Toward High‐Quality Divorce Agreements: The Influence of Facilitative Professionals

During divorce, couples typically must negotiate spousal and child support, divide jointly held property and other assets, and reorganize their living and child‐custody arrangements (Emery 1995). Conflicts about such issues often present major challenges to these couples, their family members, and the professionals assisting them (Beck and Sales 2001). The two most common means of completing divorce settlements are via litigation and mediation.

In brief, mediation is the process in which an impartial third party (the mediator or mediators) facilitates an agreement‐making process during which the decision‐making power remains with the disputants themselves (Kressel and Pruitt 1989). Mediation's foundational principle is that, by fostering parties' self‐determination, detrimental legal battles can be avoided, and resolutions can emerge that benefit the interests of both parties (Welsh 2004). Although mediation advocates proclaim its general effectiveness, much remains unclear concerning which process‐factors contribute to high‐quality outcomes following divorce mediation.

During the last three decades, researchers have focused on determining divorce mediation's advantages over litigation. Some studies reported positive outcomes from mediation (Kelly and Gigy 1989; Bailey and Robbins 2005), whereas others reported that parties faced emotional difficulties and substantive losses following mediation (Pearson and Thoennes 1988; Emery, Matthews, and Wyer 1991). These mixed results have suggested the need for more nuanced examinations of which elements of divorce mediation make the process effective or ineffective.

Several studies have examined the influence of gender on the outcome of dispute resolution. For example, in one early study, Robert Emery and Melissa Wyer (1987) found that mothers whose divorces were resolved by litigation were more satisfied with their final settlements than were mothers whose cases were mediated. Another study found that men whose divorces were litigated gained a greater understanding of their ex‐spouses' positions than did similar men who went through the mediation process, although the same study found that women whose divorces were mediated felt more empowered than did those whose divorces were resolved through litigation (Kelly and Gigy 1989). In another study, fathers who did not express concerns about participating in divorce mediation were also more likely to reach full agreements than fathers who did express concerns, but the level of concern expressed by mothers prior to the process did not seem to have any impact on whether the parties achieved settlement (Ballard et al. 2011).

Divorce mediation researchers have also studied differences between divorces initiated by both parties or unilaterally (Emery 1995; Bickerdike and Littlefield 2000). Specifically, non‐initiators have been found to be more likely to be overwhelmed by the divorce decision and to experience intensive feelings of unresolved attachment toward the ex‐partner. In contrast, initiators have been found to have had more time to adjust to the divorce and to thus experience less difficulty adjusting to the changed boundaries of the relationship (Emery 1995; Kruk 1998). Overall, initiators and non‐initiators typically experience the divorce negotiation process differently.

Variations in divorce dispute resolution outcomes can also reflect the level of marital conflict (Johnston and Campbell 1998; Ballard et al. 2011). Indeed, some researchers have argued that the competitive and adversarial nature of litigation encourages antagonistic divorces and aggravates existing conflict between spouses (Somary and Emery 1991), and high levels of conflict between spouses are predictive for failing to achieve satisfactory divorce agreements (Wall, Stark, and Standifer 2001; Kelly 2004; Emery, Sbarra, and Grover 2005). And, indeed, a founding principle of family mediation is that complicated and emotionally charged conflicts can be more easily dealt with outside of the traditional court proceedings (Gerber 1990). Accordingly, Katherine Kitzmann and Robert Emery (1994) found that disputants' perceptions of procedural justice (i.e., the fairness of the process itself regardless of outcomes) can predict the overall satisfaction of highly conflicted couples, whereas for divorcing couples with relatively low levels of conflict, their perceptions of distributive justice (i.e., the fairness of the outcome) were more predictive of their satisfaction with the mediation experience. Other researchers have found that the most extremely violent couples were more likely to be screened out of mediation by mediators (Tishler et al. 2004). And researchers have also reported that high conflict between ex‐partners can predict failure to reach mediated divorce agreements (see Kelly 1996, 2004 for a review).

