Attachment Style, Friendship Quality and the Mediating Effect of Communication Skills in Young Adults Friendship

Myriad studies on friendship quality in young adults argued for the importance of investigating attachment relationships among peers, as they play a crucial role for individuals from adolescence into adulthood. Empirical research on friendship also demonstrated the important link between attachment style and friendship quality, noting that secure attachment predicted perceptions of friendship quality. This study looked specifically at the associations between attachment, communication and the quality of relationship shared between friends using four different questionnaires; Attachment Style Questionnaire (ASQ), Relationship Questionnaire, Communication Functions Questionnaire (CFQ) ), and the Network of Relationship Inventory-revised . Findings involving 110 pairs of friendship dyads indicated individuals with good communication skills tend to have better friendship quality. In addition, securely attached individuals foster higher relationship quality with their friends, but fearfully attached are appraised by friends as having low quality friendship relationship. Communication skill was found not significantly correlated to the attachment styles possibly due to factor of short length friendship between participants affecting their level of communication despite their attachment type. Overall, results of study suggest communication skills of individuals play a role in affecting friendship quality but not the attachment styles. However, the attachment styles of friends who appraise the quality of the friendship have a significant effect on the friendship quality appraisal itself. These findings are discussed in terms of their implications for work on attachment styles, communication skills and friendship quality.


Introduction
Bowlby's attachment theory (1973,1980,1982) postulated that the prototypes for adult attachment styles are based on infancy experiences with the primary care-givers which later form the internal working models (Bretherton, 1990) of individuals. Bretherton (1990) argued that these internal working models are rules, schemas or scripts that sum up a person's relationships history. Kobak and Hazan (1991) postulated that these models influence the quality of close relationships individuals develop later in life as their views of self and others determine how ones interpret and react in social situations. Batholomew and Horowitz (1991) conceptualized individuals' differences in attachment into a fourcategory description styles: the secure, preoccupied, dismissing and fearful. Various studies showed that securely attached people tend to communicate better, have higher quality friendship relationships and enjoy more intimacy and satisfaction in relationships compared to the insecurely attached group of people. Reis and Shaver (1998) showed that secure people self-disclose more than insecure individuals, leading to greater intimacy in their friendship relationships while anxious and avoidant people who have negative view of others tend to self-disclose less, leading to lower level of intimacy in their friendship relationships.
Literature search in this study found that there is a lack of studies on the influence of attachment style of friends in a dyadic friendship relationship. Reis and Shaver (1988) as well as Kerns (1994)

Predictions
This study predicted that secure individuals will communicate better and demonstrate high levels of intimacy in their friendship relationship while fearfully attached individuals will demonstrate low level of satisfaction in their friendship relationship. Further, it is expected that communication skills and attachment style of individual will be a significant element in determining friendship quality.
(1) Hypothesis 1: It was hypothesized that better communication skills produce better friendship quality. (2) Hypothesis 2: The attachment styles of individuals will have an effect on their communication skills. (3) Hypothesis 3: The attachment styles of individuals play a role in determining friendship quality. (4) Hypothesis 4: The attachment styles of friends play a role in determining friendship quality. (5) Hypothesis 5: Fearful people have lowest friendship satisfaction compared to people of other attachment styles. (6) Hypothesis 6: Secure people will have more intimate friendship relationships.

Participants
One hundred and eleven pairs of young adults participated in this study (male, N=52; female, N=59). Mean age was 23.00 years and ranged from 18 years to 30 years. The majority of participants (68.47%) were single, the rest (31.53%) were either dating or married with none being separated or divorced. Mean length of friendship was 38.93 months with a range from 3 months to 276 months. Among these dyads, 39 pairs were male dyads, 42 pairs were female dyads and 30 pairs were mixed. However, participating dyad No.63 was removed due to many missing values in the returned questionnaires.
"Exclude cases pairwise" option was selected to deal with missing values in the variables. However, "replace with mean" option was used for "communication skills" and "friendship quality" variables as there were not many missing values for these two variables. Moreover, it prevented the small sample size from being further reduced. signed an informed consent form to indicate their agreement in their voluntary participation before the commencement of the study. Participants were debriefed and thanked at end of experiment.

