“Does training in Motivational Interviewing affect the ability to build working alliance? – an intervention study”

ABSTRACT Higher education in health science and social work are examples of professional educations were it is vital to continue to learn and develop professional knowledge. This study focused on a specific education program which is of relevance to such professionals. Professional studies in health and social work demand insight into and knowledge of disciplines that embrace physical and mental health, as well as social issues. The aim of this study was to find out whether training in Motivational Interviewing contributed to the development of the ability to build a working alliance. N = 72 students within health and social work were surveyed with the Working Alliance Inventory, a questionnaire on thoughts and feelings in relation to clients (Horvath & Greenburg, 1989). They were surveyed (in 2014 and 2015) before and after a 1-year course in MI, and 73.6% (53/72) responded at both measurement points. Analyses were conducted both on each item (Wilcoxon matched-pairs signed ranks test) and on the three sub-scales: goals, tasks and bonds (paired samples t-test). The study showed no significant association between training in MI and increased the ability to build working alliances; however, an association between training in MI and decreased scoring on the sub-scale, goals, was found.


Introduction
Higher education in health science and social work are examples of professional educations where it is vital to continue to learn and develop professional knowledge. However, the basis for professional knowledge varies among different education programmes, higher educational institutions and fields of knowledge. This study focused on an education program of relevance especially to professionals within health and social work; learning Motivational Interviewing (MI), in this article addressed as the intervention. The focus was on whether training in MI contributed to the development of the ability to build a working alliance, which is an essential skill in health care and social work.
In Norway, further education programmes are expected to focus on the social benefits and relevance of education and research. Recent years have seen a growing shift in emphasis from one-way dissemination of knowledge to dialogue between society and the public, as well as to education and development of students as active citizens (Willumsen & Studsrød, 2010). Professional health and social workers are required to have insight into and knowledge of disciplines that embrace physical and mental health, as well as social issues.
In Norway, these disciplines are undergoing constant change, in terms of legislation, organization and allocation of responsibility as well as theoretical and empirical knowledge (Bjaarstad, Kristin, Randi, Hatling, & Reinertsen, 2014. Borg, Karlsson, & Stenhammer, 2013Borg, Sommer, Strand, & Ness, 2013;Levin, 2004;Ministry of Health and Care Services, 2009;The Norwegian Directorate of Health, 2017). This means that educational institutions must be capable of providing knowledge and skills relevant to the current challenges in health and social workwhat needs to be learned. At the same time, the professions must develop competence in handling change and development in the decades ahead; regarding how learning takes place. It is therefore important to gain insight into whether educational programmes fulfil this social responsibility.
A clear characteristic of professions is that they build on a specialized education linked with a knowledge base (Freidson, 2013;Molander & Terum, 2008). A focus on the practice of tasks in a variety of professions has increased in recent years, especially in relation to changes in the rights and role of the service recipient. This is reflected in both theory (Bjaarstad et al., 2014; and legislation such as the Law on patient and user rights (Law 1999-07-02.63). In addition, a change in attitude has taken place concerning the relationship and power structure between the service provider and the service recipient (whom we hereafter refer to as the "therapist" and the "client"). The focus should be on the client's understanding, needs and goals, which means a more balanced relationship between the client and the therapist. This also means that communication shifts from being characterized by advice and instructions from an expert in a paternalistic tradition, to a dialogue between two equal parties. This changes the requirements for knowledge and competence correspondingly.
The aim of this study was to find out whether learning MI contributed to the development of an essential skill in health care and social work; the ability to build a working alliance. To find out about this, a number of students that attended the MI education in 2014 and 2015 were surveyed with the Working Alliance Inventory, a questionnaire on thoughts and feelings in relation to clients (Horvath & Greenberg, 1989).

