love and affection, commitment, and satisfaction). At each wave, participants reported that each specific affect had increased since they had last participated in the study.

I also examined whether the reports of change varied over time for the subsample of participants who remained in their relation- ship throughout the entire study and who participated at every wave (n = 35 men and 38 women from the 41 couples). A repeated measures analysis (within a multivariate analysis of variance) indicated no significant changes over time in scores on the index of perceived change for either the men or the women. A consistent overall positive change was reported at each wave for this group of most stable couples.

Hypothesis 1 also stated that stable couples would perceive greater increases in love and related affect than would actually be found in an analysis of changes in scores on contemporaneous scales between waves of the study. As the results in Table 2 indicate, the scores on the index of contemporaneous feelings did not change significantly between any adjacent waves of the study for the groups of stable couples as a whole. Furthermore, between most waves, there were no significant changes in scores on the love, commitment, and satisfaction scales. The significant changes between Times 1 and 2 were actually negative rather than positive (in love for men and satisfaction for both men and women). The only significant increases were between Times 4 and 5, and these

Six participants from five different relationships could not participate in one of the waves, and one couple (2 participants) is missing from two waves.

3 The alpha coefficient for this index and all other multiple-item scales and indexes described in this section was similarly high for the smaller subsamples of intact couples remaining in the study at later time points. The one exception was a lower alpha coefficient at Time 5 for women (for commitment and the index of contemporaneous feelings), apparently be- cause of a lack,.of variation on one or more scale items (all women scored at the highest end of the scale).

CHANGES IN RELATIONSHIPS 49

Table 1 Perception of Changes in Love, Commitment, and Satisfaction for Men and Women in Intact Couples at Four Waves of the Study and in the Breakup Sample

Measure

n (couples) Three-item index of perceived change

Men Women

Specific items Love and affection

Men Women

Commitment Men Women

Satisfaction Men Women

Intact Time 2

84

5.22 5.62

5.44 5.94

5.21 5.65

5.04 5.26

Intact Time 3

62

5.44 5.82

5.68 5.89

5.55 5.89

5.08 5.69

Sample

Intact Time 4

48

5.51 5.81

5.50 5.81

5.65 5.90

5.39 5.73

Intact Time 5

41

5.62 6.06

5.82 6.10

5.87 6.23

5.18 5.85

Breakup

58

3.32 3.44

3.92a 4.15.

3.22 3.37

2.76 2.82

Note. The above mean responses are based on a response scale ranging from 1 (decreased) to 7 [increased), where 4 = same or no change. The exact sample size for each analysis varied as a function of the number of participants at each wave (both partners did not participate at every wave) and the amount of missing data on each variable. With the exception of the means that are marked with the subscript a (for the breakup sample), each mean was significantly different from 4, the midpoint representing the perception of no change.

were for commitment and satisfaction for women. These results, when considered in conjunction with the results presented above for the reports of change, provide support for the prediction in Hypothesis 1 that the amount of change reported by participants is greater than actual change (as evidenced in change in contempo- raneous scores).

I also considered the possibility that the most stable couples

(those who stayed together throughout the longitudinal study) would experience increases in scores on contemporaneous scales over time. In fact, scores on the index of contemporane- ous feelings did increase over time for men, linear F(l , 34) = 4.47, p < .05 (means at Time 1 = 6.39, at Time 2 = 6.43, at Time 3 = 6.44, at Time 4 = 6.55, at Time 5 = 6.66), although not for women, linear F(l , 37) = 0.89, ns (means at Time 1 =

Table 2 Changes Between Adjacent Waves in Contemporaneous Scores on Love, Commitment, and Satisfaction for Participants in Intact Relationships

