Professional Documents
Culture Documents
Atkins Et Al 2005
Atkins Et Al 2005
1, 144 150
BRIEF REPORTS
Kathleen A. Eldridge
Pepperdine University
Donald H. Baucom
University of North Carolina
Andrew Christensen
University of California, Los Angeles
Infidelity is a common issue with which distressed couples and their therapists grapple. However, there are no data on the efficacy of commonly used therapies to treat couples in which there has been an affair. In the present exploratory study, the authors examined the therapy outcomes of a sample of infidelity couples (n 19) who had participated in a randomized clinical trial of marital therapy (N 134). Results show that infidelity couples began treatment more distressed than noninfidelity couples; however, evidence suggests that couples who had an affair and who revealed this affair prior to or during therapy showed greater improvement in satisfaction than noninfidelity couples. Implications for therapy with infidelity couples are discussed.
The world breaks everyone, and afterward, some are strong at the broken places. Ernest Hemingway, A Farewell to Arms
For many couples, the discovery of a partners infidelity1 is experienced as a betrayal that shatters one of the most fundamental assumptions of a relationship: its exclusivity. Yet, therapists have little empirical guidance in treating couples who have had an
David C. Atkins, Travis Research Institute, Fuller Graduate School of Psychology; Kathleen A. Eldridge, Graduate School of Education and Psychology, Pepperdine University; Donald H. Baucom, Department of Psychology, University of North Carolina; Andrew Christensen, Department of Psychology, University of California, Los Angeles. Some of the results in this article were presented in a symposium at the 36th Annual Conference of the Association for the Advancement of Behavior Therapy, Reno, Nevada, November 2002. This research was supported by National Institute of Mental Health Grants MH56223 (awarded to Andrew Christensen at the University of California, Los Angeles) and MH56165 (awarded to Neil S. Jacobson at the University of Washington for a two-site clinical trial of couple therapy; after Jacobsons death in 1999, William George served as primary investigator at the University of Washington). The article is based in part on David C. Atkins dissertation, which was submitted in partial fulfillment of the requirements for a doctoral degree in clinical psychology at the University of Washington. We thank John Gottman, Peter Fehrenbach, Bob Kohlenberg, Sona Dimidjian, and Sara Berns for their insightful comments and feedback. Correspondence concerning this article should be addressed to David C. Atkins, Travis Research Institute, Fuller Graduate School of Psychology, 180 North Oakland Avenue, Pasadena, CA 91101. E-mail: datkins@ fuller.edu
affair. At the present time, there are no data on the efficacy of widely used couple therapies to handle instances of infidelity. In the present study, we provide preliminary data on treatment outcomes of 19 couples who had an affair and who had participated in a randomized clinical trial of marital therapy (Christensen et al., 2004). In examining the efficacy of couple therapy, meta-analyses (e.g., Dunn & Schwebel, 1995) and qualitative reviews (e.g., Baucom, Shoham, Mueser, Daiuto, & Stickle, 1998) have shown that couple therapy can be a powerful treatment for relationship discord; however, couple therapy is not a panacea for all relationship problems. In looking at clinical significance data (reviewed in Baucom et al., 1998), most therapists have found that approximately two thirds of couples improve during therapy and that between one third and one half of couples are considered recovered at the end of therapy. Follow-up data, when it exists, demonstrate that a proportion of couples (as many as one third) do not maintain the improvement that they accomplished in therapy. Thus, there is a sizable proportion of couples who receive no benefit from therapy and others who are not able to maintain their therapeutic gains over time. When infidelity has been mentioned in couple therapy studies, it has typically been included in a list of presenting problems for therapyjust another issue on the list of problems that couples bring to therapy. However, infidelity may alter the likelihood that couples respond positively to treatment.
Many different terms have been used in the research literature to refer to infidelity. For consistency, infidelity and affair are used to refer to a sexual and/or secret emotional relationship with someone other than ones spouse.