Working alliance
The concept of working alliance has been found to have a major impact on the quality of the relationship and dialogue between the client and the therapist (Cook, Heather, McCambridge, & Petry, 2015;Hoffart, Borge, Sexton, & Clark, 2009;Horvath, Del Re, Flückiger, Symonds, & Hilsenroth, 2011;Norcross, Wampold, & Hilsenroth, 2011). Ralph R. Greenson (Greenson, 2008) launched the working alliance as a concept and according to Horvath & Greenberg; it was strongly influenced by Carl Rogers' person-centred therapy (Horvath & Greenberg, 1994). Rogers did affirm the subjective personal experience of the patient as the basis and standard for living and therapeutic effect (Rogers, 1951(Rogers, , p. 1952). Rogers identified six conditions that are needed to produce personality changes in clients (Rogers, 1957). Edward S, Bordins' division of the working alliance into three dimensions has played a key role in the later interpretation of the alliance concept (Bordin, 1979;Horvath & Greenberg, 1994). Bordin maintains that a working alliance consists of (1) goals; that is, what the therapist and client agree on as the objective of the treatment, (2) tasks; a common understanding and agreement on the individual's tasks for achieving the goal and (3) bonds; the therapeutic and personal link that develops between therapist and client during therapy (Bordin in Horvath & Greenberg, 1994). It is worth mentioning that Bordin differs with the clientcentred method, as he emphasizes an explicit negotiation of goals and tasks (Bordin, 1994). There seems to be broad support to the correlation between working alliance and client outcome, more so early on in the therapy (Horvath & Greenberg, 1994;Wampold & Imel, 2015). Working alliance is regarded as a general, or common, factor regardless of the type of intervention or therapy (McCarthy & Barber, 2009;Moyers, Miller, Hendrickson, & La Greca, 2005;Wampold & Imel, 2015). On the other hand, the working alliance seems to be strongly influenced by qualities related to the therapist (Del Re, Flückiger, Horvath, Symonds, & Wampold, 2012). Clients also seem to attribute the main responsibility for the quality of the working alliance to the therapist (Bedi & Mallinckrodt, 2006). Therapists in mental health are expected to acquire the knowledge and skills they need to develop working alliances with their users; however, studies show that this is not always the case (Moyle, 2003). Moyers et al. (2005) claim that little attention has been paid to attributes or techniques of the therapist that may strengthen or damage the working alliance.
Given the important role of the working alliance in health care and social work regardless of the practical or theoretical framework, it is important that the concept should be addressed in the education of professionals in social work and health care. Such education should contribute to students' ability and skills in establishing and nurturing the working alliance (Del Re et al., 2012;Norcross et al., 2011).

Motivational interviewing
An example of a further education programme primarily intended for health and social work professionals is training in MI as a counselling method. Motivational Interviewing was first introduced by William R. Miller (Miller, 1983) and has since then been further developed and described by Miller, particularly in collaboration with Stephen Rollnick (Miller & Rollnick, 1991, 2002, 2013. MI is recommended as a method in several specific disciplines in the health and social sector, with welldocumented benefits (Hettema, Steele, & Miller, 2005;Lundahl, Kunz, Brownell, Tollefson, & Burke, 2010;Smedslund et al., 2011). Some suggest that MI-related therapist behaviours have a positive effect on working alliances (Boardman, Catley, Grobe, Little, & Ahluwalia, 2006). One study showed a significant association between students attending further education in MI and perceived improvement in one's own alliance building capability (Berge & Breiseth, 2015). Miller & Rollnick acknowledge working alliance as a key aspect of relational work and collaboration (2013). However, they do not provide a more detailed description of what a working alliance consists of (Miller & Moyers, 2006;Miller & Rollnick, 2013). Moyers et al. (2005) maintained that the emphasis of MI on specific relational qualities is consistent with empirical evidence linking working alliances to improved therapeutic outcomes in more general therapeutic literature. MI is a counselling method that focuses on change and motivation (Miller & Rollnick, 2013), and like the concept of working alliance, MI theory has roots in Carl Rogers' patient-centred theory (Barth, Børtveit, & Prescott, 2013;Rogers, 1951). Key elements in the spirit of MI are a partnership on an equal footing, acceptance, compassion and evocation (Miller & Rollnick, 2013). The focus is always on evoking and exploring the client's perspective, in terms of collaboration, the goal and how to achieve the goal; which are the tasks in Bordins' terminology. Other central concepts in MI are equality, mutual trust, hope, empathy, autonomy and affirmation (Barth et al., 2013). When working with elaborating the client's perspective using MI, it is essential to ask the client about his or her understanding of both the problem or situation, the goal and how to get there. When using MI, the therapist does not diagnose the problem, the goal or the tasks. The therapist wants to understand the clients' perspective when working together with the client to find alternative perspectives and ways of dealing with the issues. The clients' autonomy will be crucial in MI; the right of clients to make his or her own choices regardless of what the therapist may find wise or not (Miller & Rollnick, 2013). Using MI, the therapist does not tell the client what to do or not, according to the recommendations of Miller and Rollnick (2013), as this will decrease the risk of resistant towards change within the client. As mentioned, MI is documented to benefit client's motivation and willingness to change, and the emphasis is on the client's own goal and task as an essential element (Hettema et al., 2005;Lundahl et al., 2010;Smedslund et al., 2011). Miller and Rollnick (2013) also suggest that one important part in MI is "a decrease in unhelpful counsellor responses" (2013, p. 381), such as telling people what to do, confronting and directing. However, it is still a lack of knowledge when it comes to the impact MI has on the therapist's assessment of own ability to create a good working alliance. Thus, it is interesting to examine whether this education program helps to strengthen students' skills in developing a working alliance.

