This study examined the hypothesis that the three behavioural systems of attachment, caregiving, and sexual mating first described by Bowlby (1969/72) each make a unique contribution to intimacy in romantic relationships.

One hundred and nine undergraduate psychology students from Griffith University were administered questionnaire measures of attachment style, caregiving style, sexual interaction, sexual communication satisfaction, attachment function, and psychosocial intimacy. The measures were analysed using an hierarchical multiple regression. Results indicated that attachment, caregiving, and sexuality each make a unique contribution to intimacy, accounting for 68.7% of the total variance (Adj.R² = .665). Unique significant contributions to intimacy were made by the attachment style/dimension of "Confidence"; the caregiving styles of "Proximity vs Distance" and "Sensitivity vs Insensitivity"; "Sexual Communication Satisfaction"; and the attachment function measure of "Attachment to Partner". The best predictor of intimacy in romantic relationships was "Attachment to Partner".

The significance of these findings will be discussed in light of the methodological limitations of this study. Suggestions for further research addressing these limitations will also be made.

Attachment Theory

Attachment theory was developed by Bowlby (1969/1972) who proposed that attachment behaviour is organised primarily on a goal-corrected basis. Attachment behaviour evolved to mitigate conditions of danger or threat by ensuring that close proximity is maintained between infants and their caregivers. According to Bowlby, children develop working models of the self and others by internalising their experiences with caretakers over time. These early attachment relations come to form a prototype for later adult relationships. Two key features define the internalised "working models" of attachment behaviour. The first is the child’s image of other people, and the second is the child’s image of the self. Ainsworth, Blehar, Waters, and Wall (1978) significantly extended Bowlby’s work and identified three patterns of infant-caregiver attachment behaviour. They labelled these patterns secure, anxious/ambivalent, and avoidant, and found that these were related to caregivers’ accessibility and responsiveness.

A basic principle of attachment theory is that the working models developed in early childhood continue to be important throughout the lifespan (Bartholomew & Horowitz, 1991). In recent years more research has been carried out to determine whether Bowlby’s theory of childhood attachment applies equally well to adulthood attachment (e.g, Bartholomew & Horowitz, 1991; Feeney, Noller, & Hanrahan, 1994; Hazan & Shaver, 1987).

Conceptualising the pair bond: Attachment, Caregiving and Sexuality

Hazan and Shaver (1987) first conceptualised romantic love as an attachment process. This pivotal study found that adults could be classified into three categories corresponding to childhood attachment styles. More importantly, it was found that secure, anxious/ambivalent, and avoidant individuals utilised three different love styles. Shaver, Hazan, and Bradshaw (1988) found similar features and dynamics between infant-caregiver attachment and adult romantic love. They hypothesised that the experience of establishing a pair bond (Money, 1980) based on the notion of romantic love, is made up of the integration of three behavioural systems — attachment, caregiving, and sexuality (see Figure 1). Of the three behavioural systems, attachment develops first, adapting to whatever caregiving environment is encountered. Caregiving, the second behavioural system to emerge, is learned by modeling the behaviour of the primary attachment figure. Finally, in early adulthood, the sexual system becomes more fully developed. When the caregiving system and the sexual system become fully developed, the working models of the individual’s social world, constructed during the early years of childhood, are likely to be activated (Shaver, Hazan, & Bradshaw, 1988). In this way, adult functioning in the pair bond is, to a certain degree, influenced by early attachment.

More recently, a four-category model of attachment styles based on Bowlby’s (1969/1972) conceptualisation of the two internal working models held by an individual has been proposed by Bartholomew and Horowitz (1991).They hypothesised that if a person’s abstract image of the self, and a person’s abstract image of the other are dichotomised with positive and negative poles, then four combinations of attachment style can be conceptualised. Bartholomew and Horowitz labelled the four attachment patterns secure, preoccupied, dismissing, and fearful, and posited that the dimensions could be defined in terms of dependency on a horizontal axis and avoidance of intimacy on a vertical axis (see Figure 2). More recent studies assessing adult attachment style (Feeney, Noller, & Hanrahan, 1994), communication and satisfaction in the early years of marriage (Feeney, Noller, & Callan, 1994), and caregiving in romantic relationships (Kunce & Shaver, 1994), have provided support for the two-dimensional model of adult attachment.

