Family mediation: The rhetoric, the reality and the evidence

This paper examines the research evidence which has been assembled internationally in recent decades in respect of family mediation, and relates it to the rhetoric which has encouraged people to regard mediation as a dispute resolution process which is superior to others. A process that promotes co-operative dispute resolution is not new: it recognises and respects the right of each party to a dispute to fashion a mutually acceptable agreement in a consensual manner which meets their particular needs. Continuing conflict can be damaging to interpersonal relationships, so a process which focuses on co-operation and problem-solving is an important option.

As the incidence of divorce throughout the Western world rose during the latter half of the last century, mediation was increasingly used to help couples reach mutually acceptable agreements about arrangements for children, the distribution of property and the allocation of finances after separation. Court-connected mediation services were first established in California in 1939 and are now commonplace across the USA. Over the last thirty-five years, family mediation has become an integral feature of family justice systems across Europe, North and South America, Africa, Asia and Australasia, and a wide variety of models exist. For the most part, unless there is a requirement for couples to meet with a mediator to explore whether mediation is a suitable mode for resolving their disputes, mediation appears to play a strategically important but relatively minor role in the separation and divorce process. This has caused concern in countries where it was widely anticipated that mediation would be the preferred route through separation and divorce. The research evidence helps to explain the reasons why not all couples need or want to mediate, but why having the option to mediate is very important.

Although the social, economic and cultural contexts in which family mediation is practised vary considerably, there are many common factors which render the learning gained from international experience particularly helpful. The international literature, however, does not refer to a uniform body of work, nor to a homogeneous service. Family mediation is a shorthand term for a varied and somewhat fragmented approach to dispute resolution, which developed as an alternative to litigation through the courts and arm’s-length negotiation via partisan lawyers. In the latter, each party instructs a lawyer and all negotiations between the parties are conducted via the lawyers. In this approach there is no direct communication between the parties and each is dependent on lawyers to convey messages accurately. In practice, such negotiations have tended to escalate conflict. In many jurisdictions mediation services are attached to the courts and form an integral part of the legal process; other mediation services are community-based, enabling all couples with issues in dispute to access them whenever they wish; mediation is also offered privately by psychologists and lawyers, particularly in the USA and the UK; and in a few jurisdictions, couples are required to meet with a mediator prior to or during court proceedings to attempt to resolve the disputes. Increasingly, lawyers acting for the parties refer their clients to a mediator before they get involved with court processes. The focus in all these services in on empowering parents to take responsibility for their children’s and their own futures.

Sifting the evidence

Many of the early research studies focused on determining client satisfaction and the extent to which mediation processes were beneficial for the parties, as well as for settlement rates (Beck & Sales, 2000; Kelly, 1988, 1990; Pearson & Thoennes, 1989; Walker et al., 1994).The first-generation research conducted in the USA, Canada, Australia and the UK constitutes the largest body of empirical mediation research. Most studies were initiated to assess the often sweeping claims about the benefits of mediation. Much of the research did not, however, distinguish between or make allowances for different approaches to mediation practice (such as single sessions, multiple sessions, co-mediation) and different settings (such as court-connected and community-based services). Nevertheless, there was a remarkable degree of convergence within the studies on many issues. This is important because there has been relatively little new research in the last decade.

There are, however, significant gaps in the evidence. There is a paucity of research that investigates the relative benefits of different types of mediation: problem-solving, transformative, therapeutic, facilitative, and evaluative. We know relatively little about the efficacy of these processes or for whom they might be more beneficial. Kelly (2004) has also noted the lack of research about mediator behaviours and interventions, participant characteristics and behaviours, and the interactions and relationship of all these with and to outcomes. It may be some time before we have a clearer view about «the interaction of emotions and personality attributes that individuals bring to the mediation setting» (Kelly, 2004, p. 31) and about the approaches that are more likely to result in positive outcomes. This paper briefly considers the key evidence in respect of mediation processes and outcomes, and refers to specific issues such as the role of children in mediation and new directions in mediation practice.

Mediation processes

Mediation practice is infinitely variable. While we can distinguish between approaches which tend to be very brief (often, single mediation sessions run by court-connected mediators) and those which take longer (often associated with community-based programmes), more subtle distinctions which focus on the content of mediation are less easy to discern. In her review of the first decade of mediation research, Kelly (1996) emphasised the lack of attention that had been given to studying the mediation process and attempting to understand the impact of process on the outcomes identified. Nevertheless, there have been some studies which have examined mediator characteristics and some which have examined mediator styles, on the basis of direct observations of practice and content analyses of conversations in the mediation room. These studies are important because they enable us to think about appropriate models of mediation and reconsider the skills needed by mediators at a time when many of the cases presenting to mediation services are particularly complex and/or involve couples who are in high conflict, often with long-standing disputes.

Mediator styles

Although the number of robust process studies remains relatively low, research has attempted to uncover why, how, and when mediators use certain techniques and skills and how they influence processes and outcomes (Davis, 1988a,b; Dingwall & Greatbatch, 1993, 2001; Hayes, 2005; Pearson & Thoennes, 1988; Slaikeu et al., 1985a,b). Early studies of mediation process distinguished between two basic styles: bargaining and therapeutic (Silbey & Merry, 1986). The bargaining style assumed that the parties had the capacity to negotiate a settlement and focused on joint decision-making. The therapeutic style assumed that the parties would not be ready to participate in joint decision-making until underlying emotional issues had been explored and dealt with. An analysis of audiotapes in England & Wales also identified these two styles. Dingwall & Greatbatch (1991; Greatbatch & Dingwall, 1989)) set out to examine two specific claims made for mediation: first, that party-controlled settlement-seeking (private ordering) enhances parental responsibilities; and second, that mediation is uniquely capable of taking account of children’s interests and providing outcomes which are individually tailored to each family’s circumstances.