Although the first studies comparing divorce mediation and litigation were conducted more than three decades ago, many gaps in our knowledge remain (Beck and Sales 2001; Shaw 2010). For instance, to date, only one quantitative meta‐analysis comparison of family mediation and litigation has been undertaken, and it comprised merely five studies. This study corroborated that there were moderately better results when family mediation was used instead of divorce litigation in the following areas: children's psychological needs, spousal relationships, and disputants' emotional satisfaction, and their satisfaction with both the process and the outcome (Shaw 2010). But several reviews of the mediation research have noted the methodological shortcomings of previous comparative studies. The three most frequent criticisms have been a failure to examine the role of the mediator, the use of a global or vague assessment of satisfaction, and that samples were too small and unrepresentative (Kelly 1996, 2004; Kressel 1997; Beck and Sales 2001; Emery, Sbarra, and Grover 2005). The present study attempts to address these shortcomings.

Despite some inconsistencies in the research findings, divorce mediation seems to be an effective method for producing settlements — in other words, it seems to work. The logical subsequent question then is: which specific elements may be responsible for mediation's positive effects? To date, the relative effectiveness of dispute resolution at achieving settlement when compared with litigation has been credited to the presumably facilitative nature of mediation versus the presumably adversarial features of litigation. Such a simple dichotomy, however, may oversimplify the process and thus bias the interpretation of studies' findings in mediation and litigation research (Shestowsky 2004). Indeed, although many mediators work facilitatively, others may use more non‐facilitative evaluative strategies (Riskin 1996, 2003; Shestowsky 2004). A more evaluative mediator will place a greater emphasis on the legal implications of a dispute (Mayer 2004). Yanick Sarrazin and his colleagues (2005) found that in divorce mediation, facilitative approaches were more effective than directive approaches, although they found little difference in the effectiveness of the two approaches when the divorcing couples no longer had significant emotional involvement with each other. Interestingly, however, mediators are reluctant to define their own approaches as evaluative (Charkoudian et al. 2009), and express broad differences of interpretation when defining precisely what constitutes a facilitative or evaluative approach to mediation (Picard 2002).

The responsibility of a lawyer in the Western legal tradition is to maximize the interests of his or her own client (i.e., zealous partisanship) without regard to the impact on the client's opponent (Macfarlane 2008). But the growing use of alternative dispute resolution has not gone unnoticed by divorce attorneys, many of whom have begun to adopt more facilitative approaches in their practices, such as trying to deescalate the conflict and improve the client's relationship with his or her ex‐spouse (Beck and Sales 2001; Wright 2007).

Overall, such stylistic variations thus imply that divorcing individuals may have been assisted by divorce professionals who vary in the degree of facilitation they provide. When comparing outcomes from mediation versus litigation, researchers should be comparing cases among professionals who displayed a similar level of facilitation, that is, comparing apples with apples not with oranges (Beck and Sales 2001).

Clearly, where a divorce falls on this continuum can influence the nature and quality of the eventual settlement. The traditional adversarially oriented divorce attorney seeks to achieve the best possible settlement for his or her client via either effective argument in court or effective negotiation during settlement talks. A major premise of those who favor more evaluative mediation is that, by offering advice and helping more actively shape the settlement discussion, they can help the parties achieve more mutually satisfactory and more durable settlements. In contrast, traditionally facilitative mediators and nontraditional facilitative divorce attorneys seek to help divorcing individuals more actively control the process and create their own agreements (Welsh 2004). Which approach works best and for which clients is an open question: we found little or no research examining the relationship between the degree of facilitation offered by both mediators and attorneys and the quality of divorce settlements.

Determining empirically which mediation practices produce the highest quality dispute resolution outcomes is a challenge for mediation researchers (Charkoudian et al. 2009; Poitras and Le Tareau 2009; Ballard et al. 2011).