Feeney, Noller and Hanrahan (1994) developed a five dimensional approach to adult attachment using a construct that covered the major features described in both three- and four-group models of adult attachment. They labelled these dimensions Confidence, Discomfort with Closeness, Need for Approval, Preoccupation with Relationships, and Relationships as Secondary. The Confidence dimension assesses attitudes to both self and others, and clearly represents individuals who have a positive self- and other-model of their social world. The other four dimensions represent particular aspects of Bartholomew and Horowitz’s (1991) model of adult attachment. The advantage of this approach to adult attachment is that it provides a clearer delineation of the attachment groups using discrete dimensional profiles rather than attachment styles (Feeney et al., 1994).

Figure 2. Model of adult attachment (after Bartholomew and Horowitz, 1991).

While most of the attachment literature has focused on attachment style based on internalised working models, attachment can also be examined in terms of its function, that is, what "set-goals" attachment behaviour is designed to achieve. Bowlby (1969/1972) described four attachment functions: proximity seeking (approaching, staying near, and making contact with a significant other), safe haven (turning to a significant other for comfort, support, or reassurance), separation protest (resisting or being distressed by separation from a significant other), and secure base (using a significant other as a base from which to engage in non-attachment behaviour). Hazan and Zeifman (1994) posited that these attachment functions are essentially the same in infancy and adulthood, that is, in the development of reciprocal attachments in adulthood, the attachment functions emerge in the same sequence as they do in infant-caregiver relationships.

Although Hazan and Zeifman (1994) suggest that proximity seeking has sexual attraction as a primary motivating force in late adolescence and adulthood, proximity seeking can also be interpreted as being analogous to intimacy as both are conceptualised in terms of a desire for closeness and connection with a romantic partner. Moreover, the "set-goals" of the other three attachment functions, that is, seeking comfort, support, and reassurance (safe haven), resisting or being distressed by separation from a loved one (separation protest), and using a loved one as a safe base from which to extend the boundaries of the relationship (safe base), can be conceptualised as being indicative of some of the intimacy themes in a romantic relationship, especially the themes of closeness, and the interdependence of partners.

Kunce and Shaver (1994) defined caregiving as the provision of care, that is, noticing and responding to another’s needs, as opposed to the seeking of support. Using the two attachment dimensions proposed by Bartholomew and Horowitz (1991), Kunce and Shaver found that individuals who had positive working models of the self and others (i.e., those individuals who are comfortable with intimacy and autonomy) scored significantly higher on the caregiving scales of Proximity vs Distance, and Sensitivity vs Insensitivity than individuals who were dismissing or fearful of intimacy or who were preoccupied with relationships. Scores on the Compulsive Caregiving scale for individuals comfortable with intimacy were significantly lower than individuals who were preoccupied with relationships. However, there were no significant differences on these scores between individuals comfortable with intimacy and those who were either dismissing or fearful of intimacy. Scores for the four groups were not significantly different on the Cooperation vs Control scale. Kunce and Shaver’s findings therefore indicate that provision of proximity and sensitivity characterise caregiving behaviour for individuals who are comfortable with intimacy and closeness more so than individuals who are not comfortable with intimacy.

The third behavioural system discussed by Bowlby (1969/1972) is that of sexual mating. In contrast to infant-caregiver attachments which are complementary, adult attachment relationships are expected to be reciprocal and are assumed to be formed primarily with sexual partners (Hazan & Zeifman, 1994). In the course of a long-term sexual relationship, attachment between partners tends to build up, and the attachment and caregiving components interact to create a reciprocal relationship (Ainsworth, 1989). Thus sex becomes an integral part of attachment in the course of normative development.