Dingwall & Greatbatch (1991) found that some mediators mainly used a bargaining style and neither reviewed the couple’s relationship nor encouraged discussion of emotional issues. Other mediators adopted a therapeutic style which encouraged discussions of the couple’s relationship and conduct. However, when it came to the stage in the mediation when negotiations needed to take place, all the mediators used techniques that kept the parties focused on reaching agreements. Dingwall & Greatbatch referred specifically to the mediators’ use of sanctions, which involved mediators directly challenging the parties if they remained preoccupied with their own interests and concerns rather than focusing on the best interests of their children. The researchers commented that the prospects of successful mediation seemed to be enhanced by the extent to which the mediators were willing to use more direct interventions. References to children’s interests, it seems, were made in the context of mediators pressing for particular outcomes, thereby applying a degree of moral pressure on parents to behave in certain socially acceptable ways.

In later studies, Dingwall & Greatbatch (2000) argued that mediators in England had become more sophisticated in both their management of the mediation process and their guidance of clients towards preferred outcomes. While they found that there was still a wide variation in the practice of mediation, the process was strongly influenced by the nature of the issues under discussion. Mediator pressure to reach particular outcomes relating to finances was less marked than in respect of arrangements for children. The mediators found it easier to remain impartial in respect of property and finance disputes than in disputes about where children should live and how both parents would retain contact. These observations are similar to those of other researchers. Studies in the USA found that mediators in cases which settled, were more in control of the interaction and that they offered interventions and elicited options which moved the parties towards particular settlements (Pearson & Thoennes, 1988).

Strategies and techniques

Kruk (1998) in Canada and Hayes (2002a, b) in England examined the strategies, techniques and interventions used by mediators in varying client and dispute situations via self-report methods. In Canada, the more experienced the mediators, the less likely they were to adopt a neutral position in relation to process and the more likely they were to be interventionist. Significant differences were found between lawyer mediators and mental health mediators: the latter were more likely to take a therapeutic stance and the former were more focused on negotiating settlements. Most mediators, however, considered structured negotiation to be the foundation of their practice. Kruk concluded that the diversity of mediation practice constituted both a strength and a weakness: a strength in that the diversity of mediation client populations and the complexity of dispute characteristics call for a plurality of methods; and a weakness to the extent that mediators continued to be divided with regard to the approaches which were considered appropriate and clients were not necessarily aware of the kinds of service they could expect. Mediators have been particularly divided about whether mediation should be therapeutic.

A more recent study in England & Wales (Hayes, 2002a, 2005) examined the strategies, techniques and interventions used by family mediators in a mix of community-based and private mediation practices, at different stages in the mediation process and in different client and dispute situations. The study revealed that the focus of the mediation process was highly pragmatic and that managing conflict and facilitating communication to facilitate settlement-seeking were integral to the process, irrespective of the characteristics of the clients and the disputes. Problem-solving techniques and active negotiation were used by mediators in various combinations. The results demonstrated that mediators saw their role in the process as active, but appreciated the importance of clients maintaining control over the content. Moreover, particular background characteristics created systematic variations in practice: professional background, practice setting and mediator experience were the most influential characteristics.

Hayes’ observations confirmed that mediators with a social welfare/mental health background were much more comfortable dealing with children and parenting issues and seemingly less at ease addressing finance and property disputes than mediators with a legal background (Walker & Hayes, 2006). Mental health/social welfare professionals were more likely to use active listening, counselling and therapeutic techniques in mediation. Lawyer-mediators, by contrast, were more likely to use reality-testing techniques and made greater use of direct questioning. There is strong evidence from this study that lawyer-mediators and mediators from a social welfare background practise in systematically different ways.

Co-mediation

One way of optimising mediation practice has been to promote co-working, whereby mediators from different backgrounds work in partnership. In complex cases in which all the issues consequent on divorce are addressed in mediation, the co-mediation model seems to have worked smoothly and efficiently and has been empowering for all the participants (Walker, et al., 1994). While Dingwall and Greatbatch (2000) acknowledged the benefits of co-mediation they were concerned that clients may be put under undue pressure from co-mediators and that mediation might be perceived as coercive and intimidating.

A pilot comprehensive co-mediation project in Winnipeg, Canada involving lawyers and family relations specialists working in partnership, having been trained together in mediation, enabled family relations specialists to understand legal and financial matters, and lawyers to understand child development, family dynamics and the emotional content of the separation process (McKenzie, 2001; McKenzie & Pedersen, 2003). This resulted in a more collaborative approach to mediation but there are issues relating to the costs of delivering a co-mediation model.