1
144
BRIEF REPORTS
145
For instance, Bennun (1985) found that jealousy was a particularly intractable problem for traditional behavioral couple therapy2 (TBCT; Jacobson & Margolin, 1979), and Whisman, Dixon, and Johnson (1997) found that couple therapists report infidelity as one of the most difficult problems to treat. Moreover, Cano and OLeary (2000) have recently shown that discovery of infidelity is associated with increased levels of clinical depression in the noninvolved spouse. Research focused on infidelity and other difficult problems in couple therapy may provide critical information to improve the overall impact of couple therapy. At the present time, only a single pilot study to our knowledge has been conducted that directly assessed the efficacy of couple therapy with infidelity. Gordon, Baucom, and Snyder (2004) have developed a multitheoretical treatment that views affairs as a relationship trauma. In their pilot study, the authors used a replicated case study design with 13 infidelity couples, 9 of whom completed therapy. Therapy effectively helped noninvolved partners cope with and come to some understanding of the affair, with subsequent improvement in marital satisfaction; however, involved partners perceptions of the relationship and individual distress did not change during treatment. Gordon et al. (2004) noted that the treatment focused heavily on the affair, and both partners may benefit from working on wider relationship issues as well as on the affair. With the exception of the Gordon et al. (2004) study, to our knowledge there are no empirical studies of infidelity to guide couple therapists treatment of infidelity couples. Using data from a randomized clinical trial of marital therapy (Christensen et al., 2004), we examined in the present, exploratory study the initial level of distress and course of treatment in couple therapy for infidelity couples compared with distressed couples who had no affair. On the basis of clinical experience, we hypothesized that affair couples would be more distressed initially and improve at a slower rate during treatment when compared with nonaffair couples.
session, separately for 1 session each, and then together for a feedback session in which the therapist presented an assessment of the couples problems and an overview of treatment. All remaining sessions were in a conjoint format. Therapy with infidelity couples used the techniques of a given therapy to address the affair; thus, couples in TBCT were taught communication and problem-solving skills with which to discuss the affair and handle issues related to it, whereas couples in IBCT focused more on the emotional impact of the affair and an understanding of its origin and meaning.3 Couples were assessed at four time points during therapy with self-report questionnaires: (a) prior to treatment, (b) 13 weeks after the pretherapy assessment, (c) 26 weeks after the pretherapy assessment, and (d) at the couples final therapy session (M 35.0 weeks, SD 9.2). The time between assessments was roughly the same for couples across the first three time points, but the last time point varied among couples.
Measures
DAS. The DAS (Spanier, 1976) is a commonly administered, 32-item self-report measure of relationship satisfaction with excellent reliability ( .96) and discriminant validity. In the present study, we used the total DAS score as the primary outcome measure of relationship satisfaction. Infidelity questionnaire. To gather data about the specifics of the affairs, we designed a questionnaire for each therapist to complete for every infidelity couple that he or she counseled. The questionnaire assessed the following: (a) which partner had the affair, (b) length of the affair, (c) when the affair began, (d) when the affair was revealed, (e) number of affairs, (f) degree of physical involvement, (g) degree of emotional involvement, (h) percentage of time in therapy spent on the affair, and (i) therapists perceptions of the couple. Some questions were based on items from Glass and Wrights (1992) study and Buss and Shackelfords (1997) study.4 The primary means of identifying affairs in the marital therapy study were through therapists reports. Therapists were asked to report on any of the couples whom they treated in which there was a sexual and/or emotional affair (Glass & Wright, 1992). A relationship was deemed an emotional affair if it involved secrecy, contained romantic or sexual feelings, and interfered with the primary relationship. There was only a single affair reported in the present study that was purely emotional, without any sexual component.
Method Participants
All of the data for the present research came from an ongoing study of marital therapy. Participants in this study were 134 heterosexual, married couples who sought therapy for marital problems. To be eligible for the study, couples had to meet criteria for relationship distress on the Dyadic Adjustment Scale (DAS; Spanier, 1976) and the Global Distress Scale of the Marital Satisfaction InventoryRevised (Snyder, 1997). Exclusion criteria included the following: (a) evidence of a psychotic disorder in either partner, (b) alcohol or drug abuse or dependence, and (c) greater than moderate levels of intimate partner violence. Nineteen couples (14.2% of the total sample) had at least 1 partner with a reported infidelity. For a complete description of the study methods, see Christensen et al.s (2004) study; only those measures and procedures directly relevant to the present study are addressed here.
