Journal of Personality and Social Psychology 1990, Vol. 59, No.


Copyright 1990 by the American Psychological Association, Inc. 0022-3514/90/$00.75

Influence of Attachment Styles on Romantic Relationships
Jeffry A. Simpson
Texas A&M University
This investigation examined the impact of secure, anxious, and avoidant attachment styles on romantic relationships in a longitudinal study involving 144 dating couples. For both men and women, the secure attachment style was associated with greater relationship interdependence, commitment, trust, and satisfaction than were the anxious or avoidant attachment styles. The anxious and avoidant styles were associated with less frequent positive emotions and more frequent negative emotions in the relationship, whereas the reverse was true of the secure style. Six-month follow-up interviews revealed that, among those individuals who disbanded, avoidant men experienced significantly less post-dissolution emotional distress than did other people.

In recent years, a growing number of researchers have become interested in the processes by which people develop, maintain, and dissolve affectional bonds within close relationships (see Bretherton, 1985; Clark & Reis, 1988). Empirical research in this area was spawned by the pioneering theoretical work of John Bowlby (1969,1973,1980), who sought to determine how and why infants become emotionally attached to their primary caregivers and why they often experience emotional distress when physically separated from them. Bowlby identified a clear sequence of three emotional reactions that typically occur following the separation of an infant from its primary caregiver: protest, despair, and detachment. Given the remarkably reliable nature of this sequence across a variety of different species, Bowlby developed a theory of attachment grounded in evolutionary principles. Specifically, he argued that an attachment system composed of specific behavioral and emotional propensities designed to keep infants in close physical proximity to their primary caregivers might have been selected during evolutionary history. By remaining in close contact with caregivers who could protect them from danger and predation, infants who possessed these attachment propensities would have been more likely to survive to reproductive age, reproduce, and subsequently pass these propensities on to future generations. Empirical research examining tenets of Bowlby's theory has focused mainly on different styles or patterns of attachment in young children. Ainsworth, Blehar, Waters, & Wall (1978) have identified three primary attachment styles: anxious/ambivalent
This research was supported by a grant from the Computing Services Center at Texas A&M University. I thank Lana Aron, Sheri Baker, Holly Bogart, Debbie Grudrein, Mindy Hall, Lisa Hutchins, Jim Lyon, Paul Nicholai, Karen Owens, Faith Short, Sarah Sloan, Nicole Streetman, and Richard Williams for their assistance in Phase 2 of this study. I extend a special thanks to Margaret Lerma for her invaluable assistance in both phases of this research. Finally, thanks go to Steven W Gangestad, Caryl Rusbult, and several anonymous reviewers for their helpful feedback on earlier versions of the manuscript. Correspondence concerning this article should be addressed to Jeffry A. Simpson, Department of Psychology, Texas A&M University, College Station, Texas 77843.

(characteristic of infants who intermix attachment behaviors with overt expressions of protest and anger toward the primary caregiver when distressed), avoidant (characteristic of infants who avoid the caregiver and exhibit signs of detachment when distressed), and secure (characteristic of infants who successfully use the caregiver as a secure base when distressed). During social development, people presumably construct internal affective/cognitive models both of themselves and of typical patterns of interaction with significant others (e.g., Ainsworth et al, 1978; Bowlby, 1973; Main et al, 1985). These mental models are believed to organize the development of personality and to guide subsequent social behavior. People who possess a secure attachment style tend to develop mental models of themselves as being friendly, good-natured, and likable and of significant others as being generally well intentioned, reliable, and trustworthy. Those who display an anxious style tend to develop models of themselves as being misunderstood, unconndent, and underappreciated and of significant others as being typically unreliable and either unwilling or unable to commit themselves to permanent relationships. And those who have an avoidant style typically develop models of themselves as being suspicious, aloof, and skeptical and of significant others as being basically unreliable or overly eager to commit themselves to relationships. A growing body of empirical research has documented the existence ofthese mental models in adults (Collins & Read, 1990; Feeney & Noller, 1990; Hazan & Shaver, 1987). Recent research has begun to explore whether an individual's attachment history might influence his or her attachment style toward romantic partners during adulthood (e.g., Collins & Read, 1990; Feeney & Noller, 1990; Hazan & Shaver, 1987; Kobak & Sceery, 1988; Main et al, 1985). On the basis of descriptions of the behavioral and emotional characteristics of avoidantly, securely, and anxiously attached children provided by Ainsworth et al. (1978), Hazan and Shaver (1987) developed a single-item, self-report measure of attachment style adapted to adult romantic relationships. Individuals are asked to indicate which one of the three attachment styles best characterizes their general orientation toward romantic involvements. Securely attached people indicate that they find it relatively easy