A staged approach

Another process issue which has emerged is whether each party should be invited to an intake meeting without the other party being present. Views about the efficacy of this model are divided. Research has indicated that many clients prefer this staged approach (Walker et al., 1994, 2004; Walker, 2001). Davis et al. (2000) argued that any client who might be reluctant or unsure about embarking on mediation would probably prefer to be offered an individual appointment rather than being put in the position of having to express concerns in front of the other party. Where both parties are prepared to attend an intake session, either individually or jointly, mediation is usually the option chosen to resolve disputes. Persuading both parties to consider mediation seriously is an important indicator of mediation take-up, providing support for mandatory referral to mediation, at least to an intake meeting, particularly when clients are legally aided and settling their disputes from public funds.

Choosing to mediate

Research has indicated that couples facing separation and divorce want someone to help them to ‘sort out the mess’ and restore some kind of order (Davis, 1988a; Walker et al., 1994, 2004), and that they value legal advice in addition to the kind of legal information which is customarily provided by mediators. Family lawyers continue to provide security, advice, comfort, and a safety-check on private ordering. Mediation is now considered to be an additional resource within the family justice system and not an alternative to legal advice and representation. While the aim has been to encourage couples to consider mediation before they resort to litigation, research indicates that such aspirations are unrealistic unless mediation becomes a mandatory requirement. Research undertaken to test the implementation of mandatory information meetings under the Family Law Act (England & Wales) 1996 found that of 1,838 people who had received extensive information about mediation, just 7 per cent had used a mediation service in the months following, and some of these couples had dropped out of the process or found that mediation did not work for them (Stark & Birmingham, 2001). When these research respondents were followed up two years later, just 10 per cent of the 1,491 who responded had been to mediation. A further 2 per cent had attended a preliminary meeting with a mediator but had not gone on to enter the mediation process (Walker et al., 2004).

Reflecting on a decade of mediation research, Pearson & Thoennes (1988) examined the evidence from three early mediation projects in the USA: the Denver Custody Mediation Project (CMP), which involved 160 couples who undertook child-focused mediation during 1979–81; the Divorce Mediation Research Project (DMRP) which involved 450 mediation clients using court-connected mediation in three localities (Los Angeles, Connecticut and Minneapolis); and a study of the mandatory mediation of child support issues in 320 cases in Delaware. In the CMP study, approximately one third of individuals who were offered free mediation refused it. Typically, one party was willing to try mediation but the other was not. Women mostly cited mistrust of their partner or a desire to avoid contact as the main reasons for refusing mediation. Men typically explained that they expected to ‘win’ in the legal process or were sceptical about whether mediation would work. Men who mediated tended to be in higher status occupations and generally had higher incomes than those who did not mediate and reported better spousal communication patterns. They had usually been encouraged to mediate by their respective lawyers. Pearson & Thoennes concluded that as long as mediation is a voluntary option, it will be used primarily by couples in higher socio-economic groups who are managing to retain some level of ongoing communication.

Mediation users in Northern California (Kelly & Gigy, 1989) gave a range of reasons for choosing to mediate. These were to: reach agreements satisfactory to both parties; reduce or avoid hostility; reduce the cost of obtaining a divorce; reduce contact with lawyers and courts; seek a fair property division; and retain a friendly relationship with the other party. One group were motivated by wanting a divorce process that was cost-efficient, amenable to more personal input and control, avoidant of lawyers and legal procedures, and advantageous in terms of support and property agreements; another were primarily motivated by wanting to end the relationship with the minimum of hostility, obtain mutually satisfactory agreements, improve communication, and remain friendly in the future. In both groups, men were significantly more positive about entering the mediation process than women. Indeed, the relationship between having some control over the decision to divorce and a positive attitude to mediation was particularly strong for men, while for women a belief in the other party’s integrity was more strongly correlated to the willingness to mediate.

Kelly & Gigy concluded that while a mutual decision to divorce enhances both men and women’s willingness to mediate, women’s attitudes are shaped by their feelings (of anger or otherwise) towards their partner and their perception of his integrity and level of co-operation, whereas men’s attitudes are shaped by their recognition of their marriage being poor. These findings substantiate those from other studies that show that certain predisposing conditions are relevant in the decision to mediate, and indicate that not all couples are suitable for or will want to take advantage of the mediation process (Stark & Birmingham, 2001). Other influencing factors include the length of the marriage/relationship, the ages of children, and the circumstances of the relationship breakdown. These factors shape the way in which issues in dispute are perceived, presented and settled in any kind of dispute resolution process. Most mediation services have witnessed a broadening of the client base over the last decade: couples across a wider age spectrum; a wider distribution across the length of the marriage; more cohabiting (never-married) couples and more couples who are separating for the second (or third) time; more high-conflict cases; and a wider range of issues and concerns being presented.

Completing mediation

Not all couples who enter mediation actually complete the process. Davis et al. (2000), found that in 32 per cent of the cases that went from an intake session to the mediation process itself, one or both parties subsequently withdrew from mediation during the process. In their comparative study of all issues and child-focused mediation in England & Wales, Walker et al. (1994) found that 46 per cent of couples who went to all issues mediation did not complete the process, as against just 12 per cent of those who went to child-focused mediation. The researchers hypothesised that this considerable difference in completion rates was primarily associated with the complexity of the case: the more complex the issues, the longer the mediation process and the more likely that one or both parties will decide to withdraw prematurely. Withdrawal does not necessarily indicate that mediation has failed or that the parties are unhappy with it. While some couples withdrew because they felt that nothing was being achieved, others felt they had made progress but were unlikely to be able to resolve all the issues.