Data Analysis
Hierarchical linear modeling (HLM; also called multilevel modeling or mixed-effects modeling) effectively and accurately models data with cor-
Procedure
Couples were randomized to either TBCT (Jacobson & Margolin, 1979) or integrative behavioral couple therapy (IBCT; Jacobson & Christensen, 1996) and received up to 26 sessions at no cost. Both treatments began with an assessment phase in which partners were seen together for the 1st
2 There has been an evolution in the names of couple therapies and TBCT in particular. Originally, this therapy was referred to as behavioral marital therapy and later as behavioral couple therapy when the field as a whole recognized that there are other committed relationships besides marriage. More recently, Christensen et al. (2004) have referred to it as TBCT to clearly delineate it from their present work. 3 We do not explore potential predictors of treatment response among infidelity, including treatment condition, in the present study because of the small sample size. Whereas it might be logical to explore factors such as treatment condition and when the affair was revealed as predictors of treatment response, even crossing these two factors leads to cell sizes with only a single infidelity couple. Thus, we believe it is more conservative and appropriate to not explore predictors of treatment response in the present, small sample. 4 A copy of the questionnaire can be obtained by contacting David C. Atkins.
146
BRIEF REPORTS
related groups such as spouses and repeated measures (Atkins, in press; Raudenbush & Bryk, 2002). In addition, HLM is often robust to missing data (see Atkinss [in press] study for a discussion). For the present study, only 7% of the total data were missing. In the present analyses, we used a three-level model to represent couples change over time in relationship satisfaction. Following the notation of Raudenbush and Bryk (2002), the model is displayed in Equation 1.5 Level 1: Ytij 0 ij 1 ij time tij 2 ij time 2 tij tij Level 2: 0 ij 00 j r 0 ij
1 ij 10 j 2 ij 20 j
Level 3: 00 j 000 u 00 j
10 j 100 u 10 j 20 j 200 u 20 j
(1)
in which t indexes time, i indexes individuals, and j indexes couples. tij is the Level 1 residual error term that describes the scatter of each individuals data around his or her estimated regression line; r0ij is a random intercept that allows individuals within the same couple to have separate intercept values; u00j, u10j, and u20j are the random effects at the couple level that allow different couples within the study to have distinct intercept, slope, and quadratic values. An infidelity indicator variable was added (0 no infidelity, 1 infidelity) as a predictor at Level 3 to detect whether couples who had an affair reliably deviate from noninfidelity couples in one or more components of their relationship satisfaction trajectory.6 Analyses were conducted with S-plus 2000 Professional Release 2 (Mathsoft, 1999) and made extensive use of the nlme library (Version 3.3.1) of functions for mixed-effects modeling (Pinheiro & Bates, 2000).
Note. For all affair couples, the median length of affair 6 months (range 124 months); the mode for degree of physical involvement sexual intercourse; the mode for degree of emotional involvement moderate emotional involvement; the mean percentage of time in therapy spent on affair 37.9 (SD 31.2). TBCT traditional behavioral couple therapy; IBCT integrative behavioral couple therapy.
ples. Approximately two thirds of couples with affairs received TBCT, and slightly more than 50% of involved spouses were men. There was considerable variation in the duration of the affair relationship (Mdn 6 months). The great majority of affairs began prior to therapy. We find it interesting that only about one third of affairs were revealed prior to beginning therapy, and one fourth were never revealed during therapy. Most involved spouses had a single affair involving sexual intercourse and moderate emotional
5 Most published couple research with HLM has used the two-level multivariate model presented by Raudenbush, Brennan, and Barnett (1995). The three-level model used in the present study is more parsimonious (8 random effects parameters vs. 22 parameters in the two-level model). One significant restriction of the present model is that spouses are assumed to have a common slope and quadratic. This assumption is reasonable in a therapy study as it is very uncommon to have partners relationship satisfaction moving in opposite directions. See Atkinss (in press) study for a discussion of two-level versus three-level models for couple data. 6 An issue that may be a factor in the analysis is that 5 of the infidelity couples never discussed the infidelity during therapy. The outcomes of these secret affairs do not address the primary question because in these instances therapy never addressed the infidelity. Models were run with and without these couples (i.e., with 19 infidelity couples and with 14 infidelity couples); however, the results in the text focus on the 14 infidelity couples whose affair was addressed in therapy. Of the 5 couples whose affair was never brought up during therapy, 2 individuals mentioned it to the therapist during the individual session. The affair was finished by that time, and they indicated that they did not wish their partners to know about it. The remaining 3 couples never revealed the affair to spouse or therapist, and it was not until therapy had ended that the research team or therapist learned of the affair.
Table 1 Demographic Information for Couples in the Marital Therapy Study (N 134)
Husbands Variable Age Education Monthly income Years married No. of children Ethnicity, n (%) Caucasian African American Asian AmericanPacific Islander Latino Other M 43.49 17.03 4,642 SD 8.74 3.17 3,787 M 41.62 16.97 3,696 8.50 1.00 Wives SD 8.59 3.23 4,085 7.61 1.03
Note. Years married and number of children are identical for husbands and wives. The value for the couple is reported in the wives column.