Measure

Three-item index of contemporaneous feelings

Men Women

Specific scales Love

Men Women

Commitment Men Women

Satisfaction Men Women

Tl

6.39 6.57

6.31 6.45

6.03 6.32

4.44 4.51

Intact Time

T2

6.23 6.42

6.10 6.39

5.98 6.27

4.29 4.35

df

80 79

74 75

78 79

78 79

2

t

-1.67 -1.88

-2.63** -1.03

-0.42 -0.43

-2.40* -3.21**

T2

6.45 6.52

6.26 6.49

6.23 6.43

4.44 4.44

Intact Time

T3

6.33 6.54

6.18 6.39

6.17 6.37

4.32 4.42

df

56 58

51 56

57 58

55 58

Sample3

3

t

-1.19 0.26

-0.74 -1.42

-0.46 -0.48

-1.48 -0.33

T3

6.42 6.68

6.28 6.50

6.27 6.55

4.42 4.47

Intact

T4

6.47 6.58

6.31 6.44

6.38 6.56

4.46 4.44

Time

df

42 47

39 46

44 47

42 47

4

t

0.43 -1.61

0.23 -0.85

0.84 0.05

0.52 -0.60

T4

6.46 6.67

6.31 6.55

6.30 6.69

4.46 4.51

Intact Time

T5

6.52 6.78

6.36 6.64

6.58 6.85

4.51 4.63

df

36 38

35 38

36 38

36 38

5

t

0.54 1.70

0.49 1.51

1.85 2.39*

0.85 2.57*

Note. The exact sample size for each analysis varied as a function of the number of participants at each of the two waves included (both partners did not participate at every wave) and the amount of missing data on each variable. Tl = Time 1; T2 = Time 2; T3 = Time 3; T4 = Time 4; T5 = Time 5. a The sample size for Intact Time 2 = 84 couples, for Intact Time 3 = 62 couples, for Intact Time 4 = 48 couples, and for Intact Time 5 = 41 couples. * p < . 0 5 . **p < .01.

50 SPRECHER

6.70, at Time 2 = 6.63, at Time 3 = 6.77, at Time 4 = 6.66, at Time 5 = 6.77). However, the means were higher for women than for men in this subsample even early in the study. An analysis of the three specific scales indicated that commitment increased significantly over time for this subsample: linear F(l, 34) = 9.52, p < .01, for men, and linear F(l , 37) = 6.32, p < .05, for women. (For a table of these means, write to Susan Sprecher.)

Perceived Change Among the Breakup Couples

The second hypothesis stated that breakup couples will perceive that their love, commitment, and satisfaction decreased in the period prior to the breakup. Data were available after the breakup for both partners of 47 couples and one partner from 11 other couples who ended their relationships. Among this breakup sam- ple, the mean score on the index of perceived change was 3.32 for men and 3.44 for women. These scores were significantly below the midpoint of 4 (no change) and in the direction of the anchor decreased. The difference between men and women was not significant, and there was little similarity between ex-partners in perceptions of change (r = —.08). The final column of Table 1 also presents the mean scores for the individual items on change (love, commitment, satisfaction) for the subsample of breakup couples. Love was reported to decrease the least; in fact, the mean for this item for both men and women was at 4, no change. (Follow-up analyses indicated that this mean score was the result of most participants scoring at or near the midpoint of no change rather than a result of a bimodal distribution on the item.)

Because participants in the breakup sample did not complete the contemporaneous scales of love, commitment, and satisfaction in their breakup questionnaire, it was not possible to determine in any straightforward way whether they over- or underestimated the change that occurred in their relationships prior to breaking up. However, it can be assumed that positive affect decreases for couples just prior to a breakup.

Associations Between Reports of Change and the Contemporaneous Measures

The final issue examined was the degree to which participants' beliefs about past changes in their love and related affect are associated with current feelings in the relationship and with future outcomes, including actual changes in feelings and the stability of the relationship. Only the analyses conducted with the three-item index of perceived change and the three-item index of contempo- raneous feelings are presented. (The analyses were also conducted with the love, commitment, and satisfaction scales and the corre- sponding individual reports of change, but unless noted, the results were very similar.)

First, I found that those individuals who reported more positive change for the period since the last data collection also experi- enced more positive current feelings. At Waves 2-5, scores on the index of perceived change were significantly correlated with scores on the index of contemporaneous feelings (correlations ranged from .64 to .71 for men and from .33 to .65 for women).

In a second analysis, I examined whether actual change in contemporaneous scores between adjacent waves was associated with a perception of change for the same period. Lagged regres-

sions were conducted in which the score on the index of contem- poraneous feelings at Time N (where N could be 2, 3, 4, or 5) was predicted by the score on the index of contemporaneous feelings at Time N — 1 and the index of perceived change at Time N. In this analysis, a significant beta for the index of perceived change indicates that the perception of change is associated with actual change between two waves on the contemporaneous scores. With one exception (i.e., women at Time 5), in all regressions the index of contemporaneous feelings was predicted by both the index of contemporaneous feelings at Time N — 1 (significant betas ranged from .29 to .54 for men and from .26 to .66 for women) and by the index of perceived change at Time N (significant betas ranged from .42 to .62 for men and from .37 to .52 for women). These results, then, indicate that reports of change at Time N are asso- ciated with actual change in the contemporaneous scores between Time N - 1 and Time N 4

