Abstract
This study aimed to explore the associations among attachment system type, sexual satisfaction, and marital satisfaction in adult couples in stable relationships. Participants were 294 couples between the ages of 20 and 70 years who answered self-administered questionnaires. Hierarchical linear modeling revealed that the anxiety and avoidance, sexual satisfaction, and marital satisfaction dimensions were closely related. Specifically, the avoidance dimension, but not the anxiety dimension, corresponded to lower levels of sexual and marital satisfaction. Moreover, for the sexual satisfaction variable, an interaction effect was observed between the gender of the actor and avoidance of the partner, which was observed only in men. In the marital satisfaction dimension, effects were apparent only at the individual level; a positive relation was found between the number of years spent living together and greater contentment with the relationship. These results confirm the hypothetical association between attachment and sexual and marital satisfaction and demonstrate the relevance of methodologies when the unit of analysis is the couple.
APPENDIX
The basic multilevel model was defined by considering the results of the individual, the unit of analysis of Level 1, and the couple, the unit of analysis of Level 2, and allowing random intercepts in the models. Separate models were estimated for each variable. For each variable outcome, a null model or an unconditional and conditional model with covariates was estimated.
The equation of the null or unconditional Level 1 model was specified as follows: (1)
Here, Yij represents the variable outcome of individual i in couple j, βoj represents the average of the pair j in the variable outcome, and r represents the error term that is typically assumed in distributions with a mean of 0 and a variance of σ20.
The equation of the null or unconditional Level 2 model was specified as follows: (2) in which Yoo represents the average variable outcome and uo represents the deviation of the couple's average from the total average.
The conditional model incorporating the predictors of the study was broadly specified for Level 1 as follows: (3)
Here, the result of the variable outcome Yij additionally depends on β1, which represents the coefficient regression associated with covariates Z, including a list of other covariates in the models.
The Level 2 equations are presented as follows: (4) (5) (6)
Here, random intercepts are allowed for βoj, and only fixed effects are allowed for each of the coefficients associated with the co-variables. Covariates are not included in Level 2.
Notes
This variable presented statistically significant differences; lower scores were observed in women (M = 29.82, SD = 5.19) than men (M = 30.61, SD = 4.32), t(586) = 3.43, p <.01, although the size of the effect was small (d =.17).