In general, mediation and litigation are often compared in terms of the number of settlements they generate. An assumption of such a comparison is that a higher number of signed agreements indicate a better dispute resolution process (Donohue, Lyles, and Rogan 1989; Herrman et al. 2003). One concern, however, is that an exclusive focus on settlement rates favors the needs of policymakers and practitioners over what is best for divorcing couples and their families (Wright 2007). As a result, satisfaction with dispute resolution outcomes is often the preferential quality indicator for divorce researchers (Kelly and Gigy 1989; Wissler 2004). But the use of a single global assessment can limit construct validity because the item may be inadequate to measure what was intended to be measured (Beck, Sales, and Emery 2004).

Jean Poitras and Aurélia Le Tareau (2009) have proposed alternative issues to consider in evaluating the effectiveness of mediation generally. Reviewing previous studies in mediation, the authors developed a five‐dimensional common model of quality of agreement. Their dimensions for measuring quality are parties' assessment of their level of confidence in the implementation, their satisfaction with the arrangements, their assessment of the mediator's effectiveness and of the fairness of the procedure, and how well the parties reconciled their differences (Poitras and Le Tareau 2009).

Combining disputants' process evaluations with their evaluations of outcomes into a single quality of agreement construct could be problematic. Indeed, although processes considered to be equitable and fair often produce high‐quality agreements, studies have also indicated that mediations characterized by high procedural fairness can still result in stalemate (Pearson and Thoennes 1989; Kressel 1997). In addition, just because someone agrees to sign a settlement does not mean that he or she subjectively experiences it as a high‐quality agreement (Poitras and Le Tareau 2009). Consequently, it is important to define what constitutes a high‐quality divorce agreement.

According to Donald Saposnek (in Moore 2003), high‐quality agreements share the following characteristics: they are clear, detailed and specific, balanced and fair, and embody positive attitudes. Although sometimes partial agreements may be preferable to no agreement at all, achieving a comprehensive divorce resolution is also an important goal for divorcing couples and mediators (Gibson 1999; Moore 2003). Developing and implementing research methodologies that examine all relevant aspects of divorce agreement quality is an ongoing challenge for researchers in this field.

The third methodological challenge for divorce mediation researchers concerns the representativeness of research findings. Ethical and logistical considerations often prevent researchers from gaining access to data from actual mediations, and much mediation research has relied on small samples (Beck and Sales 2001), which can diminish the external validity of findings.

In a similar fashion, it is important to specify the idiosyncratic policy framework that guides the mediation processes under examination. For example, findings generated in countries or states with mandatory mediation or with extensive traditions of no‐fault divorces, for example, may not generalize to regions or countries operating under different rules and traditions. This study, for example, examines the experiences of a sample of divorced Flemish speakers in Belgium. During the late 1990s, the first mediation legislation appeared in the Belgian chamber and senate, and the Belgian judicial code was amended in 2001. Under this new law, court proceedings can be suspended by consensual agreement if the disputing parties agree to pursue mediation.

A turning point was reached in 2005 when parliament passed a second and more wide‐ranging mediation act (Casals 2005). This mediation act embodied a number of key principles: namely that mediation be a confidential and privileged process, that the mediator be an independent, impartial, and competent mediator, and that it be voluntary. A Federal Mediation Commission was established that institutionalized mediator certification procedures and mediation policies.

In a similar vein, changes made to divorce laws in 2007 further altered the divorce landscape. Specifically, this new law allows couples to either legally divorce on the basis of “mutual consent” or on grounds of “irretrievable breakdown.” (The latter is also known as the no‐fault procedure.) Using the irretrievable breakdown option, a person can attain divorce subsequent to providing proof of an irreconcilable marital disruption, after separation of a specific duration, or following a repeated and well‐considered request to divorce (Swennen 2010). Thus, both parties do not have to agree to the divorce before it can take place. By contrast, the mutual consent procedure requires couples to agree on all relevant family issues (i.e., children, finances, and property) prior to any judicial divorce pronunciation. The impact of both these legal changes on dispute resolution outcomes is not yet fully understood.

Studies that examine how the behavior of divorce professionals affects the quality of divorce agreements in a representative community sample of divorcing individuals are nearly nonexistent. The present study is designed to help fill this gap in the research.