anxiously. Hence. This latter prediction. people who possess secure.1980). They also reported how frequently each of 28 different emotions were experienced within ' Avoidant people might experience emotions less frequently in their relationships. however. people who manifest an anxious style also should be involved in affectively unpleasant relationships. satisfaction. whereas those who display an avoidant style should develop emotionally distant relationships defined by lower levels of trust. people who exhibit secure. Anxiously attached people indicate that they find others are reluctant to get as close as they would like. 1969) suggest that two major dimensions underlie the experience of emotions: magnitude of affect (intense vs. and quality of communication in established relationships. 1973. negative). and satisfaction.1980) and research (e. they may not necessarily experience pronounced distress following relationship dissolution if their relationships possess less desirable qualities. (1978) revealed that securely. and often want to become extremely close to their partners. 1985). Hypotheses Concerning Nature of the Relationship Considering the different types of mental models they harbor.g. & Wall. their insecurity about the stability of relationships in general should preclude them from developing relationships characterized by high levels of trust. Thus. committed. interdependence. In view of their tendency to eschew intimacy and commitment and to maintain guarded emotional distance from their romantic partners. Waters. SIMPSON to get close to others. 1973. To the extent that they develop interdependent. 1957. Collins and Read (1990) recently provided preliminary evidence suggesting that certain attachment styles do covary with measures of trust. avoidant.1973. and those who manifest anxious styles should experience more intense distress. commitment. despite the fact that avoidant people may appear to have emotionally quiescent relationships. Although the avoidant style is characterized by behavior indicative of interpersonal detachment. Even though they may yearn to develop stable. avoidant. despair. Hypotheses Concerning Emotional Distress Following Relationship Dissolution Ainsworth et al. mild) and hedonic sign of affect (positive vs. 1983). 1983). 1987). avoidant children typically express extreme levels of detachment (reflected in emotional and behavioral withdrawal) when separated from caregivers. People who manifest an anxious style ought to exhibit considerable ambivalence toward their romantic partners. interdependence. The study of individual differences in attachment styles is likely to contribute significantly to our understanding of why close relationships vary in both their quality and their interpersonal nature (Shaver. Berscheid. several hypotheses can be generated about the nature and emotional quality of romantic relationships possessed by people who exhibit different attachment styles.1979. and detachment). 1978. Avoidantly attached people indicate that they are uncomfortable being close to others. Hazan. Despite the fact that anxiously attached people yearn for extreme levels of commitment and emotional closeness. 1987) concerning attachment processes. To test these hypotheses.g. 1988). are comfortable depending on others and having others depend on them. and satisfaction are evident. Blehar. Ainsworth et al.972 JEFFRY A. requires some qualification. two of the most important global dimensions underlying relationship development and everyday functioning (Kelley. Davitz. even though securely attached people are likely to be involved in relationships characterized by more desirable attributes. & Bradshaw. Moreover. and anxious attachment styles should develop relationships that systematically differ in their emotional tone. people who exhibit an avoidant attachment style should experience the opposite pattern of emotions. Berscheid. 1988). 1983. Bowlby. participants responded to a series of measures that served as indicators of the amount of trust. interdependence. To date. however. On the basis of previous theory (e. People who exhibit a secure style should gravitate toward and develop stable.g. Hypotheses Concerning Emotions Experienced Within the Relationship According to attachment theory (Bowlby. ..1979). they should be susceptible to experiencing frequent negative affect. Although all children tend to experience the same sequence of basic responses (protest. 1973). and anxious attachment styles should be involved in different kinds of romantic relationships (Bowlby. and are nervous when anyone gets too close. commitment. and trusting relationships. commitment. Bretherton. whereas anxious children often exhibit extreme levels of protest and anger. the positive nature of their mental models should buffer them from excessive distress after dissolution (cf. 1978.1 And given the uncertainty they harbor concerning the stability and dependability of their relationships. find it difficult to completely trust and depend on others. and satisfaction. frequently worry that their romantic partners don't really love them or won't remain with them. commitment. adults who exhibit avoidant attachment styles toward romantic partners should experience less emotional distress following relationship dissolution. Kobak & Sceery. and don't worry about being abandoned or about someone becoming too emotionally close to them. 1969. supportive relationships in which relatively high levels of trust. interdependence. interdependent relationships typically experience stronger emotional distress following relationship termination (Simpson. Several emotion theorists (e. it also is characterized by the presence of and potential for frequent negative emotions (Ainsworth. During Phase 1. according to attachment theory (Bowlby. people who display a secure attachment style should be involved in emotionally pleasant relationships characterized by frequent occurrences of mild and intense positive emotion and by less frequent occurrences of mild and intense negative emotion (cf. Block. supportive relationships. no one has examined whether different attachment styles may precipitate different levels of interdependence and commitment. and satisfaction that existed in their current relationship. and avoidantly attached children display different emotional reactions following separation from adult caregivers. Hazan & Shaver. Past research has shown that people involved in close. 144 dating couples were studied longitudinally.