Research questions
The research question explored in this article is: "Do students report higher self-perceived working alliance with their clients, after learning MI?" This article focuses on the effect of learning and training in MI on students' perceptions of their own ability to build working alliances with their clients.

The intervention
The intervention, getting knowledge about and learning how to practice the MI method, was divided into five intensive learning gatherings. These gatherings took place over a calendar year. To qualify as a participant, one would need to have a bachelor's degree in a relevant topic and also working with people especially focusing on motivation and change. Each gathering lasts for two or three days. They are all led by the same (two) teachers, except for one gathering when professionals share their experiences from practising MI.
The overall aim of the intervention was for participants to acquire and develop attitudes, skills and strategies essential in MI. Hence, the participants are expected to get insights and knowledge about research and theory in MI, the ethical foundation on which the method is based, and where and when it is appropriate to use MI. The participants were introduced to the MI method, and during the intervention also trained in how to practice MI on how to adjust the method to specific tasks. During the intervention the participants got lectures, they got demonstrations of MI and supervision about their achievements of MI qualification. This happened through critical reflections and dialogue, role play and real play. All participants were required to practice MI in between the gatherings with their clients. These client conversations where transcribed and the participants received supervision from trained MIpractitioners in groups.
In this type of further education programme, the aim of the learning activities is that, combined with the pedagogy, they will contribute to a transformation from knowledge to skills. The intervention follows the progress as illustrated in Figure 1.
During the intervention, the participants are expected to achieve insight, profound understanding and become more reflective about their interaction with clients when they use an MI approach. Particular emphasis was placed on a client-focused position, where the key is to explore the client's thoughts, values, change goals and how to achieve these goals. This is precisely what MI is about; being able to understand the client's perspective; how does she/he perceive the situation, what goals, if any, do the client want to achieve and how does the client see him/herself achieving these goals. These aspects are also essential for working alliance, according to Bordin (1979) an ability to build bonds, to identify common goals and reach agreement on tasks aimed at achieving the goal.
Although the intervention did not focus on working alliance as such, the overlap of essential aspects was expected to increase the participants' ability to establish a working alliance after learning MI. The questionnaire in use focuses on the therapist and client finding a "common ground" where agreement upon goals and tasks and trust and appreciation of each other are essential. MI does not focus as much on mutual "good feelings" or the therapist agreeing with the clients' choices. Instead, the method tries to take a more unbiased standpoint, helping the clients to find his/ her own way, enhancing autonomy and supporting choices. Given these differences, the questionnaire will not provide answers to whether the programme actually does teach the students how to conduct MI. Hence, the questionnaire is not customized to evaluate the goals of the programme as such, even if it provides knowledge relevant for the goals of the programme.

The participants
In 2014 and 2015 all students in one MI education were invited to participate in this study. Students that were invited, are expected to attend 80% of the learning activities of the course, according to the education plan of this education (Nord University, 2018). The study does not provide much information about the participants apart from gender and work experience. The majority of participants in the survey were women (79%) from the age of 25 to 60. The average number of years of work experience was 13 years as a health and social care worker (range from zero to 35). The students that participated had an educational and professional background from health and social service, such as nurses, social workers, physical therapists, occupational therapists. A minority of the students had a professional background of being a teacher, police or manager.

The questionnaire
The questionnaire used was the Working Alliance Inventory (Horvath & Greenberg, 1989). According to Martin, Garske, and Davis (2000), several scales have been developed over the years to measure working alliance. Penn scale, Vanderbilt scale, Working Alliance Inventory and California Psychotherapy Alliance Scale has shown the highest relation to outcome and Martin et al. (2000) suggest that the WAI is likely to be appropriate for most research projects. Boardman et al. (2006) used WAI in their study and found that MI-related therapist behaviours have a positive effect on working alliances. All in all, we found it most expedient to use WAI. A short version (WAI-S) developed especially for therapists by Tracey and Kokotovic (1989) was used. The form is based on Bordin's division of the working alliance into three dimensions: goals, tasks and bonds (Bordin, 1979) and was translated into Norwegian by Svartberg & Saxton in 1994. Horvath claims that there is some unclearness regarding the distinctiveness of the subscales of goal, task and bond (Horvarth & Greenberg, 1994). At the same time, there seems to be some evidence of the correlation between the measures and some preliminary support to the structure in one-factor analytic study (ibid). As the original scale remains the same even though several different versions have been developed, and because Berge and Breiseth (2015) used WAI-s in their study, we chose to use the WAI-s. The aims and learning outcomes as presented in the programme description deviate from the objectives and indicants of the questionnaire (Nord University, 2018). While WAI-s is very specific when it comes to indicants, the programme descriptions are more at an overall level, describing shortly the essence of MI as well as an expected learning outcome. The learning outcome includes knowledge, skills and general competence.