We hypothesize that the degree of facilitation provided by the divorce professional will have an impact on the quality of the divorce settlement. Specifically, our hypotheses are that a more facilitative practice (in litigation as well as in mediation) will be positively correlated to the quality of divorce settlements (Hypothesis One.) We further hypothesize that divorcing couples who participate in mediation will achieve higher quality divorce settlements than will similar couples who litigate their divorces (Hypothesis Two). Furthermore, because previous studies have shown that such individual variables as the disputant's gender and status as initiator or non‐initiator of the divorce, as well as such couple interaction variables as the level of pre‐divorce conflict in the relationship, can all affect the quality of divorce agreements, we have controlled for these variables in our analysis.

This study is part of the Interdisciplinary Project for the Optimization of Separation Trajectories Project (IPOS), which is a collaboration among psychologists, lawyers, and economists at Ghent University and the Catholic University of Leuven. All spouses who divorced between March 2008 and March 2009 in the courts of four Flemish cities (Antwerp, Ghent, Kortrijk, and Mechelen) were invited to participate in the project. The IPOS data set contains 2,146 surveys of which 1,849 were fully completed. This represents an overall participation rate of approximately 30 percent. The average duration of participants' marriages was approximately 14.5 years (median 13.08 years), which is consistent with the median marriage duration of thirteen years that was reported for marriages ending in divorce by the Belgium National Institute of Statistics (NIS 2007).

After collecting demographic data (family background, the type of legal procedures selected, and economic status), we randomly divided participants to answer either questions about the relationships between the divorcing parents and their children, the relationship between the ex‐partners, or the personal qualities of the attorney or mediator who assisted in achieving an agreement. (The perceptions of disputants on the behavior or personal qualities of lawyers who may have sat in on mediation sessions as counsel were not collected or analyzed in this study.) For analysis, we selected 469 participants from the main sample and 117 participants from the subsample examining the personal qualities of the divorce professional. We only retained those participants who completed their questionnaires, had reached a written divorce agreement, and had received the assistance of a mediator or lawyer.

Because this study was largely intended to expand on previous research, we also needed to operationalize the quality of divorce agreements and facilitative practice as experienced by participants. To this end, two groups of divorce professionals with various professional backgrounds (law, sociology, psychology, and social work) helped us to pretest and revise all items in the survey. Furthermore, we determined the psychometric characteristics of all scales using an exploratory factor analysis and calculating the Cronbach's alpha internal consistency coefficient. Each scale reached an optimal structure with factor loadings that well exceeded 0.50 and Cronbach's alpha values ranged from 0.80 and 0.89. See Table One for an overview of Cronbach's alpha values and factor loadings for each item.

Our dependent variable was the quality of divorce agreements. Respondents were asked to rate their satisfaction with four agreement characteristics on a seven‐point Likert scale ranging from 1, “completely unsatisfied,” to 7, “completely satisfied.” These characteristics were selected according to the criteria identified by Saposnek (in Moore 2003): was the agreement tailored, fair, comprehensive, and clear?

Our independent variables were whether the professional engaged in facilitative practice, the professional's personal qualities, the couple's divorce trajectory, their levels of pre‐divorce conflict, and participant's gender and divorce initiator status. We measured the first independent variable, facilitative problem solving, by asking participants to agree to six statements using a seven‐point Likert scale ranging from 1, “completely disagree,” to 7, “completely agree.” The statements inquired how interest‐based, how informative, and how structured they thought their meetings with their mediators or attorneys had been.

Based on the work of Joan Kelly and Lynn Gigy (1989) and Paul Dierick and Germain Lietaer (2008), we developed eight statements for measuring respondents' perceptions of how empathic, authentic, accepting, and impartial the attorney or mediator was. Each item was rated by the participants on a four‐point scale ranging from 1, “not applicable,” to 4, “extremely applicable.” Using exploratory factor analysis, we identified one six‐item factor that can stand for several basic relationship attitudes we refer to here as “Rogerian” after psychologist Carl Rogers (1961), who argued for their value in therapeutic practice. We developed a second two‐item factor to represent the professional's impartiality.