for women. 1989). three conceptually related yet empirically distinct measures were used to assess it: Rubin's Love Scale (Rubin. elated. mild positive quadrant: calm. 1977). Method Phase 1 Data Collection Participants and Procedure One hundred and forty-four dating couples (144 men and 144 women). the measure posRempel et al. determine the internal reliability of each style. ing both members of each dyad rate 13 sentences contained within the Insecurity also may play a critical role in the formation and developHazan and Shaver (1987) adult attachment measure on Likert-type ment of trust in relationships (Berscheid & Fei. wanted/cared slight alterations in wording designed to reduce item-response diffifor.78). Cronbach's alpha was . Even though Hazan and Shaver's attachment measure was develtrust was assessed by two sets of measures: the Trust Scale (which oped according to descriptions of avoidantly. couples were contacted by telephone to determine whether they were still dating and..ATTACHMENT 973 the relationship. jealous. respectively. rangwhich discrete classes underlie adult individual differences are rare. First. Internal analyses revealed belonging to one of three mutually exclusive attachment categories that all four measures correlated highly (for both men and women. Participants indicated how often they relations between multiple-item measures designed to assess specific. rs without indicating the extent to which the chosen category characterranged from . intense gnettes were decomposed into 13 individual sentences. the three Hazan and Shaver (1987) attachment viof emotions—intense positive emotion. principal-axis factor analyses izes them. 1977). measurement of individual differences in adults traditionally has focused on linear Frequency ofEmotion Index. r = .82. Measures Trust measures. satisfied. Finally. the Dependency Scale (Fei & Berscheid. similarity of values. Trust in a romantic partner tends to be a multidimensional construct consisting of at least three components: predictAttachment style measures. Moreover. ranging from strongly disagree (1) to strongly agree (7). Each emotion was rated on 7-point Likert-type scales. sad. participated in this investigation. Second. dependability. worried. and optimistic. & Archer. 1970). Cronbach's alpha was . including dating partners. happy.74). and faith. 1985) and the Insecurity Scale (Fei & Berscheid. Internal analyses revealed that all three measures were highly correlated (for both men and women. 1983). people classify themselves as Insecurity Scale. 1985) and the Investment Scale (Rusbult. Couples reported to a large experimental room in groups of 5 to 15. and disaccording to how they typically felt toward romantic partners in gengusted." (f) "I'm somewhat uncomfortable being too close to others. and mild negative emotion—were represented. Cronbach's alpha exceeded . The experimenter guaranteed that all responses would be confidential and that they would not be revealed to anyone.g. 1987). Six months later. hostile. The 28 emotions used were selected so that all four major quadrants presumed to underlie the experience Accordingly.83.41 to . Seven tapped the biases. 1985)." (1) "I rarely worry about my partner® leaving me.53 to . each of which negative emotion. Commitment measures." (b) "I'm not very guilty. (1978). Cronbach's alpha exceeded . was responded to on a 7-point Likert-type scale. as part of Phase 2. at least one member of whom was enrolled in introductory psychology at Texas A&M University.64). Accordingly. Cronbach's alpha exceeded . attractiveness." Items a-e were taken from Hazan and Shaver's "secure" vignette description. surprised. Simpthat can be conducted.. Couples then were physically separated to ensure that they could not communicate while completing the survey." (h) "I'm nervous whenever anyone gets too close to me. having others depend on me. To control for acquiescence response Seven emotions tapped the intense positive quadrant: excited." (g) "I find it difficult to trust others completely. scales. 3 sentences were worded in a negative direction. and this desire sometimes scares them away. For each of the trust subscales.82. kindness. respectively. all gorical classification method assesses an individual's standing on only factor loadings exceeded 60). As a result. 1985). loyalty. and faith (Rempel. and anxiously consists of three subscales: predictability." (c) "I'm comfortable . depressed. Satisfaction was assessed by an 11 -item. attached infants provided by Ainsworth et al. For the 1987. rs ranged from . Furthermore. Hendrick & Hendrick. 1980). Fourth." (e) "I don't like people getting too close to me.7 years. joyful. Third. eral: (a) "I find itrelativelyeasy to get close to others. if not.m were taken from the "avoidant" and "anxious/ambivalent" vignettes. ing from never (1) to very often (7). fearful.85. this catesures loaded highly on the same factor (for both men and women. how much emotional distress each partner experienced following the breakup. For all three scales. experienced each of 28 different emotions within their current relacontinuously distributed constructs (Jackson. delighted. securely. Two measures.87). therefore. Attachment style was measured by havability.83. sesses some unfortunate psychometric properties (see Hazan & Shaver. Internal analyses indicated that these two measures were highly correlated (for men. Mean length of dating relationship was 13. Given that interdependence is a multidimensional construct (Berscheid. 1983)." (j) "Others often arereluctantto get as close as I would like. The mean age of men and women was 19. & Zanna. irritated. Items f-i and j .70. one attachment style despite the fact that some adults may be best Satisfaction measure." (d) "I rarely worry about being abandoned by others. needed. And seven assessed the mild negative quadrant: rejected. In addition." and (m) "I often want to merge completely with others. and passionate. principal-axis factor analyses conducted separately on men and women indicated that all three measures loaded highly on the same factor (for both men and women. Over 92% of the couples had dated for more than 1 month. were used to assess commitment: the Commitment Scale (Lund. and the Self-Disclosure Scale (Miller. Instances in tionship. this method does not allow one to son." (i) "Others often want me to be more intimate than I feel comfortable being. meaningful individual difference variability conducted separately on men and women indicated that all four meathat exists within each category cannot be assessed. content. Berg. this method of valid measure of satisfaction with the partner on 11 different dimenclassification places severe limits on the types of statistical analyses sions (e. serene. For each scale. Commitment tends to be reflected in both avowed intentions to remain in a relationship as well as in extrinsic and intrinsic investments made within it (Kelley. 1977).5 months. The experimenter told participants they would be asked to complete an initial questionnaire survey that inquired about their dating relationship. mild positive emotion. comfortable having to depend on other people. 1971).4 and 18. Participants rated the following items rant: angry. all factor loadings exceeded . Seven assessed the intense negative quadculty were made for 2 sentences. r=. distressed. facecharacterized as a blend of two or more styles. and lonely. She then passed out the survey in number-coded packets so that each dyad could be identified later for purposes of data analysis. dependability." (k) "I often worry that my partners) don't really love me. Holmes. Interdependence measures. disappointed.

and women who scored lower on the anxious index were dating men who scored higher on the secure index. so that women who possessed an avoidant style tended to be somewhat more anxious. however. Moreover. commitment. composite indexes of each dimension. Both members of 132 dyads (91. and faith in the partner and in lower levels of insecurity in the relationship).2 Relations Among Attachment Indexes All three indexes were then correlated both within and between men and women. the items corresponding to each vignette were identified. trust.67%) were successfully reached. keyed in the proper direction. people who scored higher on the avoidant index reported that they were involved in relationships defined by lesser amounts of interdependence. Higher scores reflected greater security. two factors emerged. Additional items. all comparisons of effect sizes between the three attachment indexes must be made bearing their differential reliabilities in mind. everyday functioning and routine?" Participants responded to these three items on 8-point scales. there was no relation between scores on the avoidant and anxious indexes." and "Who initiated the breakup: you. The second factor appeared to tap an anxious-nonanxious dimension. and trust correlated highly within each domain." and (c) "How upset were you immediately after the breakup?" Participants responded to these three items on 7-point scales." and (f) "How long did the breakup disrupt your typical.36% of those contacted) no longer were dating. everyday functioning and routine. or your partner?" negatively correlated with those on the avoidant index. that the secure and anxious indexes were somewhat less reliable than was the avoidant one. The Avoidant Attachment Style Index was developed by aggregating Items f through i (Cronbach's alpha = . Measures Intensity and Duration of Emotioned Distress Index. results are reported for all three attachment indexes.91). ranging from no time at all (1) to more than 2 months (8). Note.51). Once all scales were keyed in the proper direction. to what extent did it disrupt your typical. commitment. Men who scored higher on the anxious index indicated that 2 Factor analyses also were performed on all 13 attachment items. Higher scores reflected greater anxiousness (Cronbach's alpha = . Conversely. men who scored higher on the avoidant index were dating women who scored somewhat higher on the anxious index. Three items assessed the duration of distress: (d) "How long did it take you to make an emotional adjustment following the breakup. Participants also answered two additional questions: "When (during the 6-month period) did you stop dating this person.974 Phase 2 Data Collection Participants and Procedure JEFFRY A. Telephone interviewers first determined whether the couple was still dating. Within both men and women.79)." (b) "Immediately after the breakup occurred. scores on the Secure Attachment Style Index were somewhat negatively correlated with those on the Anxious Attachment Style Index. composite indexes were constructed to provide more reliable estimates of the strength of association between each domain and the three attachment styles. greater trust (reflected in the greater predictability of. ranging from not at all (1) to a great deal/extremely (7). Higher scores indicated greater distress. Higher scores indicated greater avoidance. indicating that greater security was associated with slightly less anxiousness. dependency on.59). correlations between these composite indexes and the three attachment styles revealed that people who scored higher on the secure attachment index indicated that they were involved in relationships characterized by greater interdependence (reflected in greater love for. and satisfaction. within each sex. a slight positive relation emerged between these two styles. The Secure Attachment Style Index was constructed by aggregating Items a through e (Cronbach's alpha = . These people then responded to a telephone survey that assessed the extent of emotional distress they experienced following relationship dissolution. however. several attempts were made to contact all participants by telephone. As reported in Table 1. greater commitment (evidenced by greater commitment to and investment in the relationship). 1987)." (e) "How long were you upset after the breakup. For men. dependability of. Attachment and Relationship Indexes: Individual-Level Analyses Because the measures designed to assess interdependence. Results Phase 1 Construction of Attachment Indexes To construct continuous measures of each attachment style. Measures designed to tap each domain were then aggregated to form more reliable. and greater satisfaction. 48 couples (36. more reliable index (Cronbach's alpha=. For women. the dyadic effects were relatively small in magnitude. and aggregated to form three continuous attachment indexes. Given that the general pattern of findings did not differ depending on whether two or three attachment styles were examined (see footnote 3) and in view of strong theoretical reasons for attempting to identify three distinct attachment styles. The first factor appeared to tap a secure-avoidant dimension. so that greater security was indicative of less avoidance. According to independent reports provided by both partners. Because scales that possess lower reliabilities yield more attenuated effects. At the level of the dyad. The Anxious Attachment Style Index was created by aggregating Items j through m. each measure was transformed into a common metric (through z-score transformation) within the sexes. two statistically reliable effects emerged. both you and your partner. Scores on the secure index were very . Three items assessed the intensity of distress: (a) "Immediately after the breakup. and self-disclosure with the partner). All six items were then aggregated to form a single. Two marginally reliable effects also emerged: Women who scored higher on the secure index were dating men who scored lower on the avoidant index. SIMPSON Approximately 6 months later. As reported in Table 2. By and large. how difficult was it for you to make an emotional adjustment. Men who scored higher on the anxious attachment index were involved with female dating partners who scored somewhat lower on the secure index. The intensity and duration of emotional distress was assessed by a six-item index (Simpson.