However, can beliefs about change forecast or contribute to change at a future time in the relationship? To examine this, I conducted lagged regressions, which examined whether scores on the index of perceived change at Time N — 1 predicted the scores on the index of contemporaneous feelings at Time N, controlling for the scores on the index of contemporaneous feelings at Time N — 1. These analyses were conducted only for the intact samples at Times 3, 4, and 5, because there were no Time N — 1 reports of change for the Time 2 intact sample. In general, these results indicated that the index of contemporaneous feelings at Time N — 1 was associated significantly with the index of contemporaneous feelings at Time N (except for women at Time 5). However, the index of perceived change at Time N — 1 was generally not predictive of the index of contemporaneous feelings at Time N. The one exception was that for women, the index of contempora- neous feelings at Time 3 was predicted by the index of perceived change at Time 2 (/3 = .32, p < .05). Thus, these results suggest that positive beliefs about change in the relationship at one point do not lead to actual positive change at a future time (e.g., a year later) in the relationship.

In a final analysis, I examined whether perceptions about change in one's love and related affect, controlling for contemporaneous feelings, were related to relationship stability (vs. instability) over time. Because the 17 couples who broke up by Time 2 were not asked at Time 1 (prior to their breakup) about perceptions of change, they could not be included in these analyses. Thus, I used the Time 2 index of perceived change and index of contempora- neous feelings as predictors of whether the relationships that were still intact at Time 2 (84 of the original 101 couples) broke up by the end of the study (of these 84 couples, 43 ended their relation- ship) and also included length of the relationship at the time the

4 This association may occur even though there was little actual change in participants' scores on love and related emotions between Time N — 1 and Time N. There was variation in the degree and direction of actual change in scores on the index of contemporaneous feelings, and although a majority of the participants scored above the midpoint on the index of perceived change, participants varied in how far above the midpoint they scored. Furthermore, those who experienced sharp decreases between Time N — 1 and Time N in scores on the index of contemporaneous feelings tended to report a decrease at Time N. Hence, although in the aggregate there was little actual change but a great deal of perceived change, perceived change was correlated with actual change.

CHANGES IN RELATIONSHIPS 51

study began as a control. Because there was an almost even distribution on the dichotomous dependent variable, ordinary least squares regression was used. This regression was conducted sep- arately for men and women.

Relationship length at the time the study began was associated negatively with the likelihood that the relationship broke up over time for both men (j3 = -.25,.p < .05) and women (j3 = - .22, p < .05). That is, those who were in longer term relationships at the beginning of the study were more likely to still be together at the end of the study. Also, in partial support of Hypothesis 3, there was a trend for perceptions that one's feelings (e.g., love) were increasing to be associated negatively with breaking up. The beta for the index of perceived change was — .26 for men (p < .09) and —.26 for women (p < .07). However, scores on the index of contemporaneous feelings (at Time 2), included in the same re- gression equation, showed little association with final relationship status (J3 = —.04 for men and —.08 for women, both ns).5

Discussion

Perceived and Actual Change in Stable Couples

Hypothesis 1 predicted that the romantic partners in stable relationships perceive that their love, commitment, and satisfaction increase over time. This hypothesis was strongly supported. For the intact samples at Times 2, 3, 4, and 5, men and women perceived that their positive feelings (e.g., love) for their partner had increased since they had last participated in the study (these reports of change were not requested at Time 1). Thus, with each passing year participants believed that their love (and commitment and satisfaction) had grown. These results are consistent with other findings in the literature indicating that people perceive then- relationship and partner in idealistic ways (e.g., Murray & Holmes, 1993; Murray et al., 1996; Van Lange & Rusbult, 1995). It is also possible that people's definition of love and related affect changes over time so that compared with the type of love (commitment or satisfaction) that they are currently experiencing, what they recall experiencing at an earlier time may seem to have a different and lesser quality.