For this study, we obviously differentiated between those couples who went through litigation and those who went through mediation. We further differentiated between participants who legally divorced through the mutual consent procedure (comprehensive agreement prior to legal divorce pronunciation) and those participants who chose to divorce on the grounds of irretrievable breakdown (legal divorce pronunciation prior to a settlement).

To measure the level of pre‐divorce conflict, we asked participants to respond to three modified statements from the conflict properties subscale of the Children's Perception of Interparental Conflict (Grych, Seid, and Fincham 1992). Previous studies have demonstrated that these items are suitable for use with adults (Bickham and Fiese 1997; Kline, Wood, and Moore 2003). Participants were asked to respond on a five‐point Likert scale and represented separate conflict dimensions for frequency of conflict (“How often did you and your ex‐partner have conflicts before the breakup?”), intensity of conflict (“How intense were these conflicts before the breakup?”), and the frequency of resolution (“How often did you and your ex‐partner reach a solution for these conflicts?”).

We also wanted to control for respondents' gender and initiator status. We asked each respondent whether he or she had initiated the divorce, whether his or her ex‐partner had, or whether the divorce had been jointly initiated.

We executed exploratory factor analyses using the statistical modeling program Mplus6.11 (Muthén and Muthén 1998–2010), and we tested the research hypotheses using multiple regression analyses carried out in R2.12.2 (R Development Core Team 2011). For each variable, the variance inflation factor was below ten (see Tables Two and Three for an overview), which indicates that multicollinearity has not compromised our regression analysis.

Survey participants were 469 divorcing individuals; 56 percent (265) are women. Participants ranged in age from twenty‐one to seventy‐six with an average age of forty‐four. An overwhelming majority of participants, 80 percent (376), divorced with mutual agreement, only 20 percent (ninety‐three) of participants opted for a divorce on the basis of irretrievable breakdown. (This is in contrast with a general rate of 52 percent mutual consent and 48 percent irretrievable breakdown registered in courts.) While in the mutual consent procedure both partners need to appear before court, the irretrievable breakdown procedure does not oblige both partners to be present at court. This, we believe, reduced the availability of participants who opted for irretrievable breakdown divorces, and thus explains why our pool of participants in that category is proportionally smaller than in the general population.

A mediator assisted 30 percent (139) of participants to reach agreement, 70 percent of participants were helped by a lawyer (330). (We did not record data regarding which of these mediations did or did not have lawyers present as counsel.) For the mutual consent divorce subset, a higher proportion of participants, 35 percent (130), mediated. For the participants with irretrievable breakdown divorces, however, only 10 percent (9) were involved in mediation.

Forty‐nine percent of all participants (231) initiated the divorce; for 32 percent (149), the divorce was initiated by the spouse, and for the remaining 19 percent (89), the divorce decision was initiated by both partners.

On the whole, 215 participants (46 percent) reported high pre‐divorce conflict levels (with a score of 4 or 5 on a five‐point scale with 1 meaning low pre‐divorce conflict and 5 meaning high pre‐divorce conflict). One hundred three participants (24 percent) reported low pre‐divorce conflict levels (scores 1 and 2).

The divorce professionals were generally also perceived to be fairly facilitative. More specifically, the majority of participants (72 percent) reported that the professional displayed a high level of problem‐solving behaviors during mediation or litigation (scores 5, 6, or 7 on a seven‐point scale).

Thirty‐three participants (7 percent) reported that the mediator or lawyer displayed few problem‐solving behaviors (scores 1, 2, and 3). A similar pattern emerged for how the personal qualities of the professional were perceived. That is, whereas 12 percent of participants found their professional to display few Rogerian qualities (score 1 on a four‐point scale with 1 meaning extremely low Rogerian qualities), 12 percent of participants scored their divorce professional much higher (score 4), while the majority (77 percent) reported their divorce professional as somewhat Rogerian (scores 3 and 4). The majority of participants (81 percent) also perceived the mediator or lawyer to be very impartial (score 4 on a four‐point scale with 1 meaning not at all impartial). A small proportion of participants (12 percent) did not perceive their professional as impartial (score 1 and 2).