20** Secure .28**** .26*** .09 .19** -. Attachment and Relationship Indexes: Dyad-Level Analyses An examination of dyadic effects (i. FAVD = avoidant attachment style: women. and less satisfaction.12 -.23"* -. MSEC 2.27*** -. four composite subscales representing each of the four emotion -. All correlations are two-tailed.29*"* Women Anxious -. MAVD 4.28"" .19" . Critical value at alpha = . As displayed in Table 3.15* -. Data is presented for correlations within each sex and within the dyads. 0 1 . relationship length was partialed out of all correlations between the three styles and the four composite indexes.22*** .31*"* -.31"" -.22" -.23*" -.25*" -.28"" . * " p < .17 -. And women who scored higher on the avoidant index were dating men who reported less commitment.37**** . and avoidant attachment styles.35**** . MANX 3. although in the anticipated direction. Women who scored higher on the anxious index reported being involved in relationships denned by less commitment and less trust. MANX = anxious attachment style: men.24 — Note. commitment.05 -. women who scored higher on the secure index were dating men who reported greater commitment to the relationship and less insecurity in the relationship. To ensure that these results were not attributable to systematic differences in length of relationship.25*** -. and less satisfaction.30 — -.05.31**** -.37"" -.34"" -21*** Note.36"" .15* -.15* .05 .11 -.28"" -. commitment.15* .19" -.30**" -. n= 144 women and n = 144 men.14 -. less trust. FSEC 5.38**** .25*** .30**" -. And men who scored higher on the avoidant index tended to be dating women who reported less trust. FAVD 1 — 2 -.12 -. MAVD = avoidant attachment style: men.21" -.e. Men who scored higher on the anxious index tended to be dating women who reported less interdependence and less commitment.ATTACHMENT 975 Table 1 Pearson Product-Moment Correlations for Three Attachment Styles Variables 1. anxious.26*** . FANX 6.10. Trust.25*** .27**** -.14 . less commitment.26*" .27*** . Zero-order correlations between the three attachment styles and the specific measures used to assess interdependence. Effects for interdependence and commitment were marginally significant.22"* . greater insecurity.12 5 6 in Table 2 continued to be reliable once relationship length was controlled for.17" -. As presented in Table 4.23*** Anxious -. were not reliable. and satisfaction and vice versa) revealed several reliable within-dyad effects.14* -.37"** .15* -.13 .48**** -. and trust corroborate these findings.05 is .28**** -. All significant effects reported Table 2 Pearson Product-Moment Correlations Between Attachment Styles and Indicators of Interdependence. MSEC = secure attachment style: men.23*** .28**** -.66 .27"" -.29**** -.45**** -. Attachment and Frequency of Emotion: Individual-Level Analyses To assess the emotional nature of the relationships of people who exhibited secure.23"* -.24*** -.11 -.16* -.17 (two-tailed).54"** -.23*** Avoidant -22*** -.15* -. men who scored higher on the secure index tended to be involved with female partners who were less insecure. and greater insecurity.18 -.23*** -. correlating each male's attachment style scores with his female partner's standing on each indicator of interdependence.07 -. *"*/?<.001. FANX = anxious attachment style: women.65 .10 -.16* -.24*** .09 Avoidant -. Effects for interdependence and satisfaction. Commitment.20** .21** -. and Satisfaction Attachment style Men Variable Interdependence measure Love Dependency Self-Disclosure Composite index Commitment measure Commitment Investment Composite index Trust measure Predictability Dependability Faith Insecurity Composite index Satisfaction measure Secure . they were involved in relationships characterized by less trust and less satisfaction. . « = 144 men and n = 144 women. FSEC = secure attachment style: women. **p<.32 3 4 . trust.38"" . Women who scored higher on the anxious index were involved with men who reported less interdependence.40**** -.25*** -.26*** .27*" -. *p<.12 .27*** .