In their later follow-up study of 152 mediation users Walker et al. (2004) found that only 30 per cent of them had completed mediation to the point where all their objectives had been met. The multiple reasons which the other mediation clients gave for withdrawing from the process were that: there had been little chance of making further progress; mediation had not been achieving anything for them; mediation had been causing more rows; the other party had failed to attend; they were unhappy with the approach of the mediator(s). Other studies have found similarly high rates of non-completion. Kelly & Gigy (1989) found that 43 per cent of clients attending mediation in Northern California terminated mediation before reaching full agreements on all the issues. The most common reason given for premature withdrawal was that mediation was ‘too expensive’, while the unwillingness of one party to attend was the second most common reason given. Nevertheless, non-completion did not necessarily indicate dissatisfaction with mediation, and some couples had clearly decided that they did not need the mediator to help them reach agreements because they were able to negotiate amicably between themselves.

It is inevitable that some people will enter mediation only to discover that it is more challenging than they thought and that they want to withdraw. The effectiveness of mediation, therefore, should not be gauged by the numbers of couples who fail to complete the course or are unable to reach agreements on all issues. Many of them will have made significant progress before terminating mediation. Some will not have done, and mediators worldwide have agreed that mediation is not a panacea for all couples whose relationships are ending.

The links between processes and outcomes

Studies of mediation process, although relatively scarce, have indicated that variations in practice impact both on the experience of mediation and on the outcomes achieved. A number of factors relating to the clients, the mediation model and setting, and the disputes themselves all impact on outcomes. Variables such as the emotional stability of the parties and the couple’s commitment to divorce and to mediation were found to be associated with outcomes in some of the first generation of research studies (Kelly & Gigy, 1989). Other studies identified the intensity and duration of the dispute(s) and the quality of the relationship with the other party as important indicators of outcomes: more recent and less severe disputes were more likely to be resolved, as were those between parties who had demonstrated at least a modest degree of co-operation and ongoing communication (Hochberg & Kressel, 1983; Pearson & Thoennes, 1988).

Hayes’ (2002b) work in England has shown that particular background characteristics of mediators create systematic variations in their practice. Myers & Wasoff (2000) explicitly compared the practice of mediators in Scotland from different professional backgrounds and found a number of areas of difference in their practice. For example, lawyer-mediators were more likely than non-lawyer-mediators to offer overt direction, direct advice and direction in respect of settlements during the mediation process. By contrast, non-lawyer-mediators encouraged greater levels of client participation and client ownership of outcomes. The language used also varied: non-lawyers used the language of needs, responsibilities and preferences, while lawyer-mediators used the language of rights. However, like the mediators in Dingwall & Greatbatch’s studies, mediators in Scotland were found to hold strong values with regard to what kinds of arrangements work best for children. Dingwall & Greatbatch (2001) observed that mediators seek to influence outcomes accordingly, thus shifting the process away from one of client empowerment to one of managed settlement-seeking. Undoubtedly, more research is needed if we are to be able to make more accurate associations between processes and outcomes and to predict which cases will do well with which model of mediation.

Outcomes

Research has examined a range of key outcomes which characterise successful mediation: parties reaching agreement; parties being satisfied with the process and outcomes; and parties complying with mediated agreements and avoiding relitigation.

Reaching agreement

In her first review, Kelly (1996) reported that mediation research internationally indicated that couples reached agreement between 50 and 85 per cent of the time. She could find no clear relationship between agreement/settlement rates and the number of mediation sessions or the time spent in mediation. Differences have been found, however, depending on the focus of the mediation. Higher settlement rates have been reported for all issues mediation than for child-focused mediation. In her more recent review, Kelly (2004) reconfirmed that settlement rates generally range between 50 and 90 per cent.

A pilot study of family mediation relating to children’s issues in four courts in New Zealand (Barwick & Gray, 2007) found that full agreement was reached in 59 per cent of 257 completed cases. The Winnipeg pilot indicated full agreement in 62 per cent of mediated cases (McKenzie, 2001), and in the next stage of that programme full agreement was reached in as many as 73 per cent of the cases (McKenzie & Pedersen, 2003). A survey of 250 high-conflict parents who had attended different models of in-court mediation in three sites in England found that relatively brief intervention (up to one hour) with parents who had a history of difficulties and disputes relating to access/contact led to full agreement in 45 per cent of cases (Trinder et al., 2006a). Mediation involving a judge was the least successful model in terms of reaching agreement, thus replicating the findings of the first national study of mediation practice in England & Wales (Ogus et al., 1989). The evidence is consistent in respect of the lower settlement rates which are generally achieved in models which are high in judicial control.

In a pilot family mediation scheme established in Hong Kong between 2000 and 2003, 80 per cent of 933 mediation users reached overall agreement, with agreement being highest (94 %) in respect of child contact (access), child custody (92 %) and financial support for the children (88 %) (Sullivan, 2005). Agreements relating to property were recorded at 84 per cent and spousal support at 82 per cent. These settlement rates are at the highest end of the settlement rate spectrum, and Sullivan suggested that this may be because the parties were influenced by traditional Chinese cultural beliefs and values. The mediation process reflects these values by encouraging attainment of the common good rather than individual gain and valuing co-operative decision making, compromise, the achievement of harmony, and the avoidance of negativity and conflict. Mediation in Hong Kong has been seen as having the potential to help couples craft agreements in a private, confidential setting that honours the parties’ relational obligations to children, to each other and to extended family members. The study provides strong evidence that cultural factors are important indicators of outcomes, and that where harmony and conformity are valued family mediation resonates with these values.