Results
SPSS version 19.0 was used to analyze the dyadic data collected. Of the 110 participants, 48.1% (n=52) classified themselves as securely attached, 13.0% as preoccupied (n=14), 15.8% as fearful (n=17), and 23.1% as dismissive (n= 25), using the Relationship Questionnaire (Bartholomew & Horowitz, 1991). Two of the participants did not report their attachment style and were excluded from the study.
In the current study, the cronbach alpha coefficient for (1) The relationship between communication skill (as measured by the CommSkill_F variable using the The Communication Functions Questionnaire) and friendship quality (as measured by the FriendshipQuality_F scale of The Network of Relationship Inventory-revised) was investigated using Pearson product-moment correlation coefficient. Preliminary analyses were performed to ensure no violation of the assumptions of normality, linearity and homoscedasticity. There was a strong, positive correlation between the two variables, r = .648, n = 110, p < 0.005, with high levels of communication skill associated with high levels of friendship quality.
In addition, a one-way between-groups analysis of variance was conducted to explore the impact of attachment style on levels of communication skill, as measured by the Communication Functions Questionnaire (CFQ). Subjects were divided into four groups according to their attachment style (Secure; Preoccupied; Dismissing; Fearful). There was no statistically significant difference at the p <. 05 level in CFQ scores for the four attachment groups: F(3, 104) = 1.4, p = .24. Despite not reaching statistical significance, mean scores were different between the groups. The effect size, calculated using eta squared, was .04.
A one-way between-groups analysis of variance was also conducted to explore the impact of attachment style on levels of friendship quality appraised reflectively, as measured by the Network of Relationship Inventory-revised (NRI-r). Subjects were divided into four groups according to their attachment style (Secure; Preoccupied; Dismissing; Fearful). There was no statistically significant difference at the p <. 05 level in NRI-r scores for the four attachment groups: F(3, 104) = .12, p = .95. Despite not reaching statistical significance, there was actual difference in mean scores among the groups. The effect size, calculated using eta squared, was .003. Post-hoc comparisons using the Tukey HSD test indicated that the mean score for Preoccupied Group (M = 3.26, SD = .87) was significantly different from Secure Group (M = 3.40, SD = .87), Fearful Group (M = 3.43, SD = .80) and the Dismissive Group (M = 3.38, SD = .79). Further, a one-way between-groups analysis of variance was conducted to explore the impact of attachment style of friends on levels of friendship quality, as measured by the Network of Relationship Inventory-revised (NRI-r). Subjects were divided into four groups according to their attachment style (Secure; Preoccupied; Dismissing; Fearful). There was a statistically significant difference at the p <. 05 level in NRI-r scores for the four attachment groups: F(3, 106) = 2.9, p = .04. Despite reaching statistical significance, the actual difference in mean scores between the groups was quite small. The effect size, calculated using eta squared, was .08. Post-hoc comparisons using the Tukey HSD test indicated that the mean score for Secure Group (M = 3.59, SD = .86) was significantly different from Fearful Group (M = 2.95, SD = .94). Preoccupied Group (M = 3.48, SD = .71) and Dismissive Group (M = 3.30, SD = .70) did not differ significantly from either Secure Group or Fearful Group. A one-way between-groups analysis of variance with planned comparisons was conducted to explore the impact of fearful attachment style on levels of friendship satisfaction appraised reflectively, as measured by the 'Satisfaction of Friend' scale. Subjects were divided into four groups according to their attachment style (Secure; Preoccupied; Dismissing; Fearful). There was no statistically significant difference at the p <. 05 level in Satisfaction_F scores for the four attachment groups:  The relationship between avoidant (fearful) style of attachment (as measured by the AvoidantAttch) and reflective appraisal of satisfaction in the friendship relationship (as measured by the Satisfaction_F scale) was investigated using Pearson product-moment correlation coefficient. Preliminary analyses were performed to ensure no violation of the assumptions of normality, linearity and homoscedasticity. There was a small, negative correlation between the two variables, r = -.16, n = 109, p < 0.09, with high levels of avoidant attribute associated with low levels of friendship satisfaction.
A one-way between-groups analysis of variance with planned comparisons was also conducted to explore the impact of secure attachment on levels of friendship intimacy appraised reflectively, as measured by the 'Intimacy of Friend' scale. Subjects were divided into four groups according to their attachment style (Secure; Preoccupied; Dismissing; Fearful). There was no statistically significant difference at the p <. 05 level in Intimacy_F scores for the four attachment groups:

Discussion
The results of this study assist to clarify and extend previous findings with regard to individuals' early attachment styles with primary care-givers and qualities of close relationship formed with others later in life. In general, findings support predictions based on previous research on attachment styles, communication skills and friendship qualities.
As predicted by hypothesis 1, this study found that communication skills affect friendship quality significantly. There was a moderately strong, positive correlation between the communication skills and friendship quality. This indicates that individuals with good communication skills tend to have better friendship quality. This could due to intimacy being formed when friends reveal intimate information regarding themselves, causing them to trust one another, leading to a better friendship quality (Bauminger, Finzi-Dottan, Chason & Har-Even, 2008 illustrating that good communication creates intimacy in friendship and leads to satisfaction in friendship and results in better friendship quality (Berndt, 1996).
With regard to hypothesis 2, this study did not find a significant difference in the effect of attachment styles on level of communication. Previous research, however, found that attachment style did significantly affect level of self-disclosure. Pistole (1993) observed from a counselling perspective opined that secure people divulged more than insecure people. Furthermore, Collins & Read (1990) stated that people who were contended with closeness conveyed more self-disclosure. However, this result could have been obtained due to other factors like length of friendship between the participants, affecting the level of communication between them, and possibly cushioned off the effect of their attachment styles. Thus, even people with fearful attachment style scored high in their communication skills scale despite previous research findings and Bartholomew's and Horowitz's (1990) predictions on them as being low in self-disclosure and having the tendency to maintain distant from partners. Findings in this study showed that this contradictory outcome only appeared to those with great length of friendship. Hence future study should take the interactive effect of friendship length into consideration as well.
Similarly, in testing of hypothesis 3, there was no statistical significant difference between the attachment style of participants and the friendship quality perceived by friends. Nevertheless, the actual mean scores of the analysis showed that friends of preoccupied individuals perceived their friendship as having low quality when compared to the secure, dismissive and fearful groups. This outcome coincides with findings of earlier research which predicted that preoccupied people are highly anxious individuals who crave closeness, usually report less satisfaction in their friendship relationships (Welch & Houser, 2010).
Nevertheless, results of study with regard to hypothesis 4 showed that there was a significant effect of the friends' attachment styles on the perceived friendship quality. This indicates that the attachment style of friends affected how they perceived the quality of their friendship and hence the final appraisal of their friendship quality. In this study, friendship quality is assessed from the perception of friend. Securely attached individuals appraised their friendship with others as of high quality while fearfully attached individuals appraised the quality of their friendship with others as of low quality.
Conversely, in testing of the hypotheses 5 and 6, the relationship between satisfaction and attachment styles of individuals as well as that of intimacy and that of individuals' attachment styles were analyzed using one-way between group analysis of variance with planned comparisons. Results showed that the attachment style of participants has no significant effect on the satisfaction felt by their friends. This finding contradicts the prediction of previous research and could probably due to the small and unequal sample size of each attachment style in this study. However, a comparison of the actual mean scores of the four attachment groups indicates that friends of fearfully attached individuals experienced least satisfaction in their friendship relationship. A correlation analysis of the avoidant attributes and friendship satisfaction variables showed that there is a mild negative correlation between the two. This finding indicates that individuals with higher avoidant attributes tend to experience lower satisfaction in friendship relationships. This is consistent with Bartholomew and Horowitz (1991) postulation which opined that secure individuals display low levels of anxiety and avoidance of their view of self but high levels of hope, self-disclosure and relationship satisfaction. Conversely, fearful people do simply showed high levels of anxiety and avoidance, nevertheless, they respond with lower levels of hope, self-disclosure and relationship satisfaction (Welch & Houser, 2010).
Likewise, analysis on the attachment style of participants on the intimacy felt by their friends did not reach statistical significance. This finding is not as predicted. Again, this can be attributed to small and unequal sample size for each attachment style category. Nevertheless, friends of preoccupied group reflected perception of least intimacy in their friendship relationship as showed in figure 3. This findings was predicted by Welch and Houser (2010) who found that preoccupied individuals perceive lower intimacy in their relationships due to self-esteem issues resulted from negative views of self.
Overall, results of the study suggest that communication skills of individuals play a role in friendship quality but not their attachment styles. Nevertheless, the attachment style of friends instead of the participants was found to influence the appraisal of friendship quality. Hence, it seems that the perception of the quality of a friendship relationship depends significantly on the attachment style of the friends who appraised it rather than that of the participants.

Limitations and Future Directions
It should be noted that if this study were to be repeated, researcher should first ensure that the number of participants in each attachment style category is equal. This will enable a better and fair comparison among the groups. In this study, the sample size was fixed first. Attachment styles of individuals were only streamed after filling up the questionnaires. This approach gave rise to the limitation of small and unequal sample size in each attachment style category, leading to possible unpredicted outcomes for some of the hypotheses.
Secondly, there is the consideration on the use of self-report evaluation of questionnaires alone. There are concerns over self-report bias and the passive nature of these instruments. It was argued that self-report measures probe conscious attitudes towards current relationships based on yesterday's relationships and memories of experiences. These instruments do not detect certain attachment phenomena which require activation to manifest (Ravitz et al., 2010). Thus, future research should probably consider adding of observational study such as interviews to reduce such limitation.
Thirdly, the length of friendship posed a limitation to this study as it interferes with the effect of attachment style in area with regard to level of self-disclosing which determines level of intimacy in friendship relationships. Thus, future research should consider taking dyads with same length of friendship across the four categories of attachment styles, preferably with minimum one year of friendship instead of only three months as in this study. This is to eliminate the possible effect of short term friendship in friendship quality study.
Lastly, generalizability of results is limited by the nature of the sample. Results of study by Samter and Burleson (2005) showed that people from different cultures do emphasize different aspects of communication, that is, people communicate with different patterns in accordance to their cultures. In this study, there is a mixture of ethnic groups in the sample. This may introduce confounding variable to the study when different ethnic participants communicate with different patterns even if they share the same style of attachment. Future research should ensure ethnic homogeneity to single out confounding effect of cultures on the study of communication skills effect on friendship quality.

Conclusion
In conclusion, our findings suggest that communication skills play a vital role in mediating the effect of attachment styles on friendship quality as people with different attachment styles communicate differently in accordance to their internal working models. Further, it was found that the attachment styles of participants have no effect on the friendship quality appraised reflectively by friends; on the contrary, it is the attachment styles of friends which showed to have an effect on the friendship quality. This finding reflects the possibility that the quality of a relationship is very much depending on the attachment style of the appraiser rather than that of the partners. However, such finding needs further corroboration from future research.