19** .09 -.16* .05.20** . To examine this possibility.34**** -.19** -.09 .001..23*** .03 -. less mild negative emotion. Commitment.30**** Avoidant -.19** -. and more mild positive emotion. All four subscales were reliable (all Cronbach's alphas > .14* -. **p<.11 -.09 .22*** -.20** . All correlations are two-tailed. Composite subscales assessing the extent of emotion evident in the other three quadrants were constructed in a similar manner.10 -.79).13 -.11 -. And those scoring higher on the avoidant index were involved with women who experienced mild positive emotions less often.19** -.16* -. These effects.001.20** -. rather than a single categorical one. quadrants were constructed. Commitment. People who scored higher on the secure index reported experiencing reliably less intense negative emotion. in which people place themselves into one of three mutually exclusive attachment categories. are not specific to the use of three continuous measures of attachment. enced intense positive emotions less regularly.01 ****/><.10.01 -.01 -.07 . All correlations are two tailed.15* .14* -. Trust.3 3 The effects reported in Tables 2-7 could be attributable to the use of three continuous attachment measures.07 .09 -. The composite subscale reflecting the extent of intense positive emotion was constructed by aggregating the seven items designed to tap this quadrant.29**** _32».10. The effect for intense negative emotion among women was marginally reliable.10 -.09 -.09 .976 Table 3 JEFFRY A. ***p<. n = 144.04 .19** -.06 -. more intense positive emotion. n = 144. Table 7 reveals that women who scored higher on the secure index were dating men who reported more frequent intense positive emotion in the relationship.16* -.04 -.12 -. correlating each male's attachment style scores with his female partner's scores on each composite emotion subscale and vice versa) revealed several reliable within-dyad effects. When length of relationship was partialed out of the correlations between the three attachment indexes and the four emotion subscales.13 -.15* . all significant effects reported in Table 5 remained reliable.10 -.35***» -22*** -. As displayed in Table 6.11 Note.08 -. more mild negative emotion. Note.23*** -. **p<. Those scoring higher on the anxious index were involved with men who experienced intense positive and mild positive emotions less frequently.19** -. Each individual then was placed into a single attachment category on the basis of their highest standardized score on the three continuous attachment indexes.07 . The four subscales were then correlated with each attachment style index.14* Anxious -.32**** -.25*** -. Given the factor analytic resultsreportedin footnote 2.06 -.15* .16* .02 . ****/>< . Specifically. less intense positive emotion.13 .20** -. ***p<. In particular. The pattern offindingsand magnitude of effects closely corresponded to those reported for the three continuous measures.16* -. Trust. men who scored higher on the secure index tended to be dating female partners who reported experiencing more frequent mild positive emotions in the relationship. a set of two-dimensional analyses paralleling the three-dimensional ones also were conducted.07 .05 .e. each of the 13 attachment items that loaded highly on thefirstfactor (the 5 secure and 4 avoidant ones) were keyed in the same direction and were aggregated to form thefirstdimension (reflecting level of security/lack of avoidance).09 Anxious -29**** -. A series of one-way analyses of variance (treating attachment category as the independent variable and the four composite measures of relationship quality and the four composite subscales of emotion as the dependent measures) was conducted within each sex.01 .13 . therefore.03 . People who scored higher on the avoidant and anxious indexes reported experiencing reliably more intense negative emotion. Men scoring higher on the anxious index tended to be involved with women who experi- . a parallel set of analyses were conducted.32**** -.03 Avoidant -.05.17** -.09 -. Schefie tests (Kirk. and less mild positive emotion. Following this.00 -.22*** -. all three continuous measures of attachment were standardized (through z-score transformations) within each sex. *p<.07 -.01. Results are displayed in Table 5. */><. and Satisfaction Reported by Their Female Partner Attachment style Variable Interdependence measure Love Dependency Self-Disclosure Composite index Commitment measure Commitment Investment Composite index Trust measure Predictability Dependability Faith Insecurity Composite index Satisfaction measure Secure .05 -.06 . each of the 13 attachment items that loaded highly on the second factor (the 4 anxious items) were aggregated to form the second Attachment and Frequency of Emotion: Dyad-Level Analyses An examination of dyadic effects (i.20** -. 1982) conducted on each dependent measure confirmed the reliability of these findings. SIMPSON Table 4 Pearson Product-Moment Correlations Between Men's Attachment Styles and Indicators of Interdependence.« -.23*** Pearson Product-Moment Correlations Between Women's Attachment Styles and Indicators of Interdependence. and Satisfaction Reported by Their Male Partner Attachment style Variable Interdependence measure Love Dependency Self-Disclosure Composite index Commitment measure Commitment Investment Composite index Trust measure Predictability Dependability Faith Insecurity Composite index Satisfaction measure Secure .

p<.34.01. or who initiated the breakup. highly avoidant men tend to experience less prolonged and intense emotional distress following relationship termination. ?(46) < 1.15* . Discussion On the basis of derivations of attachment theory.21** -. a reliable relation still existed between the avoidant and emotional distress indexes. for both sexes. People who exhibit a secure attachment style tend to be involved inrelationshipscharacterized by higher levels of interdependence. respectively). People's scores on the Intensity and Duration of Emotional Distress Index were not reliably associated with either when they stopped dating the partner during the 6-month period.10. Neither the secure nor the anxious style was reliably correlated with the index of emotional distress both rs(46) < .32**** . commitment. whereas those who display anxious and avoidant styles experience the opposite pattern. 44) < 2. No reliable effects emerged for women.28**** Note. *p<.33**** Anxious Avoidant Secure Women Anxious Avoidant -.39**** .46**** -. Among men. the three attachment styles are strongly associated with different patterns of emotional experience within relationships.37**** .25*** .33) than for women (r = . ns. commitment. A test for independent correlations (Cohen & Cohen. commitment. these effects are not specific to the use of three versus two attachment dimensions.31**** -. Men who scored higher on the avoidant index experienced significantly less emotional distress following dissolution. three hypotheses concerning the relations between adult attachment styles and phenomena occurring within romantic relationships were examined. Specifically. Table 6 Pearson Product-Moment Correlations Between Mens Attachment Styles and Emotions Reported by Their Female Partner Attachment style Emotion Secure Anxious Avoidant -.41**** . r(46) = -.26*** .20** -.10 -. p < .22*** 44**** -.001.08 dimension (reflecting level of anxiousness). and satisfaction. p<.23*** .44**** -.12 Mild Positive Composite subscale Intense Negative Composite subscale -. ****/>< . F(l.12 Mild Negative Composite subscale .03. Ms = 24. the avoidant style reliably predicts the strength of emotional distress experienced after relationship dissolution. Furthermore.22*** -. <(94) < 1.10.01. ns. men and women did not differ in the amount of emotional distress experienced follow- ing dissolution. trust. All correlations are two-tailed.33.28**** -22*** .02. each style was correlated with scores on the Intensity and Duration of Emotional Distress Index.30**** .06.14* Intense Positive Composite subscale -. trust. Highly avoidant men might experience less emotional distress because they tend to be involved in short-term dating relationships characterized by lower amounts of interdependence.ATTACHMENT 977 Table 5 Pearson Product-Moment Correlations Between Attachment Styles and Emotions Attachment style Men Emotion Intense Positive Composite subscale Mild Positive Composite subscale Intense Negative Composite subscale Mild Negative Composite subscale Secure . Hence. 14. p < .001. r(46) = -. **/><.01. All correlations are two-tailed.5.32**** -.05. trust. ns. Internal analyses correcting for attenuation that was due to unreliability of the measures revealed that none of these nullresultswere attributable to the unreliability of either the emotional distress or attachment indexes.08). 1983) revealed that the correlation between emotional distress and the avoidant style was reliably larger for men (r = -. *••/?<. for both sexes. z=2. */7<.02 -. .05. anxious. Analyses revealed that the pattern and magnitude of effects for the two-dimensional analyses were highly similar to those found using the three continuous attachment measures across all dependent measures.09 -.31**** .31**** -.19** -. n = 144 women and n = 144 men. ns. Yet when the effects of the length of their relationship and their scores on the composite indexes of interdependence.77 and 25. ***•/><.23*** . *•*/><. Individual-level analyses revealed that secure. ns. and satisfaction. and avoidant attachment styles tend to be associated with romantic relationships that differ in their qualitative nature. all rs(46) <. n = 144. Phase 2 Emotional Distress Following Dissolution To determine the relations between the three attachment styles and the extent of emotional distress experienced after relationship dissolution. People who manifest a secure style have relationships characterized by more frequent occurrences of positive emotion and less frequent occurrences of negative emotion.21** -. Moreover.02.00 Note.05. and satisfaction were simultaneously partialed out.05 . Those who exhibit insecure styles—particularly highly avoidant people— tend to have relationships defined by the opposite set of features. .