Reaching partial agreements

Given the complexity of some disputes and the emotional climate within which mediation usually takes place, it is hardly surprising that full agreement is not reached by everyone. Moreover, some agreements in mediation represent final settlements and others may offer only temporary or interim solutions. Findings indicate that the majority of couples manage to reach some agreements on some issues during mediation. Partial agreement was reached in 30 per cent of the New Zealand cases (Barwick & Gray, 2007); agreement was reached on some or most issues in 32 per cent of cases in the Winnipeg pilot (McKenzie, 2001), and in the follow-on mediation programme 27 per cent of cases achieved partial agreements (McKenzie & Pedersen, 2003). Studies in the USA confirm findings from other countries (Kelly, 2004). For example, the Colorado study (Thoennes, 2002) found that while 39 per cent of cases reached full agreement, a further 55 per cent reached agreement on some issues. The snapshot studies in California revealed that 55 per cent of couples had reached some agreement (Kelly, 2004).

When full and partial settlement rates are combined, the evidence across a very wide range of studies is that relatively few couples leave mediation without reaching agreement on any issue. Looked at in this way, family mediation has been shown to be remarkably successful in terms of assisting couples to reach agreement. In Winnipeg, just 14 per cent of couples in the pilot phase and only 8 per cent in the follow-on phase did not reach any agreement. In the New Zealand study, 11 per cent of completed mediations did not reach agreement, and in the Hong Kong pilot the number of cases in which no agreement was reached was extremely small. It seems reasonable to conclude that couples are more likely to reach at least partial agreement in mediation when it is not rushed or limited to one relatively brief session in court, and when parties have sufficient time to find some elements of agreement between them. There is little research that has examined the cases which fail to reach any kind of agreement, and it would be useful to be able to determine whether there are any common characteristics among the cases, the mediation approaches and the settings.

Satisfaction with mediation

The number of agreements reached in mediation needs to be considered alongside a number of other measures of effectiveness. Kressel & Pruitt (1989) argued that there had been an overemphasis on settlement rates as an indication of success, particularly as couples who did not reach agreement clearly valued the mediation experience because it achieved other beneficial outcomes. The majority of studies have taken client satisfaction rates as an important indicator, and it can be seen that the failure to reach any agreement in mediation does not necessarily imply that the parties are dissatisfied with the process or that they have not achieved other benefits, at least in the short term.

In her initial review, Kelly (1996) found that, with one exception, all studies in all countries and settings showed that client satisfaction with both the mediation process and outcomes was quite high – within the range of 60–80 per cent. Slaikeu et al. (1985a) suggested that the best indicators of success in mediation were the clients’ perceptions of the mediator’s ability to facilitate communication, allow each party to be heard, identify options for resolution and engender mutual understanding between the parties. The Winnipeg study also revealed high levels of satisfaction: more than 80 per cent were either somewhat or very satisfied with the co-mediation process, the session content, the intake process and the overall mediation experience. Somewhat lower rates of satisfaction were attributed to the mediation location and the brevity of the intervention.

The snapshot study of mediation in the Californian courts (Depner et al., 1994) found widespread satisfaction: parents found it helpful in developing workable agreements for child custody and contact and they were satisfied with the outcomes. The general pattern of positive evaluations was stable across diverse populations, but mediation was rated more helpful by parents with less education and lower incomes, and by ethnic minorities. The type of mediation had no statistically significant effects on general satisfaction. Although levels of satisfaction were undoubtedly higher when couples were able to reach agreement in mediation, even when couples failed to reach any agreements, most felt that mediation has given them the opportunity to address the issues, receive helpful information, clarify the issues in dispute and move forward.

The generally high levels of satisfaction are reflected in the finding that most clients say they would recommend family mediation to someone in similar circumstances. Overall, the body of research suggests that satisfaction levels are higher than settlement rates, irrespective of mediation settings. Clients refer to: being heard and understood; being respected and taken seriously; being given a chance to express their feelings and concerns; receiving information about what matters to children; having a safe forum in which to communicate; having mediators who are impartial and sensitive to difficult and emotive issues; feeling empowered to find a solution; being helped to manage personal affairs; being able to focus on the needs of children; and having an opportunity to air grievances. Dissatisfaction is related to the obverse of the above, and also to: being pressured to reach agreements; experiencing tension-filled and unpleasant mediation sessions; feeling rushed; and being confused about the purpose of mediation.

Concerns about mandatory mediation have been couched in terms of the perceived disadvantages for women, pressurising them into reaching compromises which are not in their interests or are unfair and unhealthy. Yet no evidence exists that women, as a group, tend to fare worse in negotiations as a result of their greater interest in co-operation and maintaining good relationships (Rosenberg, 1992). Nor are women more likely than men to drop out of mediation. Kressel & Pruitt (1989) indicated that when parents were required to mediate (even when they would have preferred not to) between 75 and 85 per cent were satisfied with the process and glad they had been ordered to participate. Indeed, women were significantly more likely than men to say that mediation gave them an opportunity to express their point of view and helped them put aside their anger and focus on their children’s needs (Kelly & Duryee, 1992). Nevertheless, women were more likely than men to drop out of mediation because they lacked sufficient understanding of financial issues and/or were feeling confused. They were also significantly more likely to believe that their spouse had some kind of advantage during negotiations, that the issues were too complex and that they felt emotionally drained and unprotected (Kelly et al., 1988; Kelly & Duryee, 1992). Men in some studies, however, have indicated greater dissatisfaction than women about what they perceive to be a bias in the family justice system (including in mediation) in favour of mothers over issues such as custody and contact (Emery et al., 1991, 1994; Maccoby & Mnookin, 1992; Walker, 1994).