BRIEF REPORTS
147
involvement. When the affair was addressed in therapy, there was significant variability in the amount of time in therapy spent on the affair (M 37.9%, SD 31.2%).
Basic Model
Figure 1 presents mean DAS scores during therapy and 90% bootstrapped confidence intervals for distressed, noninfidelity couples (n 115), infidelity couples (n 14), and secret affair couples (n 5). Descriptively, infidelity couples began treatment more distressed than noninfidelity couples, and secret affair couples were more distressed yet. Infidelity couples had a positive trajectory of relationship satisfaction during therapy as opposed to secret affair couples who showed an increase early in therapy that deteriorated in the later portions of therapy. Table 3 presents the basic HLM model and a model including the infidelity indicator variables, excluding couples in which the affair was kept secret. The basic model shows that, on average, couples in the study began treatment with a DAS score of 85 and improved during the course of therapy, gaining 0.36 DAS points per week of therapy. There is some indication of deceleration, but the quadratic component failed to reach significance. A deviance test demonstrated that adding the infidelity indicator variables significantly improved the fit of the model, 2(3, N 129) 10.0, p .018. The coefficients of the infidelity dummy variables represent the difference between infidelity couples and distressed, noninfidelity couples in each parameter of the growth curves. Similar to the plot of means, infidelity couples began therapy significantly more distressed than noninfidelity couples, with an estimated pretreatment DAS score of approximately 78 ( p .028), which is a large
effect (d 1.12). There was a nonsignificant difference between infidelity and noninfidelity couples in their rate of change, and some indication ( p .090, d 0.72) that infidelity couples change in therapy was accelerating over time compared with noninfidelity couples.7 The estimated regression lines along with mean DAS scores at the four time points are presented in Figure 2.8
Clinical Significance
Table 4 presents clinical significance results with Jacobson and Truaxs (1991) method for study couples broken down by whether there was an affair. The amount of change made during therapy was quite similar across infidelity and noninfidelity couples ( p .86 with Fishers exact test). If the affair couples who did not address the affair in treatment are included, then the percentage of recovered infidelity couples drops to 33%, reflecting the fact that almost all of these secret affair couples were treatment failures. Finally, contrary to the findings of Gordon et al. (2004), involved spouses were more distressed initially on the DAS (M 73.26, SD 15.60) than noninvolved spouses (M 78.42, SD 17.30); however, this difference did not reach significance, t(19) 1.38, p .18, likely due to the small sample of infidelity couples. Both partners in infidelity couples achieved similar gains in therapy.
Discussion
It is not uncommon to hear the opinion that an affair is the end of a relationship. Our data speak unequivocally to this issue: infidelity is not necessarily the death knell of a relationship. At the same time, a case could be made that infidelity is not just another problem that couples bring into therapy, as infidelity couples are
7 Models that included the secret (n 5) affair couples revealed that these couples were significantly more distressed than either noninfidelity couples or infidelity couples whose affair was addressed in therapy, t(131) 2.60, p .01. There were no significant slope or quadratic differences, most likely attributable to the sample size. 8 One possible confound of the present analysis is that infidelity couples were more highly distressed at the start of therapy; thus, differences may reflect this initial distress and not infidelity per se. To address this possibility, we conducted a matched case-control analysis. Each infidelity couple was matched to one or more noninfidelity couples on the basis of average pretreatment DAS couple scores and Marital Satisfaction InventoryRevised pretreatment Sexual Dissatisfaction subscale couple scores (Snyder, 1997). Results based on an HLM analysis that incorporated the additional nesting because of matching revealed virtually identical results with the primary analysis. Thus, differences based on infidelity do not appear to be driven by initial distress. Details of the case-control analysis are available from David C. Atkins. In addition, an anonymous reviewer noted that there is not a control group in the present analyses, which makes it difficult to rule out spontaneous remission as a possible confound. Baucom, Hahlweg, and Kuschel (2003) recently conducted a meta-analysis of wait-list control groups in studies of TBCT that showed that couples receiving no treatment did not tend to improve; on the contrary, they worsened slightly. This finding argues against spontaneous remission as an explanation of the present findings.