Although the participants in intact relationships at each wave perceived that their love, commitment, and satisfaction were in- creasing over time, there was little evidence that these affects, as measured by contemporaneous measures, were actually increasing between waves of the study, also in support of Hypothesis 1. The exception was that in the most stable couples (those who stayed together throughout the study), an overall increase was found for affect, particularly commitment. The general lack of change in contemporaneous scores between waves, however, may be due to ceiling effects reached in the measures early in the study, which made it difficult for significant differences to be evidenced.6

Perceived Change Among the Breakup Couples

Most of the individuals whose relationships dissolved during the longitudinal study completed one additional questionnaire after the breakup, which included items asking how their love and related phenomena had changed before the breakup. These participants reported overall decreases in their positive feelings (especially

satisfaction and commitment) prior to the breakup. These results are consistent with the research indicating that decreases in com- mitment, satisfaction, and other positive feelings in the relation- ship occur before relationship breakdown (e.g., Rusbult, 1983). In addition, however, for some individuals the reported decrease in positive affect for the period prior to the breakup may reflect attempts, in postdissolution rumination, to make sense of the breakup (Duck, 1982).

Of the positive feelings, satisfaction was perceived to decrease the most, whereas love was perceived to decrease the least. These results suggest that people do not end their relationships because of the disappearance of love, but because of a dissatisfaction or unhappiness that develops, which may then cause love to stop growing. Research on emotional reactions after a breakup (Sprecher, 1994) has also found that love does not dissipate com- pletely in the process of relationship dissolution.

Associations Between Reports of Change and Contemporaneous Measures

Evidence was found in this study that beliefs about increases in love and related phenomena were associated with current feelings in the relationship and future positive outcomes. First, I found at each wave of the study that included collection of reports of change (Waves 2-5) that individuals in intact relationships who reported increases in their positive feelings since they had last participated in the study were also likely to experience current high levels of love, commitment, and satisfaction with their partner. However, these analyses were based on reports of perceived

5 In three additional analyses, the score on each individual contempora- neous scale (love, commitment, satisfaction), one at a time, was included along with the corresponding report of change. In each regression, the contemporaneous scale score did not predict final relationship status. These additional regressions also indicated that it was the perception that love was increasing, in particular, that was especially predictive of relationships stability.

6 Although there was very little statistical difference in scores between adjacent time points, ceiling effects may be reached on standard scales and measures very early in a longitudinal study, which can make it impossible for participants to score significantly higher at later waves even though their love and related affect may be increasing. In fact, for each scale at Time 1, one fourth to one half of the participants had one of the highest possible scores (0, 1, or 2 response points from the highest absolute summed score). To account for this ceiling phenomenon, I conducted an analysis that allowed for any slight increase in a scale score over time to count as an increase, and I then compared the proportion of participants who had a higher score on a contemporaneous scale at Time N than at Time N - 1 with the proportion of participants at Time N who reported an increase on the report of change for the same relationships phenomena. These proportions (relative to those who reported no increase or a de- crease) were compared using a nonparametric sign test, which counts the positive and negative differences between two variables. Each variable (change as represented by the report of perceived change and change as represented by the difference between contemporaneous scores between two adjacent time points) was coded using the scale 1 = no change or decrease and 2 = increase. In every single case (examined for love, commitment, and satisfaction for men and women separately at Times 2, 3, 4, and 5), there were significantly (p < .0000 top < .05) more cases in which the report was higher, showing an increase not shown by the contemporaneous reports, than was the reverse.

52 SPRECHER

change and contemporaneous measures completed at the same time point, and thus the participants' current relationship states may have influenced how they recalled change. As suggested by Ross's (1989) theory of emotions, individuals who are in a current state of happiness and love are likely to report they experienced an increase in their positive feelings if they believe that such affect increases over time in dating relationships.

Additional evidence was found that perceptions about change were related to actual change. Lagged regression analyses indi- cated that the reports of perceived change at Time N (for the period prior to Time N) predicted scores on the contemporaneous scale for the same relationship phenomena at Time N, controlling for scores on the contemporaneous scale at Time N — 1 (i.e., change in the contemporaneous scores). McFarland and Ross (1987) also found that individuals who come to like and love their partner more over time reported experiencing more love and liking in the past.

These results, however, may only indicate that people are ac- curate in their estimates of change and do not necessarily demon- strate that beliefs lead to future change. Thus, additional lagged regressions were conducted to examine whether Time N — 1 reports of perceived change were associated with change in con- temporaneous scores between Time N — 1 and Time N. In these analyses, almost no support was found for the idea that beliefs about positive change contribute to actual positive change at a later time.