Regarding the outcome of the dispute resolution, only 52 participants (11 percent) rated their agreement as low‐quality (scores 1, 2, and 3 on a seven‐point scale), while 307 participants (67 percent) were much more satisfied, giving their agreements scores of 5, 6, or 7 on a seven‐point scale. (See also Table Two for an overview of all sample and subsample characteristics.)

Our results support our hypothesis that the mission of the divorce professional correlates significantly with participants' satisfaction with their divorce agreements. Specifically, participants who had meetings with a mediator (i.e., mediation) reported higher‐quality agreements than did corresponding individuals in litigation (beta = −0.33; standard deviation: 0.13; probability < 0.01). What is more, our results indicate that the professional's level of perceived facilitativeness has an impact on the quality of the divorce agreement. Indeed, the more that divorcing individuals perceived that their professionals were displaying problem‐solving behaviors, the more highly they rated their divorce agreements (beta = 0.37; standard deviation: 0.05; probability < 0.001). We also found that the legal trajectory of the divorce had no significant impact on the quality of the agreement — that is, participants who obtained a mutual consent divorce reported similar satisfaction with their agreements when compared with individuals who divorced on the grounds of irretrievable breakdown (probability > 0.1).

At the level of individual and couples characteristics, our analysis revealed that the initiator status of the participant did not predict the quality of agreements (probability > 0.1) nor did the gender (probability > 0.1). In contrast, our analysis did reveal that the level of marital quality significantly predicted immediate dispute resolution outcomes. The more unresolved, intense, and frequent were the couple's conflicts, the worse the quality of their divorce agreements (beta = 0.13; standard deviation: 0.06; probability < 0.05). Table Three summarizes our regression analysis.

Specifically, the more impartiality (beta = 0.33; standard deviation: 0.16; probability < 0.05) as well as the more Rogerian qualities (beta = 0.47; standard deviation: 0.17; probability < 0.01) the professional was perceived to display, the more fair, detailed, tailor‐fit, and comprehensive the agreements were judged to be by the participants. Our subsample results matched the larger sample in that we found a significant positive effect for mediation when compared with litigation (probability < 0.05), and we found no significant effect for the type of divorce (mutual consent versus irretrievable breakdown) (probability > 0.1), initiator status (probability > 0.1), or gender (probability > 0.1).

In contrast with our main sample results, however, we did not find the pre‐divorce conflict levels for the subsample to have a significant effect on the quality of divorce agreements, although the fact that beta coefficients for pre‐divorce conflict in both samples are identical suggests that this may reflect that the sample size of the subsample is too small to determine statistical significance, and this result may be considered inconclusive. An overview of the output for the multiple regression analysis on the subsample can be found in Table Four.

In Belgium, recent legislative changes and growing public awareness have encouraged greater use of mediation in divorce, and the practice is no longer rare, but empirical knowledge of what contributes to high‐quality divorce agreements is scarce. With the current study, we seek to build on the results of earlier mediation research in several ways. First, we present a research framework that not only accounts both for differences between litigation and mediation but also considers the actual behaviors of professionals (mediators or attorneys) assisting divorcing couples; specifically, we are concerned here with the professional's use of facilitative (problem‐solving) practices. Second, rather than relying on settlement rates alone, which have been the main concern of policymakers, we have sought to appraise the quality of divorce agreements from the perspective of divorcing individuals themselves. Some interesting findings have emerged.

A more facilitative practice on the part of mediators and attorneys was found to correlate with higher‐quality divorce agreements. That is, by helping disputants focus on interests, exchange information, and by structuring the process, mediators and attorneys helped improve dispute resolution outcomes. These problem‐solving behaviors have a long tradition in interest‐based negotiations and are associated with several popular mediation models (Fisher and Ury 1981; Hensler 2000).

The professional's Rogerian personal qualities of empathy, acceptance, and authenticity were also found to influence the quality of divorce agreements. Within psychotherapy, these qualities were initially set forth to counter the omnipresence of the authoritarian and detached psychoanalyst (Rogers 1961). Similarly, our results suggest that dispute resolution professionals should shift from traditional adversarial and reserved professional attitudes toward behaviors that convey more warmth, authenticity, and caring. What is more, when the professional was perceived to display a higher level of impartiality, the disputants were more likely to perceive their agreement as higher quality.