Figure 1. Mean Dyadic Adjustment Scale scores with 90% bootstrapped confidence intervals during therapy by infidelity status.
148
BRIEF REPORTS
Table 3 Hierarchical Linear Modeling Results for Basic Model and Model With Infidelity
Basic model Variable Intercept Infidelity Slope Infidelity Quadratic Infidelity AIC B 84.67 0.36 0.002 SE B 1.02 0.08 0.002 7,431.6 t 82.8 4.6 1.1 B 85.90 7.10 0.39 0.26 0.003 0.009 SE B 1.08 3.27 0.09 0.26 0.002 0.006 Infidelity model t 79.6 2.2 4.5 1.0 1.4 1.7 7,427.6 p .0001 .028 .0001 .318 .162 .090 d 1.12 0.37 0.72
Note. t t statistic based on regression coefficient divided by its standard error; p exact p value (except where noted) based on t statistic with 127 degrees of freedom for intercept and 700 degrees of freedom otherwise; d effect size calculated as fixed-effect regression coefficient divided by the square root of the corresponding random effect; AIC Akaike Information Criterion.
notably and reliably more distressed than their noninfidelity peers at pretreatment. Especially given their initial distress, the results of the present research are quite optimistic with respect to the efficacy of couple therapy with couples who had an affair. Infidelity couples start treatment more distressed than their noninfidelity
peers; however, if anything, they improve in therapy at a greater rate, particularly at the end of therapy. Thus, at the end of treatment, the average outcome of couples who had an affair is indistinguishable from the outcome of distressed couples without affairs.
Figure 2. Predicted regression lines from hierarchical linear modeling analysis with infidelity predictors along with mean Dyadic Adjustment Scale (DAS) scores. Squares represent the DAS mean at four assessment points for couples without infidelity; triangles represent the DAS mean at four assessment points for couples with infidelity.
BRIEF REPORTS
149
Table 4 Clinical Significance Classifications for Infidelity and Noninfidelity Couples (N 125)
Noninfidelity couples Classification Deteriorated Unchanged Improved Recovered n 14 25 17 56 % 12 22 15 50 n 2 3 3 6 Infidelity couples % 15 21 21 43
Note. Four couples, including one infidelity couple, dropped out of the study prior to the end of treatment, and it was impossible to compute clinical significance statistics for them. Only the 14 infidelity couples in which the affair was addressed in treatment are included.
However, there is a note of caution in our findings as well: Affairs that remain hidden appear devastating to the relationship. This finding is mostly observational rather than statistical as the small sample size (n 5) of secret affair couples prevents any meaningful statistical analyses. However, with a mean DAS score of 76 at the end of treatment, there appears to be considerable consistency and little optimism among this small cohort of couples. In some instances, the affairs were ongoing while the couple was in therapy. For these couples, there is strong reason to believe the involved partner was not fully committed to the therapeutic process. For couples in which the affair was finished but the involved partner refused to address it in therapy, it may be that the secrecy is a marker for a general lack of trust and openness in the relationship. Undisclosed affairs would be a profitablealbeit quite challengingarea for future research. The present results also differed somewhat from those in Gordon et al.s (2004) study. The two primary findings from Gordon et al.s study were that the noninvolved spouse was more distressed initially and also made greater gains in therapy relative to the involved spouse. In the present study, the involved spouse was found to be more distressed than the noninvolved spouse, and both partners made similar gains in therapy. Gordon et al.s study focused almost entirely on the affair, which may explain the differential effects between involved and noninvolved spouses. In the present study, we used two broad-based couple therapies, and our work with infidelity couples involved nonaffair issues. Comparing the findings from the two studies, it appears that focusing on the relationship as a whole may be particularly helpful for the involved spouse. In addition, Gordon et al.s study specifically recruited couples with affairs, which may also help to account for differences between the studies. Whereas the present study is quite encouraging with respect to infidelity and couple therapy, these results are preliminary because of several limitations of the present research. First and foremost, the total sample of infidelity couples (n 19) is quite small, and further research is needed to replicate the present findings. Second, the rate of affairs in the present sample (14.2%) is lower than we expected in a distressed sample of couples, though it is similar to the rate of infidelity in a nationally representative sample of married couples (Atkins, Baucom, & Jacobson, 2001). Because there are no data on the rate of infidelity among distressed couples,
it is difficult to assess the rate of affairs in the present sample. In addition, it is possible (even likely) that there are some affairs in the past (or present) that went undetected in our sample. Thus, we are most confident about the known infidelity couples and their response to treatment, whereas the noninfidelity couples may include several past (or ongoing) affairs that were never made known. Finally, the overall sample of participants are not completely representative of distressed couples as a whole because of the exclusion criteria of the study and the demographic characteristics of those who chose to participate in the research. In conclusion, it is worth highlighting the optimistic findings of the present research with respect to infidelity and couple therapy. This study is modest in terms of the number of affair couples, and speculation should be tempered accordingly. Nonetheless, the clear indication is that two of the existing, broad-based couple therapies can be effective with couples who are struggling with infidelity, particularly if the infidelity is addressed during treatment. This is an important message for couples who may be seeking couple therapy for infidelity and for the therapists who are providing the therapy. In addition, it is an encouraging indication of the efficacy of couple therapy.