Additional analyses indicated that participants' beliefs about change were associated with final relationship status. With the subsample of couples who were still in intact relationships at Time 2, I conducted a regression for each gender in which rela- tionship status by the end of the study was regressed on the Time 2 index of contemporaneous feelings and index of perceived change. The length of the relationship at the time the study began was controlled for in these analyses and was expected to be associated with final relationship stability. These results indicate that beliefs about increases in positive feelings in the relationship were asso- ciated with relationship stability, controlling for relationship length and contemporaneous feelings, and demonstrate the importance of positive beliefs and optimism for the maintenance and continua- tion of romantic relationships. The finding that the length of the relationship was associated with stability is consistent with a similar finding from several other studies examining predictors of the stability of dating relationships (e.g., Felmlee, Sprecher, & Bassin, 1990; Simpson, 1987). Duration of relationship is thought to represent how invested one is in the relationship (Rusbult & Martz, 1995), and investments are theoretically associated with longevity of relationships (e.g., Rusbult, 1983).

Methodological Strengths and Limitations

There were several strengths of this study. No previous study had measured individuals' perceptions about changes in their romantic relationships at multiple time points and also compared these perceptions with changes in the same variables for the same period of time using a standard longitudinal design with contem- poraneous scale scores. Other strengths include the lengthy time span of the longitudinal investigation and data being collected from both partners.

Of course, there were also limitations. First, the data were

limited to the heterosexual, romantic couples who were initially dating. Whether similar changes would be found in other types of relationships (e.g., homosexual, friendship) or for other stages of heterosexual romantic relationships (e.g., long-term marriage) needs to be investigated in future research. Furthermore, the sam- ple was rather homogeneous on background and demographic characteristics (most were Caucasian, middle class, and in college, and all were from the United States). It could also be argued that relationship change was studied in a limited way. Perceived change was assessed for a long period of time (e.g., a year), and the longitudinal design was based on data collected approximately once a year. Such data may mask fluctuations that occur within that period of time. For example, even though individuals may believe that their love increases over a year, data collected more frequently may show that downturns are also experienced.

In conclusion, this study indicates that while involved in a romantic relationship, people perceive that their love and related affect increase year after year. Although the participants in this study did not generally score higher on contemporaneous scales of love and related phenomena with the passage of time, those who perceived greater change also showed relatively greater actual change. Participants who loved more at one time point also tended to love more at a later time point, but the perception of an increase in love at one time point did not predict degree of love experienced at a later time point. This study also provided interesting findings about emotions involved in the breakup process. After a breakup, people perceived that their commitment and satisfaction, but not their love, declined in the period prior to the breakup. Furthermore, the perception that love had stopped increasing (or was decreasing) was moire predictive of final relationship status than actual scores on love and related phenomena. In the study of relationship change, it may be just as important to ask the partners about their perception of change as it is to assess change through scores on contemporaneous measures at multiple time points.

References

Aron, A., & Aron, E. (1986). Love and the expansion of self: Understand- ing attraction and satisfaction. New York: Hemisphere.

Berg, J. H., & McQuinn, R. D. (1986). Attraction and exchange in con- tinuing and noncontinuing dating relationships. Journal of Personality and Social Psychology, 50, 942-952.

Braiker, H. B., & Kelley, H. H. (1979). Conflict in the development of close relationships. In R. L. Burgess & T. L. Huston (Eds.), Social exchange in developing relationships (pp. 135-168). New York: Aca- demic Press.

Cate, R. M., Huston, T. L., & Nesselroade, J. R. (1986). Premarital relationships: Toward the identification of alternative pathways to mar- riage. Journal of Social and Clinical Psychology, 4, 3-22.

Cate, R. M., & Lloyd, S. A. (1992). Courtship. Newbury Park, CA: Sage. Duck, S. (1982). A topography of relationship disengagement and disso-

lution. In S. Duck (Ed.), Personal relationships: Vol. 4. Dissolving personal relationships (pp. 1-30). New York: Academic Press.

Felmlee, D., Sprecher, S., & Bassin, E. (1990). The dissolution of intimate relationships: A hazard model. Social Psychology Quarterly, 53, 13-30.

Hendrick, S. S. (1988). A generic measure of relationship satisfaction. Journal of Marriage and the Family, 50, 93-98.

Honeycutt, J. M. (1993). Memory structures for the rise and fall of personal relationships. In S. Duck (Ed.), Individuals in relationships (pp. 60-86). Newbury Park, CA: Sage.