.15* Attachment Styles and Relationship Quality People who possessed avoidant and anxious attachment styles reported being involved in a relationship characterized by less interdependence. If the three styles primarily stem from unique person-situation interactions. the precise origins of these effects cannot be determined. SIMPSON although not impervious to change (see Main et al. Attachment researchers have discussed the three styles interchangeably. committed involvements. and satisfaction. Collins & Read.01. This finding replicates the work of Collins and Read (1990).15* -. • p < . 1990. 1977). who revealed that the extent of anxious attachment displayed by women assumes a greater role in affecting global satisfaction within relationships.10 -. The interdependence and commitment results extend our knowledge concerning what other critical indicators of relationship functioning reliably covary with the three styles. the specific causal pathways may be different for anxious and avoidant people. Intense Positive Composite subscale . for instance. Hazan & Shaver. Within-dyad correlations did reveal that women who were anxious were dating more avoidant men. This pattern of results corroborates previous research (e.35**** Mild Positive Composite subscale . one would expect certain styles to correlate highly with one another within couples. highly avoidant people may produce heightened distrust in their romantic partner. By eschewing closeness and commitment. Nonetheless. they probably indicate that an individual's attachment style has less impact on the partner's perceptions of the relationship than it has on his or her own perceptions. Men's standing on the anxious style did not covary with their female partner's level of satisfaction with the relationship. and trustworthiness. More important. whether their partner's lower satisfaction generates their anxious behavior. 1985).19** -. and men who were anxious were involved with less secure women.978 Table 7 Pearson Product-Moment Correlations Between Women's Attachment Styles and Emotions Reported by Their Male Partner JEFFRY A. Over time. revealed an interesting trend.12 . **p<. Feeney & Noller. ****p<. Within-dyad results also revealed one reliable sex difference.26*** Intense Negative Composite subscale -. sometimes using them to describe features of people and other times using them to characterize features of relationships (cf. One salient issue in the adult attachment literature concerns whether individual differences in attachment style are traitlike in nature or whether they reflect unique person-situation interactions. This pattern of results is not surprising given that the quality and emotional tenor of relationships should be more strongly and directly influenced by one's own attachment style rather than the style adopted by one's partner. their partner's lack of interdependence/commitment has made them anxious. The frequency of emotion findings transport the study of adult romantic attachment into a new domain. trust. Interestingly. Highly avoidant people reported their relationship was somewhat less interdependent and committed than did highly anxious people.08 . highly avoidant people ought to elicit anxious propensities from their romantic partner and vice versa. one more closely linked to recent conceptualizations of attachment as a system governing affect regulation in relationships (Sroufe & Waters. Finally. This finding also is in accord with research indicating that female possessiveness tends to be negatively associated with male satisfaction. but male possessiveness does not tend to be associated with female satisfaction (Davis & Oathout. n = 144. which has shown that avoidant people tend to be preeminently concerned about avoiding excessive intimacy and commitment in relationships and that anxious people tend to be extremely preoccupied with issues surrounding their partner's predictability.10 Note. the emotional distress results address one of the most fundamental predictions stemming from attachment theory. Conversely. Highly anxious people. All correlations are two-tailed. Given the correlational nature of this investigation. Hazan & Shaver. Dyadic effects were less robust than were individual-level ones. Attachment style Emotion Secure Anxious Avoidant -. who found similar associations between the three attachment styles and measures of trust and satisfaction. commitment. may have partners who report less interdependence and commitment for several different reasons. 1 0 .16* Mild Negative Composite subscale -. or whether additional variables account for these relations.11 . Although these findings may reflect the fact that withinindividual measures shared common method variance that was not shared by the within-dyad measures. highly anxious people may generate decreased closeness and commitment from their partner. or they habitually become involved with partners who are less able or willing to develop close. whereas highly anxious people reported their relationship contained somewhat less trust. ***p<. The anxious style was associated with somewhat less partner-reported interdependence and commitment. These effects.g. 1990. 1987). 1987). including that their anxious behavior reduces their partner's level of interdependence/commitment. whereas women's standing on the anxious style was negatively correlated with their male partner's degree of satisfaction. are not highly contingent on the specific style displayed by current romantic partners. 1987).05. whereas the avoidant style was associated with less partner-reported trust. How does this investigation contribute to our understanding of attachment phenomena within adult romantic relationships? The relationship quality findings clarify and extend those of Collins and Read (1990). however. Considered together. by displaying a general lack of trust. Closer examination of these effects. Future research must determine whether the behavior of anxious women elicits lower satisfaction in their male partner. however. for example. within-dyad results revealed an opposite pattern of effects.001. however. these findings suggest that individual differences in attachment style. dependability.08 . they did not increase significantly in size with relationship length. were small in magnitude.