Much of the attachment literature dealing with the sexual component of the pair bond deals almost exclusively with the biological nature of sexual mating, concentrating on different types of sexual behaviours in relation to attachment style (e.g. Hazan, Zeifman & Middleton, 1994; Shaver, 1994). Therefore the links between intimacy and the sexual mating component of the pair bond are not immediately obvious. However, the intimacy literature makes specific links between sexual involvement and intimacy. According to Schaefer and Olson (1981), one common notion of intimacy is that the greater the sexual involvement between partners, the more intimacy is present in the relationship. They also suggest that intimacy is a process that occurs over time thus making a link to the development of intimacy in a relationship.

Waring, Tillman, Frelick, Russell, and Weisz (1980) also make an explicit link between intimacy and sexuality. Their study revealed that sexuality is considered part of intimacy by most people although it is not considered to be the primary component. Moreover, sexual satisfaction was often defined as a consequence of commitment and compatibility more than any specific sexual behaviour. Nevertheless sexuality was the third most frequently reported aspect of a feeling of intimacy by husbands and wives in the study.

Rationale, Aims and Hypotheses

This study sought to elucidate a number of themes linking intimacy, and attachment in adult romantic relationships. The significant themes that have emerged are: (1) individuals with a positive working model of the self, and a positive working model of the other are more comfortable with intimacy, tend to be more confident in their relationships, and score higher on the caregiving scales of proximity and sensitivity; (2) the biologically-based behavioural systems of attachment and caregiving contribute to intimacy in loving relationships. Moreover, sexuality, the third behavioural system variable can also be considered to contribute to intimacy given the number of studies that have linked the two constructs; and (3) the presence of all four attachment functions in a romantic relationship is indicative of a fully formed attachment bond between the partners.

The main aim of this study was to investigate the relative contributions made by attachment, caregiving, and sexuality as predictors of intimacy in adult romantic relationships. Although Shaver, Hazan and Bradshaw (1988) consider attachment and caregiving to be possible aspects of intimacy, no study has yet assessed the predictive power of the three behavioural systems of attachment, caregiving, and sexuality on intimacy in romantic relationships.

A secondary aim of this study was to examine the unique contribution of attachment function to intimacy in romantic relationships.

More specifically it is predicted that:

1. Of the 5 attachment dimensions of the Attachment Style Questionnaire (ASQ), the ‘Confidence’ dimension will make a significant unique contribution to intimacy as measured by the Psychosocial Intimacy Questionnaire (PIQ) while the ‘Discomfort with Closeness’ dimension will be negatively related to intimacy as measured by the PIQ.
2. Of the 4 caregiving styles of the Caregiving Questionnaire, the ‘Proximity vs Distance’ and ‘Sensitivity vs Insensitivity’ styles will make a significant unique contribution to intimacy as measured by the PIQ.
3. Of the aspects of sexuality examined, the ‘Pleasure Mean’ and the derived ‘Sexual Behaviour Index’ measures of the Sexual Interaction Inventory (SII) and the Sexual Communication Satisfaction measure will make a significant unique contribution to intimacy as measured by the PIQ.
4. Attachment function assessed by the ‘Attachment to Partner’ score on the WhoTo Inventory will be make a significant unique contribution to intimacy as measured by the PIQ.

Participants

One hundred and nine undergraduate Griffith University psychology students (women = 81, men = 28) participated in the study. Various attempts were made to recruit more males for the study, but they were more reluctant to participate than females. The sample over-represents females (74.3% of the sample) and beginning students (67.9% of the sample), and under-represents males (25.6% of the sample) and continuing students (32.1% of the sample).

Participation in the study was dependent on a number of criteria. First, the participants had to be at least 18 years of age. Second, the participants had to be currently involved in a heterosexual romantic relationship of at least three months’ duration with one partner. Third, as research involving the assessment of sexual behaviours suggests that relationship changes could occur if both members of a pair bond keep records of their sexual activity (e.g., LoPiccolo & Steger, 1974; Mann, 1973), only one partner of the pair bond was allowed to participate in the study and the questionnaire was not to be discussed with the partner whilst being completed.

The majority of participants (64%) were aged between 18 and 23. The mean time in the current relationship was 3.73 years (SD = 4.16) with a range of three months to 19.83 years. The vast majority of respondents (82.9%) reported having no children.