Compliance and Relitigation

One of the key tests of the effectiveness of mediation is the extent to which agreements reached stand the test of time. There is mixed evidence from a range of studies. Pearson & Thoennes found that 79 per cent of clients in the Denver Custody Mediation Project who reached agreements in mediation reported that their spouse was in compliance with the terms of the agreement. This was higher than compliance with court-adjudicated outcomes (67 %). Only 6 per cent of the mediation group had experienced serious disagreements over the settlement in the months following mediation, whereas 33 per cent of the litigation group had done so (Pearson & Thoennes, 1982, 1984a,b). There was evidence, also, of poorer compliance in the litigation group with child-support and access orders, although non-compliance was also problematic in the mediation group. Over longer periods, both groups reported frequent disagreements. Nevertheless, Pearson & Thoennes were of the view that successful mediation clients were more able to work through their disagreements and did not necessarily turn to the courts to resolve them. Although there were inconsistent patterns in the evidence, they concluded that mediated agreements were no less stable than those drawn up through lawyers or ordered by the courts.

The Winnipeg study evidenced better compliance in the mediation group than in the non-mediation group, and this appeared to have a positive impact on parenting behaviour (McKenzie, 2001). The follow-up study indicated that levels of co-parental interaction had increased and that satisfaction with the arrangements remained high and there had been a lower rate of court filings in the mediation group (McKenzie & Pedersen, 2003).

The study of in-court mediation in England (Trinder et al., 2006a) found that agreements were still intact six months later, irrespective of the mediation model. The researchers found that two factors were significant predictors of durability: not being eligible for legal aid/state funding, and having reached full rather than partial settlements in mediation. The ‘easiest’ cases were the most likely to remain intact. The very low number of cases returning to court was viewed as a positive outcome from a complex sample. Of course, the non-use of the courts does not necessarily indicate that agreements are being complied with, and the research data in this area are not robust enough to enable us to be certain that all mediated agreements are durable. Family relationships change considerably after separation and divorce and transitions are often complex and challenging. Agreements made during mediation may no longer be appropriate. For all but a hard core of parents who repeatedly litigate over custody and access/contact, the passage of time tends to reduce the problems associated with divorce so that it is more difficult to discern differences between mediation and litigation groups. Nevertheless, re-litigation tends to be higher in non-mediation samples than among mediation clients.

There is evidence, particularly in the USA and Canada, that settlement/outcome patterns vary between mediation and litigation groups, which may explain the low levels of re-litigation in the mediation groups. In mediation, parents were: more likely to agree to joint legal custody than those using the courts (Emery, 1994; Pearson, 1991; Richardson, 1988) able to develop parenting plans that provided more contact time for non-custodial parents (Kelly, 1993) and able to agree about decision-making in respect of children (Kelly, 2004). Given that the chief aim of family mediation is to resolve disputes, settlement rates and the durability of settlements are the obvious measures of effectiveness. An understanding of the factors which account for ‘success’ or ‘failure’ in terms of reaching agreements is still somewhat confused. Outcome success seems to be associated with a clear focus in mediation, a process which is allowed to unfold gradually, and the support of partisan lawyers advising the clients.

The wider benefits of mediation

Satisfaction with the mediation process is generally high and the reasons given for this suggest that it is the process which allows the achievement of wider benefits which really makes a difference. Over thirty years ago, Margolin (1973) found that mediation clients reported greater child satisfaction, greater enjoyment on the part of parents and children, and more appropriate child behaviour than litigation parents. Nevertheless, there is little evidence from the early research that changes in children’s adjustment to separation and divorce or improved psychological functioning can be attributed to mediation (Emery, 1994; Kelly, 1990; Walker et al., 1994). These researchers agreed that claims that mediation would improve psychological adjustment were quite unrealistic.

Achieving reductions in conflict and improving communication between the parties are considered by mediators to be important outcomes. The Winnipeg pilot found that mediation had reduced conflict around custody and time-sharing for 59 per cent of the participants. There was a consistent and statistically significant decline in conflict at follow-up, including a significant decline in the conflicts between parents who had tended to put children ‘in the middle’, and mediation clients showed a more positive post-separation relationship pertaining to communication and problem-solving activities. Significant improvements were evident in terms of child-coping and the achievement of full agreement in mediation appeared to result in more positive outcomes. While the passage of time is acknowledged as an important factor, the Winnipeg study provides clear evidence that mediation contributed to the positive changes identified. McKenzie concluded that comprehensive mediation can help to support good-quality post-separation parenting and that, in those cases where it reduces conflict, the process of separation should be less damaging to children.