1989. Thus. the secure style tends to be associated with more adaptive styles of love (Hendrick & Hendrick. Third. Men tend to be more conflict avoidant and to typically suppress strong emotions. 1973). 4 . intense positive emotion as reported by the partner. 1988) as well as greater self-esteem.. are described as less ego resilient and more hostile. 1977) have suggested that attachment theory should be viewed more globally as a theory of affect regulation. the avoidant style appears to parallel the tendency of men to suppress emotions. Given their mental models of others as being predominantly well-intentioned. one should not necessarily expect them to be overly distraught following relationship dissolution. 1987. the attachment style an individual exhibits in a given relationship may be contingent on their own relationship history in conjunction with the style possessed by their current romantic partner. and anxious people tend to be characterized as less ego resilient and more anxious.4 For men. 1985). at least three different factors are likely to systematically influence the amount of distress experienced: attachment style. Although not definitive. and less anxiety (Feeney & Noller. these results may be limited to recently formed dating relationships. However. Kaplan.ATTACHMENT 979 Attachment Styles and Emotionality The frequency of emotion effects reported in Phase 1 were fairly large. this investigation revealed that partners' scores on the three attachment indexes were predominantly orthogonal. the inclination to suppress negative emotional events (George et al. whereas the anxious style appears to parallel the proclivity of women to more openly express them. Main etal. This finding is noteworthy because it provides additional evidence that the nature of an individual's attachment style is not merely a function of the quality of their most recent romantic relationship. defensively inhibit negative emotional experiences (George. however. Nevertheless. the lack of emotional distress experienced by avoidant men appears to stem from their attachment style per se. and the quality of the former relationship. Avoidant people. attachment styles have been conceptualized and discussed as possessing relatively stable. and trustworthy (Collins & Read. they often appear to adopt the same style in different relationships Attachment Styles and Postdissolution Emotional Distress Highly avoidant men reported experiencing significantly less postdissolution distress. Hazan & Shaver. For avoidant women. particularly given the modest reliability of the measures. Caveats Several caveats must be considered when interpreting these findings. Feeney & Noller. 1990. 1978). Wells. Given the tendency of avoidant people to On the face of it. we do not propose that the concept and use of discrete attachment categories be abandoned. 1980). According to this perspective. was not witnessed for women. & Alvarez. however. They also should be more adept at translating negative affect into positive affect (Bowlby. despite the fact that compelling psychometric reasons exist for treating attachment styles as continuous dimensions and despite the fact that virtually identical findings emerged regardless of whether attachment styles were treated as discrete or continuous measures. 1987). in relation to avoidant and anxious ones (Kobak & Sceery. Future research must examine how positive and negative emotions differentially affect the quality of ongoing relationships. one might suggest that anxious women ought to experience reliably more postdissolution distress than other people. Furthermore. whereas women tend to be more conflict confronting and to openly vent strong emotions (Gottman & Levenson. Moreover. 1990. Recent research suggests that secure people do tend to have more affectively positive and well-adjusted relationships. First. even though people can experience changes in attachment style over time and across generations within families (Main et al. securely attached people should be predisposed to cope more effectively with negative affect in relationships than should their insecurely attached counterparts. 1978. Hazan & Shaver. Secure people are described by their peers as being more ego resilient. greater selfconfidence. perhaps through selective or biased attributional processes that allow them to habitually perceive and interpret stressful events in a positive. Peplau. and not necessarily the presence of negative emotion.. 1985. the avoidant attachment style continued to reliably predict the degree of emotional distress even when measures of relationship length and quality were controlled for. traitlike qualities. reliable. these findings seem to suggest that the absence of positive emotion. although sound theoretical reasons exist for hypothesizing that attachment styles may directly affect the quality and emotional tenor of relationships. Among both men and women. for instance. Hazan & Shaver. gender. highly avoidant men should experience less postdissolution distress. Although this argument is a plausible one. Kelley et al. 1985. and less hostile. within romantic relationships. 1985) should be at odds with the propensity on the part of women to express strong affect. 1983). different styles of attachment may reflect different patterns of ego resilience. 1987. Because anxious women tend to be involved in relationships characterized by less desirable qualities. If this argument is viable. less anxious. 1988). Finally. Main et al. Second. Within-dyad correlationsrevealedthat the three attachment styles tended to covary more strongly with how frequently partners experienced positive rather than negative emotions in the relationship. on the other hand. Such an effect. higher scores on the anxious style werereliablyassociated with less frequent.relationship-enhancingmanner (Harvey.g. 1985) coupled with the tendency of men to typically suppress strong affect. & Main. Some investigators (e. Sroufe & Waters. may serve as one principle source of discontent underlying the relationships of anxiously attached people. One explanation for this disparate pattern of results stems fromresearchon sex differences in emotion during interpersonal conflict. or the ability to effectively regulate and mitigate negative affect when it arises in established relationships (see Block & Block. 1990). these correlational data do not permit causal inferences. Levy & Davis. 1988. This scenario may account for the lack of relation between the avoidant attachment style and the extent of postdissolution distress within women. rather than from the less desirable qualities of their romantic involvements.