All participants were asked to complete a set of questionnaires in booklet form. The following questionnaires were included in the booklet:

1. The Demographics Questionnaire included personal information (e.g. gender, age group, education), and relationship information (e.g. length of current relationship, relationship status, number of children, frequency of intimate activities).
2. The Quality of Marriage Index (Norton, 1983) is a 6-item scale that measures global relationship satisfaction. Feeney, Noller and Callan (1994) reported the reliability coefficient of the QMI as .88 indicating that the scale is highly reliable.
3. The Attachment Style Questionnaire (Feeney, Noller, & Hanrahan, 1994) is a 40-item measure that assesses an individual’s attachment style on five dimensions – "Confidence" (in self and others), "Discomfort with Closeness", "Need for Approval", "Preoccupation with Relationships", and "Relationships as Secondary". Feeney, Noller and Hanrahan reported reliability coefficients alphas for the five scales as .80, .84, .79, .76, and .76 respectively indicating good realiability.
4. The WhoTo Questionnaire (Hazan & Zeifman, 1994 adapted by Feeney & Hohaus) is a 12-item scale that measures the attachment functions of "Proximity Seeking", "Safe Haven", "Secure Base", and "Separation Protest" in a relationship as well as overall "Attachment to Partner". This is a relatively new instrument currently undergoing validity and reliability testing. An overall Cronbach’s alpha of .95 was obtained in this study.
5. The Sexual Interaction Inventory (LoPiccolo and Steger, 1974) is a 102-item measure that assesses self- and partner-acceptance, as well as aspects of sexual behaviours including dissatisfaction with the frequency and range of behaviours, pleasure obtained from sexual activities, knowledge of partner’s preferred sexual activity, and total dissatisfaction with the sexual relationship. A "Sexual Behaviour Index" (number of sexual behaviours engaged in * quantity of each behaviour engaged in) was calculated for this study as a measure of overall sexual behaviour. LoPiccolo and Steger reported internal consistency coefficients ranging from .85 to .93 indicating good internal consistency for all the SII scales.
6. The Sexual Communication Satisfaction Questionnaire (Wheeless, Wheeless, and Baus, 1984) is a 22-item measure that assesses satisfaction with communication about sexual behaviour and willingness to communicate about sex with one’s partner. Internal reliability was reported at .94 for the 22-item scale.
7. The Caregiving Questionnaire (Kunce and Shaver, 1994) is a 32-item measure designed to assess the four caregiving dimensions of "Proximity vs Distance", "Sensitivity vs Insensitivity", "Cooperation vs Control", and "Compulsive Caregiving". Cronbach’s alpha reliability coefficients reported were .83, .83, .87, and .80 respectively.
8. The Psychosocial Intimacy Questionnaire (Tesch, 1985) is a 60-item measure that assesses intimacy in friendship, dating or marital relationships via a total intimacy score. Tesch reported Cronbach’s alpha for the PIQ with an opposite-sex sample as .98.

A number of other questionnaires were also administered but were not relevant to this paper.

Participants were contacted via e-mail and during regularly scheduled lecture times. Information was provided about the sensitive nature of some of the questionnaire items included in the study and the need to sign an informed consent form prior to receiving the questionnaire booklet. Upon receipt of a signed informed consent form, the participants were given the questionnaire booklet and a self-addressed envelope in which to enclose the completed questionnaire. No information which could identify any participant was required to be included on the questionnaire. Average time to complete the questionnaire was approximately 45-50 minutes.

The Statistical Package for the Social Sciences (SPSS), version 6.1.1. for the Power Macintosh was used for all data analysis. Prior to analysis, SPSS DESCRIPTIVES and SPSS FREQUENCIES were used to examine the data for missing values, and fit between the variables’ distributions and the assumptions of multivariate analysis. SPSS CORRELATIONS were used to calculate the Pearson product-moment correlations between selected demographic variables and the PIQ Total Intimacy Index score, as well as the predictor variables and the PIQ Total Intimacy Index score.