The wider benefits of mediation have been widely welcomed, and they provided additional evidence for those advocating that mediation is a better way of resolving family disputes. Nevertheless, in some jurisdictions, as mediation has become increasingly integrated with legal processes, it has been increasingly regulated and appears to have narrowed its remit. In England, the introduction of public funding for mediation resulted in fewer agreements and increased dissatisfaction because agreements were unenforceable; agreements had been made under pressure; there was frustration with the way mediators had handled the process; and complaints that mediators did not provide any advice when complex issues surfaced during mediation (Walker, 2001). The findings from this study indicated that 60 per cent of mediation users felt they had not been helped to improve communication; 60 per cent thought mediation had not helped parties to share the decision-making; 58 per cent believed that mediation had not helped to reduce conflict; 58 per cent said that mediation had not helped to make divorce less distressing; and 60 per cent believed that mediation had not avoided them going to court. While those who had reached agreement were significantly more likely to feel that they had benefited in other ways, 38 per cent nevertheless disagreed with the proposition that mediation had helped to improve communication, and 38 per cent said it had not reduced conflict. The emphasis in mediation appeared to shift from promoting the multiple benefits and more therapeutic goals associated with mediation to the reaching of agreements as the sole mark of success. Davis (2001) argued that mediation had become a kind of hybrid service – a settlement device within legal proceedings – in which mediators were required to achieve outcomes according to standards applied to lawyers.

If the ability ‘to sort out troubles’ has diminished in the quest for greater uniformity of practice, it can be argued that embedding mediation within the legal system is in danger of compromising the aspects of mediation which most practitioners and clients have appeared to value. Mosten (2004) noted that the trade-off between regulations, creativity, accountability and quality control had generated a vigorous dialogue with the USA mediation community, and coercive pressure on mediators in the USA to settle cases.

The evidence from randomised control trials and longitudinal studies

Randomised control trials (RCTs) constitute the gold standard in research, enabling us to observe the impact of a specific intervention with a far greater degree of confidence. An important study conducted by Robert Emery in Virginia in the USA adopted the gold standard using a mediation group and a control group, tracking outcomes over time. The research relating to these two cohorts over a twelve-year period provides a wealth of information about the relative effectiveness of family mediation when compared with traditional litigation, and about systemic patterns of interpersonal influence and relational dyna-mics between divorced parents.

Emery and his colleagues began their study in the 1980s in order to examine whether mediation produced more desirable outcomes than did litigation (Emery et al., 1991; Emery & Wyer 1987). The research involved 71 families who had requested a child custody hearing at a court in Central Virginia between 1983 and 1986. Of these 71 families who were in dispute about the custody of their children, 35 were randomly assigned to divorce mediation and 36 to the traditional litigation route through the court. The sample represents a relatively conflicted group of divorced parents. Research assessments took place about five weeks after dispute resolution (time 1), and again some eighteen months later (time 2). Long-term follow-up assessments were undertaken twelve years after the initial custody decision (time 3). The original sample consisted of 63 mothers and 59 fathers and the twelve-year follow-up included 50 of the mothers and 43 of the fathers. There were no significant differences between the mediation and litigation groups in respect of their background characteristics. At time 3 assessment, around half of the participants were either remarried or cohabiting.

Mediation was conducted on court premises by four male–female pairs of co-mediators and was limited to a maximum of six two-hour sessions (with an average of 2.4 sessions) (Emery, 1994). The mediation process was relatively short and the intervention contained elements of both problem-solving and therapeutic mediation (Emery et al., 2005). A number of measures were used in the research to provide objective measures of change (Shaw & Emery, 1987). The key findings at time 1 and time 2 (Emery et al., 1994; Emery et al., 2001; Kitzmann & Emery, 1994) indicated that most of the mediation group reached agreement in mediation; mediated agreements were more likely to specify joint legal custody; fathers reported much more satisfaction with mediation than with litigation; mothers were more satisfied than fathers with both mediation and litigation outcomes; the level of litigation in the two years following the settlements was lower among the mediation group; immediately after dispute resolution and at time 2, the quality of family relationships and the psychological adjustment of parents were the same in both groups, except that fathers in the mediation group reported less co-parenting conflict at time 2; and decreases in parental conflict between time 1 and time 2 predicted better psychological adjustment for children, but no differences were found between the mediation and litigation groups at time 2.

The Virginia study is unique as regards the extent to which the original groups of families have been tracked over time. The time 3 assessment allowed them to examine arrangements for children, parenting behaviours, parental satisfaction with dispute resolution, the quality of parenting, and children’s psychological adjustment. The key findings at time 3 are extremely important because they indicate long-term positive benefits. Contact with the non-resident parent (usually fathers) was higher in the mediation group than in the litigation group, where it was close to the national average. Non-resident parents in the mediation group saw their children and spoke to them significantly more often than those in the litigation group (Emery et al., 2001). Moreover, contact between the parents was also higher, and non-resident parents in the mediation group were notably more involved in their children’s upbringing and reported lower levels of co-parenting conflict. There is clear evidence that non-resident parents in the mediation group maintained higher levels of contact and involvement with their children than non-resident parents in the litigation group, and this was substantiated in the accounts of resident parents. Mediation parents were more flexible in respect of the living arrangements for their children and changes reflecting children’s developmental needs and wishes tended to be managed informally by the parents.