The use of facilitative principles may clash with some fundamental criteria of the legal profession, such as zealous advocacy. But some authors have argued that lawyers should respond more to their clients' perceptions and wishes for process ownership and self‐determination as proclaimed by facilitative mediation advocates (Macfarlane 2008). Others have suggested that lawyers may be engaging in more facilitative efforts in an attempt to prevent losing market share in the divorce industry to mediators (Wright 2007). The fact that lawyers increasingly are trained and certified as family mediators (Belgian Federal Mediation Commission 2011) may further predict such facilitative efforts.

A fundamental goal of facilitative practice in mediation is that it will generate mutually satisfactory agreements rather than merely a written settlement. This finding suggests some support for the results of other studies that have argued that divorce mediation may fail to change fundamental relationship patterns between ex‐partners (Pearson and Thoennes 1989). In another study, high levels of anger in divorcing couples before mediation were found to be predictive of inferior mediation outcomes and fewer problem‐solving behaviors during the mediation process (Bickerdike and Littlefield 2000). But despite the challenge created by high levels of pre‐divorce conflicts, facilitative mediators were nonetheless reported to be more effective than directive mediators when working with such couples (Sarrazin et al. 2005).

The negative influence of pre‐divorce conflict is an issue of great concern given that as many as 50 to 60 percent of all couples that use divorce mediation report physical aggression in their relationship (Thoennes, Salem, and Pearson 1995; Mathis and Tanner 1998; Ballard et al. 2011). Researchers have also argued that mediators do not sufficiently address violence, for example, when it is identified in a couple's relationship (Johnson, Saccuzzo, and Koen 2005).

To work with high‐conflict couples, some scholars have suggested that mediators should adapt their methods. Such adaptations could include initiating individual therapeutic meetings with each partner to address emotional issues (Irving and Benjamin 2002); using individual pre‐mediation sessions to assess whether there is violence in the relationship (Beck, Sales, and Emery 2004); and initiating joint sessions only after some initial agreements have been reached (Heister 1985). Finally, some scholars have argued that anger in divorce can be therapeutic: feelings of anger may decrease depression by increasing the partners' emotional distance from each other. (Emery 1995; Emery et al. 2001).

In contrast to previous studies (Beck, Sales, and Emery 2004), we found that the gender, initiator status, and type of legal divorce did not independently contribute to participants' satisfaction with their agreements. The reasons are unclear, but it is plausible that the influence of these factors is neutralized when the professional engages facilitatively with his or her clients. In this regard, Andrew Bickerdike and Lyn Littlefield (2000) found that while progressing to agreements, family mediators' interventions can neutralize differences between initiators and non‐initiators. As a result, any effects of having the upper hand as a self‐initiator can be expected to diminish during mediation.

Our study supports the contention that mediation, particularly mediation with a facilitative orientation, yields beneficial results for individuals going through a divorce. But determining which specific mediation behaviors are responsible is difficult because of the kaleidoscopic nature of mediation practice. Divorce mediators come from many disciplines and backgrounds including law, social work, and counseling, and many mediators are non‐professional volunteers. They also conduct their mediation practices in various settings (big offices and small offices) and under different conditions (all issues are addressed or only limited issues are considered, clients' attorneys may or may not be present). Moreover, mediators may choose a variety of styles: facilitative, evaluative, transformative, narrative, insight‐centered, etc.

Because the majority of participants in the mediation group (130 of 139) had obtained their divorces via mutual consent, we must exercise caution when generalizing the impact of mediation to individuals who legally divorced through the irretrievable breakdown procedure; the small sample size of that group in this study undermines the statistical validity of our findings for that group and calls for the replication of our findings with a larger sample.

Furthermore, we restricted our studied population to heterosexual individuals who had been legally married. It would be interesting to see if our findings replicate in different populations, such as separated gay and lesbian couples or formerly cohabiting straight individuals who never married.