A four-step hierarchical multiple regression was conducted to test the first hypothesis regarding attachment, caregiving, sexuality and intimacy. The PIQ Total Intimacy Index score was the dependent variable. The predictor variables were entered as follows. The first step entered the Attachment dimension of "Confidence"; the second step entered the Caregiving styles of "Proximity vs Distance" and "Sensitivity vs Insensitivity"; the third step entered the Sexuality measures "Pleasure Mean," Sexual Behaviour Index," and "Sexual Communication Satisfaction"; and the fourth step entered the assessment of Attachment function "Attachment to Partner."

Preliminary Analysis

Pearson product-moment correlations were calculated between the PIQ Total Intimacy Index score and the following demographic variables: gender, age group, length of relationship, number of children, education level, and religiosity. These correlations were obtained so that any demographic variables related to intimacy could be controlled in subsequent regression analyses. No significant correlations were found.

Pearson product-moment correlations were then calculated between the PIQ Total Intimacy Index score and the five Attachment dimension measures, the four Caregiving measures, the four Sexual Behaviour measures, the Sexual Communication measure, and the Attachment function measure. The results are presented in Table 1.

In order to maintain the experiment-wide error for this study at p = .05, the unweighted Bonferroni correction (a /number of comparisons) to statistical significance levels was utilised, making the critical alpha level p = .002. The variables that met this criterion were: Confidence (r = .35, p < .001), Proximity (r = .57, p < .001), Sensitivity (r = .52, p < .001), Cooperation (r = .36, p < .001), Sexual Communication Satisfaction (r = .52, p < .001), Sexual Behaviour Index (r = .37, p < .001), Pleasure Mean (r = .37, p < .001), and Attachment to Partner (r = .73, p < .001).

Attachment to Partner is singular with the attachment function subscales of Proximity Seeking, Safe Haven, Separation Protest, and Secure Base, as it is calculated by summing the subscale scores. For this reason the subscale scores were not entered into the hierarchical multiple regression equation, even though they had moderate to strong correlations with the PIQ Total Intimacy Index score (rs = .56 to .71, ps < .001). Due to the effects of singularity, entering them into the equation would reduce the degrees of freedom for error and thus weaken the analysis (Tabachnick & Fidell, 1989).

Cooperation was not entered into the hierarchical multiple regression even though it was significantly correlated with the PIQ Total Intimacy Index score, because it was not hypothesised to be a reliable predictor variable on intimacy. The remaining variables with significant correlations were selected for entry into a hierarchical multiple regression analysis. Hierarchical multiple regression was chosen based on a developmental theoretical perspective which suggests that throughout the lifespan, attachment precedes caregiving and sexuality, and that a "full-blown" attachment to a romantic partner (consisting of all four attachment functions) develops last of all. The variables were therefore entered in this order into the multiple regression equation.

For the hierarchical multiple regression the statistical assumptions of normality, linearity, homoscedasticity, and independence of residuals were not violated. No outliers were identified and multicollinearity and singularity assumptions were not violated. With 109 respondents and seven independent variables, the cases-to-IV ratio is 15.6:1, which is above the minimum requirements for multiple regression analysis (Tabachnick & Fidell, 1989).

Table 2 displays the final summary table for the hierarchical multiple regression analysis. R was significantly different from zero at the end of each step. After step four, with all independent variables in the equation 68.7% of the variance (Adj. R² = .665) was accounted for (F(7, 100) = 31.307, p < .001).

In summary, the results from the hierarchical multiple regression show that the attachment dimension of "Confidence," the caregiving styles of "Proximity vs Distance" and "Sensitivity vs Insensitivity," the "Sexual Communication Satisfaction," and the assessment of attachment function "Attachment to Partner" all make a unique contribution to the PIQ Total Intimacy Index score. Examination of the b weights indicates that "Attachment to Partner" makes the greatest contribution to the PIQ Total Intimacy Index score, followed by "Proximity vs Distance," "Sensitivity vs Insensitivity," "Confidence," and "Sexual Communication Satisfaction." Together, these predictor variables accounted for 66.5% of the variance of the PIQ Total Intimacy Index score.