The twelve-year follow-up has provided significant insights into the relationship between co-parenting conflict and the ability of each party to accept the divorce and adjust to a new life. For fathers there was a strong association between their own conflict and their ex-partner’s acceptance of the divorce. Fathers were generally less accepting of the end of the marriage, perhaps because of the perceived losses it engenders. Nevertheless, mediation parents reported decreased conflict at time 2 and litigation parents reported increased conflict (Sbarra & Emery, 2008). Importantly, parents who reported the greatest short-term decreases in conflict also reported the greatest long-term decreases in conflict. Litigation parents who reported increases in conflict at time 2 evidenced the greatest increases in long-term conflict at time 3. Sbarra & Emery (2008) concluded that mediation helped to reduce co-parenting conflict and reduced the likelihood that parents would enter adversarial legal proceedings that could inflame conflict.

The findings from this study are substantive. As the researchers have properly pointed out, however, the selective attrition in the time 3 sample limits the ability to detect longitudinal effects fully and, during the decade between time 2 and time 3, attitudes to post-divorce parenting have shifted and the role of fathers has been acknowledged in more equitable outcomes in settlement decisions (Beck & Sales, 2001; Walker et al., 2004). The study shows that mediation can have significant benefits, but it does not allow us to conclude that family mediation will always produce these benefits. Mediation is unlikely to be able to account for all the effects which are manifest over a twelve-year period, but the Virginia study offers support for a model of mediation which is not restricted to a single session nor to making agreements under pressure. It is worth noting that satisfaction with both mediation and litigation declined over time, especially among fathers, indicating that the realities of post-divorce parenting are such that any dispute resolution process is unable to counteract the challenges, transitions and losses completely.

A large body of research indicates unequivocally that conflict between parents is associated with an increased risk of psychological problems for children (Amato & Keith, 1991; Emery, 1982; Pryor & Rodgers, 2001). The Virginia study has shown that conflict was reduced in the mediation group, and an average of five hours of mediation seems to have resulted in increased and sustained contact between non-resident parents and their children. However, the study did not find any group differences in the mental health of parents or children at time 3, despite other very positive outcomes. The researchers have suggested that mediation might make an impact on mental health outcomes for parents and for children if it is packaged with other effective interventions that specifically target psychological outcomes such as parenting education or children’s groups. The evidence from the RCT study suggests that the mediation process had: promoted parental co-operation; educated parents about emotions – their children’s and their own; encouraged parents to develop businesslike boundaries around an ongoing co-parenting relationship; and helped parents to avoid becoming adversaries.

Looking to the future

The evidence from a range of studies is largely supportive of claims made for mediation and the themes which have emerged are surprisingly consistent. The behaviour of the mediators and the willingness of the parties to compromise are critical variables in shaping the mediation process and the kinds of outcomes achieved. Settlements are more likely if the disputes are less intense, if the disputants are motivated to settle, if the mediator is more experienced and is active during mediation, and if disputants participate in a co-operative manner (Wissler, 2006). Nevertheless, it is ‘difficult to discern which particular characteristic plays a critical role in the mediation process or outcome’ (Wissler, 2006, p. 140). It is important not only to examine what mediators do in mediation but also the timing, frequency and skilfulness of what they do (Kressel et al., 1994; Pruitt, 1995). Research on this aspect of practice is very limited. This may be particularly relevant as mediators look to develop new skills and directions.

Robinson (2008) has argued that creative options are now needed for dealing with an ever greater variety of family conflicts and for working in partnership with colleagues from different professional backgrounds, and has expressed the view that the most effective mediation services in the future will be those that can offer a range of flexible models, in addition to other alternative dispute resolution services, from collaborative law to contact assessments. His vision includes a process of assessment to determine the most appropriate ‘bespoke package’ for each case – fitting the process to the parties and drawing on a wide variety of mediation models and techniques. In this sense, family mediation would be engaged in extending its boundaries without compromising its underpinning values. A number of specific issues need to be considered as mediation practice moves forward, including the role of children in mediation. New developments in Australia and New Zealand are providing an opportunity for mediators to revisit this thorny issue and evidence has been emerging about innovative ways forward for involving children (Cashmore & Parkinson, 2008; Goldson, 2006).

Assessing the evidence

This review of family mediation research reminds us that ‘permitting a thousand flowers to bloom has been mediation’s history’ (Mosten, 2004). The proliferation of models, styles and approaches makes it difficult to be conclusive about the evidence available and we need to be cautious about generalising from research that, inevitably, has looked at many different models in the quest to understand what works, for whom and in what circumstances. A number of dilemmas are unresolved. Should mediation be mandatory? Should children be involved? Should cases in which there has been domestic violence be automatically screened out of mediation? When is it appropriate to mediate in high-conflict cases? More important, perhaps, for community-based mediation services are questions about how family mediation should be positioned with respect to developments in parenting education programmes, collaborative law and multi-disciplinary practices of litigators, counsellors and mediators. Is there a need to develop new, hybrid, flexible models which allow for greater personalisation of services and tailored packages of intervention? Mediation is not an option chosen by a large percentage of the separating/divorcing population in most jurisdictions, but the research indicates that it can be very effective in helping parties reach agreement and undertake more co-operative parenting. The additional contribution which can be made by other supportive interventions needs to be considered also.

Conclusions about the benefits of mediation are not always consistent across studies which have evaluated different mediation approaches within different family law contexts and with different client populations. It is important to remember that not only the nature of the mediation intervention but also the clients’ exposure to other aspects of the legal system and their personal histories and characteristics will impact on outcomes. Overall, mediation is expected to achieve a good deal – far more than is expected of the courts – and there is a substantial body of evidence that it meets these expectations for very many of those who use it.