Likewise, confining to only those participants who divorced with the explicit assistance of a mediator or lawyer does not preclude that other professionals could also have been involved. For example, the Belgian legislation allows the judge to modify any agreements if deemed in conflict with the best interest of minor children or civic policy (Casals 2005). It would be worthwhile to find out how frequently judges change agreements and how this subsequently influences the experienced quality of agreements. Given that issues related to children (child support, custody, etc.) are often the most high‐stakes issues in divorce, it would be worthwhile for future studies to determine if there are any differences in the quality of divorce settlements between those couples who have children and those who do not.

In this study, we operationalized facilitative practice as an attitude displayed by mediators and lawyers in the dispute resolution process. Some research, however, has demonstrated that mediators flexibly amend practice styles during mediation sessions (Goldfien and Robbennolt 2007; Kressel 2007). A future examination of the proximal contributing factors for the quality of divorce agreements could expand on our findings by more precisely observing and analyzing various styles during role played or actual mediation sessions. In a similar vein, future studies should go beyond a unitary measure of what constitutes a good divorce agreement (in this case, disputants' satisfaction with the agreement). For example, it is possible that subjective evaluations of more narrowly ranged issues (e.g., property‐only mediations) may differ in experienced quality from more complex mediations (e.g., mediations on parenting time or child support). It is also possible for third parties to objectively evaluate the value and fairness of property and financial agreements, and further studies could also compare disputants' subjective satisfaction with the objective assessment of third party experts.

Mediation and litigation do not occur randomly, and consequently, the samples in each category will not be truly random, that is, parties were not randomly assigned to either mediation or litigation conditions, but chose them. That choice could reflect characteristics of either the couple or the divorce that could bias results. Better than such cross‐sectional designs are studies that track individuals for several months or years preceding and following the initiation of legal divorce procedures. Such an undertaking could help to identify any possible long‐term correlates of facilitative practice or quality of agreements.

Determining via research what works best and how in any given mediation does not necessarily call for a standardization of practice because mediation necessarily requires flexibility and the needs of clients will vary according to their own circumstances and the nature of their conflict, as well as for sociocultural and economic reasons.

The principles of facilitative practice may seem simple, but they can be difficult to master (Bowling and Hoffman 2000). Our findings support the idea, well established in the mediation literature, that mediators benefit from self‐awareness in their practice. Our findings also have some interesting implications for lawyers in particular: how can lawyers balance the apparent benefits to their clients of a more facilitative practice with the advocacy requirements of the legal practice? How can they balance a mutual‐interest‐based approach with the zealous partisanship that legal professionalism requires? Does pursuing a more facilitative practice represent a conflict of interest for litigators? Or are there ways in which they can incorporate a more facilitative model while simultaneously representing their clients zealously? What happens if one party is represented by a more facilitatively oriented litigator and the other is not? Does this put one party at a disadvantage? At the very least, ethical practice requires that both mediators and litigators be transparent with clients and explain their practice orientations up front so that clients have the opportunity to choose what they feel is best for them.

Our results also support the argument that settlement rates alone should not be the sole criterion for evaluating mediation effectiveness. Indeed, divorce agreements vary considerably in their fairness, practicality, and durability. The more involved the clients are in determining the particulars of the agreement, the more satisfied they are likely to be with the final version, and the more likely it is that the agreement will not generate additional conflict or require further negotiation. Our results also suggest (and additional research could confirm) that, for couples with a history of intense conflict, professionals need to do more than just respond to clients' needs in a facilitative manner.

The results of our study indicated that facilitative practice can optimize the quality of divorce agreements. Our evidence further suggests that facilitative practice in divorce is not only being undertaken by mediators but has also become a part of current legal practice. Further research should examine such conflict resolution practices in additional contexts and more precisely pinpoint how, under which circumstances, for whom, by whom, and when facilitative problem solving in divorce can be most effectively utilized. The quality of divorce agreements, as subjectively experienced by the divorcing individuals themselves, should continue to be one important measure of the effectiveness of divorce professionals.