References
Atkins, D. C. (in press). Using multilevel models to analyze marital and family treatment data: Basic and advanced issues. Journal of Family Psychology. Atkins, D. C., Baucom, D. H., & Jacobson, N. S. (2001). Understanding infidelity: Correlates in a national random sample. Journal of Family Psychology, 15, 735749. Baucom, D. H., Hahlweg, K., & Kuschel, A. (2003). Are waiting list control groups needed in future marital therapy outcome research? Behavior Therapy, 34, 179 188. Baucom, D. H., Shoham, V., Mueser, K. T., Daiuto, A. D., & Stickle, T. R. (1998). Empirically supported couple and family interventions for marital distress and adult mental health problems. Journal of Consulting and Clinical Psychology, 66, 53 88. Bennun, I. (1985). Prediction and responsiveness in behavioural marital therapy. Behavioural Psychotherapy, 13, 186 201. Buss, D. M., & Shackelford, T. K. (1997). Susceptibility to infidelity in the first year of marriage. Journal of Research in Personality, 31, 193221. Cano, A., & OLeary, K. D. (2000). Infidelity and separations precipitate major depressive episodes and symptoms of nonspecific depression and anxiety. Journal of Consulting and Clinical Psychology, 68, 774 781. Christensen, A., Atkins, D. C., Berns, S. B., Wheeler, J., Baucom, D. H., & Simpson, L. (2004). Integrative versus traditional behavioral couple therapy for moderately and severely distressed married couples. Journal of Consulting and Clinical Psychology, 72, 176 191. Dunn, R. L., & Schwebel, A. I. (1995). Meta-analytic review of marital therapy outcome research. Journal of Family Psychology, 9, 58 68. Glass, S. P., & Wright, T. L. (1992). Justifications for extramarital involvement: The association between attitudes, behavior, and gender. Journal of Sex Research, 29, 361387. Gordon, K. C., Baucom, D. H., & Snyder, D. K. (2004). An integrative intervention for promoting recovery from extramarital affairs. Journal of Marital and Family Therapy, 30, 112. Jacobson, N. S., & Christensen, A. (1996). Integrative couple therapy: Promoting acceptance and change. New York: Norton. Jacobson, N. S., & Margolin, G. (1979). Marital therapy: Strategies based on social-learning and behavior exchange principles. New York: Brunner/Mazel.
150
BRIEF REPORTS Snyder, D. K. (1997). Marital Satisfaction InventoryRevised. Los Angeles: Western Psychological Services. Spanier, G. B. (1976). Measuring dyadic adjustment: New scales for assessing the quality of marriage and similar dyads. Journal of Marriage and the Family, 38, 1528. Whisman, M. A., Dixon, A. E., & Johnson, B. (1997). Therapists perspectives of couple problems and treatment issues in couple therapy. Journal of Family Psychology, 11, 361366.
Jacobson, N. S., & Truax, P. (1991). Clinical significance: A statistical approach to defining meaningful change in psychotherapy research. Journal of Consulting and Clinical Psychology, 59, 1219. Mathsoft. (1999). S-Plus 2000 [Computer software]. Seattle, WA: Data Analysis Products Division, Mathsoft. Pinheiro, J. C., & Bates, D. M. (2000). Mixed effects models in S and S-Plus. New York: Springer-Verlag. Raudenbush, S. W., Brennan, R. T., & Barnett, R. C. (1995). A multivariate hierarchical model for studying psychological change within married couples. Journal of Family Psychology, 9, 161174. Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods (2nd ed.). Thousand Oaks, CA: Sage.
Received November 1, 2003 Revision received April 8, 2004 Accepted May 15, 2004