Data
Data were collected through self-administered questionnaires given to the students prior to, and after,  completing the course in MI (2014MI ( , 2015. The questionnaire consists of the  Horvath & Greenberg, 1989).
Questions 4, 6, 10 and 11 correspond to the category that Bordin describes as agreement on goals, 1, 2, 8 and 12 to the category assignment on tasks and 3, 5, 7 and 9 development of bonds (Bordin, 1979).
All the 72 students at the two different years of education were invited to participate, and 58 chose to fill in the questionnaire at baseline. Five students did not return the questionnaire after completing the education; hence, the analysed sample consists of 53 students (73.6%) with valid measurements at both time points. Respondents were asked to assess statements on a 7-category Likert scale (never, seldom, occasionally, a few times, often, very often and always). All questions, apart from two, had the same direction where a higher score indicated higher agreement. We reverse coded the two questions for better interpretability and ease of communication.
Analysis of responses on individual questions before the course (baseline) and after completing the course (follow-up) was done with Wilcoxon matched-pairs signed ranks test. We also constructed three scales based on individual questions to identify latent variables (common goals, tasks, and relational bonds) and assessed the scale reliability by Cronbach's alpha. A Cronbach's alpha >0.7 is generally considered reliable. The three scales were analysed with paired samples t-tests. All analyses were conducted in Stata 15 with p < 0.05 as alpha level.

Ethics
The Norwegian Social Science Data Services (NSD), project no 41710, May 2015 approved the study and its data collection. The regional committee for medical and health research ethics (REC) was consulted and concluded that the study was not subject to submission for approval by REK (2014/1820).

Results
Descriptive statistics from the 12 questions are depicted in Table 1. Except for the question relating to doubts, the median score was 5 or higher on the baseline. This pattern was fairly similar at the second measurement, and the only significant differences between measurement 1 and 2 on individual items were appreciation as a person and different ideas. One of them, appreciation as a person, suggested a decrease whereas different ideas suggested an increase from 4.58 to 4.83. Table 1: Descriptives on single items for the questionnaires on baseline and follow-up (median, min, max, range and mean) and p-values for differences between baseline and follow-up. Table 2 shows scale reliability by Cronbach's alpha for the three latent variables; common goals, tasks, and relational bonds. Only "tasks" showed reasonable reliability on the first measurement (Cronbach's alpha = 0.76) whereas all three were reasonable at the second measurement. None of the constructed indexes were statistically different between baseline and follow-up. Table 2: Mean (SD) on baseline and follow-up for three indexes (Goal, Task and Bond), reliability (Cronbach's Alpha) on baseline and follow-up, and p-value for difference between baseline and follow-up.
We further tested differences on the indexes according to background characteristics (long versus short work experience and also according to gender) and found that females scored higher on follow-up on the Goal index (p < 0.05) while the rest were not statistically significant. A linear mixed model with a three-way interaction between time*gender*experience was also specified. We found no statistical evidence for differential effects for any subgroups.