Summary of Results

The hypothesis that intimacy in romantic relationships can be predicted from attachment, caregiving, and sexuality variables was supported. The best predictor of the PIQ Total Intimacy Index score as indicated by its beta coefficient, was Attachment to Partner. This suggests that the extent to which a partner meets the four attachment functions of Proximity Seeking, Safe Haven, Separation Protest, and Secure Base is a strong indicator of the level of intimacy in the relationship. Other significant predictors of the PIQ Total Intimacy Index score were Proximity, Confidence, Sensitivity, and Sexual Communication Satisfaction. A surprising finding was that measures of sexual behaviour such as the Pleasure Mean and the Sexual Behaviour Index did not make a unique contribution to intimacy in romantic relationships. This seems to go against a number of conceptualisations of intimacy as involving a mutually satisfactory sharing of sexual activity (Schaefer & Olson, 1981), sexual expression (Waring et al., 1980), or implying a sexual dimension (Parks & Floyd, 1996).

The finding that the Attachment dimension of "Confidence" made a unique contribution to intimacy lends support to Feeney, Noller and Hanrahan’s (1994) findings that confident individuals have positive self- and other-models and therefore correspond to Hazan & Shaver’s (1987) and Bartholomew & Horowitz’s (1991) "Secure" adult attachment style. The underlying theme of the "Secure" attachment style is someone who is comfortable with intimacy and autonomy.

The finding that the Caregiving styles of "Proximity vs Distance" and "Sensitivity vs Insensitivity" both made unique contributions to intimacy lends support to Kunce and Shaver’s (1994) findings that individuals who scored significantly higher on provision of proximity and sensitivity were those who had positive self- and other-models (and thus were comfortable with intimacy, closeness, and autonomy). The themes of proximity or closeness, and the interdependence of partners also underlie contemporary views on intimacy (e.g., Clark & Reis, 1988). Proximity to a loved one has also been hypothesised to be the first requirement for the establishment of a romantic relationship (Hazan & Zeifman, 1994).

The finding that Sexual Communication Satisfaction made a unique contribution to the prediction of intimacy in romantic relationships lends support to Wheeless et al.’s, (1984) proposal that sexual communication satisfaction is associated with intimate relationships. The communication of information about sexual behaviours is also linked to the concept of self-disclosure or expressiveness that underlies the construct of intimacy. Waring et al.’s (1980) study found that sexual satisfaction in the sense of commitment and compatibility appears to influence intimacy more than any specific sexual behaviour.

The finding that the attachment function assessed by "Attachment to Partner" made a unique contribution to intimacy indicates that individuals who reported having a "full-blown" attachment with their romantic partner scored more highly on the PIQ Total Intimacy Index score. Although Hazan & Zeifman (1994) did not directly address the concept of intimacy in their study, the components of attachment (proximity seeking, safe haven, separation protest, and secure base) that they used to define a "full-blown" attachment to a romantic partner, have elements that underlie contemporary views on intimacy, such as closeness, supportiveness, feeling distressed when separation occurs, and the development of autonomy (Perlman & Fehr, 1987). "Attachment to Partner" was also the best predictor of the PIQ Total Intimacy Index score, indicating that there are higher levels of intimacy present in relationships that contain all four attachment functions.

These findings taken together suggest that, as hypothesised, the behavioural systems of attachment, caregiving, and sexuality that Shaver, Hazan and Bradshaw (1988) conceptualised as the components of the pair bond, are also components of intimacy in romantic relationships. Therefore, the pair bond can be conceptualised in terms of the amount of intimacy that exists in a romantic relationship.

The present study contributed to the current knowledge of romantic relationships by examining the roles of attachment, caregiving, and sexuality in predicting intimacy in romantic relationships. The most important implication to emerge from this study is that support was found for the hypothesis that intimacy in a romantic relationship can be predicted from the attachment dimension of "Confidence," the caregiving styles of "Proximity vs Distance" and "Sensitivity vs Insensitivity," the "Sexual Communication Satisfaction" and the assessment of attachment function "Attachment to Partner." This study was the first attempt to investigate all three behavioural systems (attachment, caregiving, and sexuality) simultaneously.