1. Annual Review ofPsychology. M. San and relationship quality in dating couples. Christensen. L. Huston. H. J.. 3-23. Rempel. 413-484).135-146. P. In I. & Fei. In A. L. Rusbult. In G. Lonscales for predicting continuity of personalrelationships. 16. sciences (2nd ed. & Davis. Monographs ofthe Society for Research in Child Develtheory and research. E. J. 48. The dynamics of structured personality tests: 1971. R. analysis for the behavioral sciences. L. A. childsion. C. In R. and adulthood: A move to the level of representation. & Hendrick.). H. 49. 39.. 2. & Read. R. E. Social and Personal Relationships. A. 281-291. Peplau. pp.. 209).). 5. N. Grisham. Fei.). ing conflict in close heterosexual pairs..1234-1244. Wells. M. Francisco: Freeman. (1985). of Personality and Social Psychology. The language of emotion. 2. McClintock. SIMPSON Harvey.. Waters (Eds. Harvey. Collins. R. Kidd (Eds. Perceived dependency. & Alvarez. N. Kelley. Hazan. Hazan. Hillsdale. Unpublished manuscript. The attachment interview Simpson. E. Lund.Journal of don: Tavistock. 182-200). J. N. L. Levinger. (1983). Bretherton. P. & Archer. Interpersonal attraction. (1977). H. (1988). Attachment and loss: Vol. Clanton & L. . Journal of Personality and Social Psychology. 95romantic relationships: Empathy and relational competence. 683-692. In J. hood. (1970). E. (1983).439Bowlby. others. New York: Basic Books.. Close relationships (pp. Berkeley. Romantic love conceptualized as an attachment process. Hendrick. BersBerscheid.. Roles and gender. New Haven. University of Minnesota. C. 59. Berscheid. NJ: Erlbaum. Jour112. & Reis. Emotion. J. Lovestyles and attachment styles Bowlby. Berscheid. Monterey. Development. Close relationships (pp. & Bradshaw. 3. J. The making and breaking of affectional bonds. Growing points of attachment and research. & Cassidy. B. 220-264). The role of ego-control and ego-resilKobak. Attachment and loss: Vol. L. J. & Main. Waters (Eds. (1982). & Wall. J. Experimental design: Procedures for the behavioral Abnormal and Social Psychology. Journal of Kirk. G.. Z. New York: compared: Theirrelationsto each other and to various relationship Basic Books. CT: W e University in heterosexual relationships: The eternal triangle. C. (1988). Romantic love and sexual jealousy. Peplau. characteristics. Huston. Sex differences in comments made durtice-Hall. In H. Cunningham. E. E.. (1983). George. 50(1-2. 55. Berg. R. 53. J. Studies in the phenomenology of emotions.. adult romantic relationships. (1978). Kelley. R. pp. A. A. 473-491. Journal ofPersonality and Francisco: Freeman.. W (1988). Child Development. In H. Love and commitment. S. & Zanna. (1971). In P. & Oathout. E. Philadelphia: Multilingual Received November 3.1990 • (Hindy & Schwartz. H. T. 50(1-2. M. (1980). R. J. T.1184-1199. 58. the Second International Conference on Personal Relationships. insecurity. R. J. (1985). R.. Journal of Personality and Social Psychology. S. Noller & M. Commitment and satisfaction in romantic associations: A test of the investment model. M. Sternberg & M. New York: Random House. The dissolution of romanticrelationships:Factors involved inrelationshipstability and emotional distress. Jealousy. N. CA: Acaand Social Psychology. S. I. (1985). Growing points of attachment theory I. Miller. G. Lindzey & E. Attachment in late adolescence: iency in the organization of behavior. M.980 JEFFRY A. G. Blehar. NJ: Erlbaum. A. L. S. San Diego. (1980). Openers: Individuals Clark. (1985). 39-101). E. Applied multiple regression/correlation Peplau. Sroufe. H. Waters.). M. and love integration of three behavioral systems. A. Future research must track people longitudinally to unequivocally establish whether attachment styles remain stable across several relationships. 1984). Levinger. & D. & Noller. A. 229-248. (1979). D. 56. Bowlby. Archeid.. Huston. J. J. (1983). WI. (1980). E. & Block. Serial No. E. Handbook ofsocial psychology (3rd ed. Attachment style as a predictor of cial Psychology. Holmes. Harvey. 265-273. McClintock. W Ickes. J. I. 397-410. (1988).). Fitzpatrick (Eds. 4. anger. C. A.511-524. (1988). J. NJ: Erlbaum.). H. 209). & Shaver. Berscheid. H. Lefebvre. (1969). 52. J. & Levenson. (1987). working models. Peterson. T. Social Psychology. Bretherton & E. In G. Kelley. (1989). Journal ofPersonality and Social relationships. Journal of Personality and Social PsyShaver. Hillsdale. M. & Berscheid. Unpublished Press. M. 44.. J. Berscheid. (1990). 13. NJ: PrenC. Hillsdale. C. Block. A. Revision received June 7. F. manuscript. Attachment. Hillsdale. J. G. 265-314). J. 358-363. W Collins (Ed. C. Psychological Review. (1987). Sink. E. M. J. K. Adult attachment. (1990). MinneWorking models. Christensen. ChrisMadison. 235-260). Peterson. E. E. Research on love: Does it meaReferences sure up? Journal of Personality and Social Psychology. (1973). Attachment and loss: Vol. K.). Englewood Cliffs. Journal ofExperimental SoFeeney. In H. A. D. (1978). Hindy.1990 Harvey. & Cohen. M. H. & Sceery. Attribution in the context of conflict and separation in close relationships. J. Journal of Personality Davitz. tensen. NJ: Erlbaum. J. Perspectives on marital interaction (pp. A. G. Maintenance of satisfaction in relationships. C. & R. J. nal of Personality and Social Psychology. (1985). D. 784-794. S. Interpersonal processes in close who elicit intimate self-disclosure. Peplau. (1983). T. University of California. 110-168). Cohen. 609-672. L. L. logy of marriage. C. (1969). (1984). New directions in attribution research (Vol. 53. and representations of self and sota Symposia on Child Psychology (Vol. Individual differences in the Ainsworth. R. Patterns of tendency toward anxious romantic attachments. P. Measurement of romantic love.Main. H. Rubin. G. Child Development. H. A. pp. demic Press. (1957).172-186. J. J. McClintock. 66-104. Journal for adults. onson (Eds. & Schwartz.). Serial No. 644-663.. Loss: Sadness and depres. 16. & Waters. H. Levinger. E. H. San Francisco: Freeman. Sex Roles. The development of investment and commitment Bowlby. Kaplan. L. In Bretherton & E. Attachment theory: Retrospect and prospect. J. & Yablon. (1977). (1985). Kaplan. Smith (Eds. San Jackson. Paper presented at attachment. Minneapolis. Block. D. 2. R. Peterson. H. L. The social psychophysioconstruct. G. (1977). Psychology. affect regulation. CA: Brooks/Cole. P. Harvey. A. The anatomy oflove. (1978).1989 Matters. Monographs ofthe Society for Research in Child opment. P. 54. 3-35. Barnes (Eds. New York: Basic Books. Journal ofSocial and Personal Relationships. Security in infancy. J. (1987). M. Kelley.). Trust in close Davis. H. Kelley. & D. G. E. L. M. 78. Love as attachment: The chology. Close relationships (pp. & D. Separation: Anxiety and 471. A. Accepted July 2. Levy. Attachment as an organizational Gottman.

Sign up to vote on this title
UsefulNot useful

Master Your Semester with Scribd & The New York Times

Special offer for students: Only $4.99/month.

Master Your Semester with a Special Offer from Scribd & The New York Times

Cancel anytime.