Discussion
The aim of this study was to investigate whether higher education in MI as a counselling method helps to change students' self-perceived ability to build alliances with their clients.
Although Table 1 shows a slight change from measurement 1 to measurement 2 on most questions, the change is significant only for two questions: therapist appreciation client as a person and having different ideas about what the problem really is (questions 7 and 10).
The table shows a decrease in the mean on question 7 from 5.7 to 5.3. Question 7 deals with appreciating the individual as a person and forms part of Bordin's bond. MI emphasizes a neutral position in the meeting with the client; liking the client is not an end in itself. One reason for the focus on a neutral position is a desire to reduce the risk that the relationship and cooperation might be affected by positive or negative feelings related to the client. This attitude differs from what is emphasized in theories about working alliances, perhaps from other therapeutic approaches as well, where the therapist's feelings towards the client are perceived as meaningful for the relationship. Such a change in the perception of the therapist's feelings and thoughts in relation to the client might influence attitudes to the appreciation of the client as a person. Question 10, different ideas, is a factor in the "goal" category. Here we find a significant increase from the first to the second measurement. Students thus perceive that after completing the training they less often have a different view of the client's problem than before the MI education programme. This change may be related to MI's stronger focus on the client's perspective than on the therapist's perspective, which might have influenced students' acceptance of the client's definition of the problem. The therapist is responsible for helping the client towards a conscious clarification of his/her understanding of the problem but is not responsible for arriving at the "correct" formulation of the problem. Greater emphasis on the client's perspective might explain the lower level of differences related to reaching an understanding of the client's problems.
No significant changes were found for the other questions, and as Table 2 shows, significant changes were found only for the "goal" category, not for bonds or tasks. At the same time, Table 2 shows higher internal reliability within the various categories after the intervention. It thus appears that the training in MI has helped to improve internal consistency, which may be attributable to students' increased awareness of relevant factors in the working alliance.
Another relevant aspect may be that the WAI-S was developed during the 1980s and has not been changed since then. In contrast, attitudes related to the therapist-client relationship and their respective roles have been in development since the questionnaire was created. Client participation, empowerment and recovery are examples of both theoretical foundation and practice in which clients' right to selfdetermination has become ever stronger. Perhaps the questionnaire does not adequately reflect this development.
Change processes take time (Prochaska & Di Clemente, 1982). To abandon previous experience implies that it is not good enough, which is a painful and time-consuming process (Mezirow, 1990). The time needed for a change in practice and in the ability to build alliances to become noticeable may be longer than the training period.
Another explanation for the lack of change might be that, at the start of the programme, students may have had an unrealistically high rating of their own ability to build work alliances. The form was distributed in connection with a training programme by one of the lecturers involved in the programme, which might have led students to want to give a good impression toward the teacher. The students were in a setting in which it might have been regarded as desirable to be seen as skilled practitioners. The training focused on several aspects relevant to working alliances might enable students to develop a more conscious attitude towards their own skills and thus to adjust the score somewhat in the second measurement. The period of 10 months between the measurements, without students being able to see their own initial measurement, may also have influenced the results. A weakness in the study is that students were requested to fill in a questionnaire based on selfreported ability, not related to a specific client or a specific perception or experience. The period that elapses between the dialogue with the client and filling in of the questionnaire might lead to recall bias. These factors might be relevant in connection with both the first and the second measurement and, depending on the interval from the last dialogue with the client to completion of the questionnaire, might affect the scores measured. However, the greatest weakness of this study is the number of participants, N = 53. The number of participants is limited to students in a specific programme of study so that it will only be possible to increase N by repeating the study several times. But this was not possible for this study.
Another limitation is that we have no control over how many times the participants actually attended the gatherings, other than at least 80% of the time in supervision. The supervision gives some support to the fact that the students are increasingly practising MI.
In a study by Sletterød and Haugan (2009) almost 80% of the responding students in the first class of this program took the initiative themselves or were encouraged by their leader to apply for the program. Almost 20% reported that they learned about education through co-workers (Sletterød & Haugan, 2009). In this study, the responding students had high or very high expectations regarding improvement of the quality of their own work with their patients (Sletterød & Haugan, 2009). This might suggest that the students are a highly motivated group, but we do not know if this motivation has aroused from reading the information provided on the website or from recommendations about the method. The high degree of motivation could have some influence on the students learning efforts and willingness to learn and change.

Conclusion
This study does not support the findings of Berge and Breiseth (2015), Berge & Breiseth (2018) or the findings of Boardman et al. (2006) that MI-related behaviours have a positive influence on the ability to build a working alliance. There is little knowledge about how or whether skills in forming a working alliance can be learned and developed. At the same time, both working alliances and MI have well-documented positive effects on activities in health and social care. Some studies indicate that the practice of MI has a positive influence on the working alliance and that training in MI contributes to a self-perceived improvement in working alliance skills. The impact of working alliance on therapeutic work is also well documented. Due to the limitations in this study, the low N and a possible risk that the students reported an unrealistic high quality, we find it difficult to conclude that learning MI has no impact on the students' ability to build a good working alliance. Given the emphasis on both working alliance and MI, we would recommend further studies of potential associations between training in MI and the development of competence in alliance building. Further studies could also include clients' assessments of the therapist's skill in the working alliance.