The limitations of this study were predominantly methodological. One of the major concerns for this study is that a cross-sectional design rather than longitudinal design was used. A longitudinal design facilitates the study of changes people go through during the course of relationship development. People in long-term relationships may be qualitatively different to those in short-term relationships, many of which may not continue. However longitudinal studies can be expensive, can take a long time to complete, there is difficulty keeping track of participants, subject loss and non-random assignment pose threats to the internal validity of the results, and change scores are problematic (Perlman & Fehr, 1987). Although a longitudinal design would have been preferable, the exploratory nature of this study, and time and resource constraints made this impossible.

The second major concern is that time limitations and the sensitive nature of the sexuality component of the study precluded the use of anything but anonymous self-report measures. While there exists a wide choice of instruments suitable for research into romantic relationships, the exclusive use of self-report measures, and the subjective nature of participant responses can lead to problems with common-method variance. Findings in relationship literature indicate that if people are satisfied and happy in a relationship, they tend to report every aspect of that relationship as positive and "glowing" — a kind of "halo effect." Social desirability may also mask the reporting of actual attitudes and behaviours by individuals, although the anonymity of the questionnaire tried to control for this.

A third concern was the size and composition of the participant pool. The data collected from the 109 participants who fully completed the questionnaire were satisfactory in terms of being able to conduct the analyses appropriate to the study. However, a larger sample would have improved the generalisability of the findings. Furthermore, the participant pool was wholly drawn from a university population and was biased in terms of gender and age. Moreover, the study only used one partner from a romantic relationship because of possible problems with the sexual behaviour measures. Sexual behaviour literature (e.g., LoPiccolo & Steger, 1974; Mann, 1973) suggests that merely keeping records of sexual activity leads to marked changes in sexual behaviour.

Finally, the sexual behaviour measures, especially the Sexual Interaction Inventory were explicit in nature. Although some participants may have found this part of the questionnaire offensive, the high completion rate indicated that the majority of participants coped well with the content of the questionnaire. Attendant to the area of sexual behaviour measures, is the notion of bias on the part of the individuals who participate in a study compared to those that do not or who terminate prematurely. This bias may account for the small number of men who participated in the study. Ultimately it may limit the generalisations that can be made about conclusions from the study.

The present study found that intimacy in a romantic relationship can be predicted from the attachment dimensions and functions, the caregiving style, and the sexual behaviour of individuals. This is the first time that all three behavioural systems described by Bowlby (1969/1972) have been used simultaneously to predict intimacy in romantic relationships. As this study was exploratory, employed only self-report measures, and used a cross-sectional design, future longitudinal studies are desirable to determine whether the current findings are transitory or whether they continue throughout the development of a romantic relationship.

A larger, more representative sample accounting for a broader sociodemographic population would improve the generalisability of the findings. Future studies should also consider a more homogeneous participant pool.

Future studies could also address the constraints placed on this study by the use of self-report measures. Non-self-report measures such as ratings from friends, relatives, or partners can provide a means of checking for the reliability of participant responses, especially on sensitive measures such as those used in this study to assess sexual behaviours.

The explicit nature of the sexual behaviour and satisfaction measures used in this study may have biased participant response rates. Future research should address this issue, perhaps through the use of a more sensitive measure of sexual behaviour, though the questions that need to be asked to assess sexual behaviours, by their very nature, could create response bias.

Finally, this study accessed only one partner from a relationship, due to concerns about the possible effect that the sexual behaviour measures could have if both partners in a relationship participated in the study. A number of advantages ensue from using both partners in any future studies. As mentioned earlier, partner ratings can be used to check reliability of responses to any measures used. It is also common for studies examining relationship development to use the attachment style and caregiving style of both partners to predict the dependent variable, something which could not be done in this study. Conclusions drawn from studies involving both partners in a relationship may afford more sensitivity when generalised to romantic relationships.

Ainsworth, M. D. S. (1989). Attachments beyond infancy. American Psychologist, 44(4), 709-716.