Honeycutt, J. M., Cantrill, J. G., & Allen, T. (1992). Memory structures for

CHANGES IN RELATIONSHIPS 53

relational decay: A cognitive test of sequencing of de-escalating actions and stages. Human Communication Research, 18, 528—562.

Honeycutt, J. M., Cantrill, J. G., & Greene, R. W. (1989). Memory structures for relational escalation: A cognitive test of the sequencing of relational actions and stages. Human Communication Research, 16, 62-90.

Huesmann, L. (1980). Toward a predictive model of romantic behavior. In K. Pope (Ed.), On love and loving (pp. 152-171). San Francisco: Jossey-Bass.

Karney, B. R., & Bradbury, T. N. (1997). Neuroticism, marital interaction, and the trajectory of marital satisfaction. Journal of Personality and Social Psychology, 72, 1075-1092.

Karney, B. R., & Coombs, R. H. (1998). Memory bias in evaluations of close relationships: Consistency or improvement? Manuscript submitted for publication.

Knee, C. R. (1998). Implicit theories of relationships: Assessment and prediction of romantic relationship initiation, coping, and longevity. Journal of Personality and Social Psychology, 74, 360—370.

Lund, M. (1985). The development of investment and commitment scales for predicting continuity of personal relationships. Journal of Social and Personal Relationships, 2, 3—23.

Mathes, E. W., & Wise, P. S. (1983). Romantic love and the ravages of time. Psychological Reports, 53, 839-846.

McFarland, C , & Ross, M. (1987). The relation between current impres- sions and memories of self and dating partners. Personality and Social Psychology Bulletin, 13, 228-238.

Metts, S., Sprecher, S., & Cupach, W. (1991). Retrospective self-reports. In B. Montgomery & S. Duck (Eds.), Studying interpersonal interaction (pp. 162-178). New York: Guilford Press.

Murray, S. L., & Holmes, J. G. (1993). Seeing virtues in faults: Negativity and the transformation of interpersonal narratives in close relationships. Journal of Personality and Social Psychology, 65, 1 OH—122.

Murray, S. L., Holmes, J. G., & Griffin, D. W. (1996). The benefits of positive illusions: Idealization and the construction of satisfaction in close relation- ships. Journal of Personality and Social Psychology, 70, 79-98.

Murstein, B. I. (1987). A clarification and extension of the SVR theory of dyadic pairing. Journal of Marriage and the Family, 49, 929—933.

Ross, M. A. (1989). The relation of implicit theories to the construction of personal histories. Psychological Review, 96, 341—357.

Rusbult, C. E. (1983). A longitudinal test of the investment model: The development (and deterioration) of satisfaction and commitment in heterosexual involvements. Journal of Personality and Social Psychol- ogy, 45, 101-117.

Rusbult, C. E., & Martz, J. M. (1995). Remaining in an abusive relation- ship: An investment model analysis of nonvoluntary dependence. Per- sonality and Social Psychology Bulletin, 21, 558—571.

Simpson, J. A. (1987). The dissolution of romantic relationships: Factors involved in relationship stability and emotional distress. Journal of Personality and Social Psychology, 53, 683-692.

Sprecher, S. (1994). Two sides to the breakup of dating relationships. Personal Relationships, 1, 199-222.

Sprecher, S., & Felmlee, D. (1992). The influence of parents and friends on the quality and stability of romantic relationships: A three-wave longi- tudinal investigation. Journal of Marriage and the Family, 54, 888-900.

Sprecher, S., & Felmlee, D. (1993). Conflict, love, and other relationship dimensions for individuals in dissolving, stable, and growing premarital relationships. Free Inquiry in Creative Sociology, 21, 115-125.

Sprecher, S., & Felmlee, D. (1997). The balance of power in romantic heterosexual couples over time from "his" and "her" perspectives. Sex Roles, 37, 361-379.

Surra, C. A. (1990). Research and theory on mate selection and premarital relationships in the 1980s. Journal of Marriage and the Family, 52, 844-865.

Van Lange, P. A. M., & Rusbult, C. E. (1995). My relationship is better than—and not as bad as—yours is: The perception of superiority in close relationships. Personality and Social Psychology Bulletin, 21, 32—44.

Received July 15, 1997 Revision received May 30, 1998

Accepted August 4, 1998