Ainsworth, M. D. S., Blehar, M. C., Waters, E., & Wall, S. (1978). Patterns of attachment: A psychological study of the strange situation. Hillsdale, New Jersey: Lawrence Erlbaum Associates.

Bartholomew, K., & Horowitz, L. M. (1991). Attachment styles among young adults: A test of a four-category model. Journal of Personality and Social Psychology, 61(2), 226-244.

Bowlby, J. (1969/1972). Attachment and loss volume 1: Attachment. Middlesex, England: Penguin.

Clark, M. S., & Reis, H. T. (1988). Interpersonal processes in close relationships. Annual Review of Psychology, 39, 609-672.

Feeney, J. A., Noller, P., & Callan, V. J. (1994). Attachment style, communication and satisfaction in the early years of marriage. Advances in Personal Relationships, 5, 269-308.

Feeney, J. A., Noller, P., & Hanrahan, M. (1994). Assessing adult attachment. In M. B. Sperling, & W. H. Berman (Eds.), Attachment in adults: Clinical and developmental perspectives (pp. 128-152). New York: Guildford Publications.

Hazan, C., & Shaver, P. (1987). Romantic love conceptualized as an attachment process. Journal of Personality and Social Psychology, 52(3), 511-524.

Hazan, C., & Zeifman, D. (1994). Sex and the psychological tether. Advances in Personal Relationships, 5, 151-177.

Hazan, C., Zeifman, D., & Middleton, K. (1994, July). Adult romantic attachment, affection, and sex. Paper presented at the 7th International Conference on Personal Relationships, Groningen, The Netherlands.

Kunce, L. J., & Shaver, P. R. (1994). An attachment-theoretical approach to caregiving in romantic relationships. In K. Bartholomew, & D. Perlman (Eds.), Advances in Personal Relationships Vol. 5 (pp. 205-237). London: Jessica Kingsley Publishers.

LoPiccolo, J., & Steger, J. C. (1974). The Sexual Interaction Inventory: A new instrument for assessment of sexual dysfunction. Archives of Sexual Behavior, 3(6), 585-595.

Mann, J. (1973). The effects of erotica. Sexual Behaviour, 3, 23-29.

Money, J. (1980). Sexosophy: The principles of erotic sexuality. In J. Money (Ed.), Love and lovesickness: The science of sex, gender differences, and pair-bonding (pp. 43-72). Baltimore: John Hopkins University Press.

Norton, R. (1983). Measuring marital quality: A critical look at the dependent variable. Journal of Marriage and the Family, 45, 141-151.

Perlman, D., & Fehr, B. (1987). The development of intimate relationships. In D. Perlman, & S. Duck (Eds.), Intimate Relationships (pp. 13-42). Newbury Park, CA: Sage.

Schaefer, M. T., & Olson, D. H. (1981). Assessing intimacy: The PAIR inventory. Journal of Marital and Family Therapy, 7, 47-60.

Shaver, P. R. (1994, August). Attachment, caregiving and sex in adult romantic relationships. APA Invited Address, Division 8, Los Angeles, CA,

Shaver, P., Hazan, C., & Bradshaw, D. (1988). Love as attachment: The integration of three behavioural systems. In R. J. Sternberg, & M. Barnes (Eds.), Anatomy of love (pp. 68-99). New Haven, CT: Yale University Press.

Tabachnick, B. G., & Fidell, L. S. (1989). Using multivariate statistics (2nd ed.). New York: HarperCollins.

Tesch, S. A. (1985). The Psychosocial Intimacy Questionnaire: Validation studies and an investigation of sex roles. Journal of Social and Personal Relationships, 2, 471-488.

Waring, E. M., Tillman, M. P., Frelick, L., Russell, L., & Weisz, G. (1980). Concepts of intimacy in the general population. The Journal of Nervous and Mental Disease, 168(8), 471-474.

Wheeless, L. R., Wheeless, V. E., & Baus, R. (1984). Sexual communication, communication satisfaction, and solidarity in the developmental stages of intimate relationships. Western Journal of Speech Communication, 48, 217-230.