The main aim of the present study was to examine whether adding motivational strategies to a standard BIP increases participant-facilitator working alliance and participants’ protherapeutic behaviors. To this end, a randomized controlled trial was conducted, comparing a standard BIP (control group) with an experimental condition in which an individualized motivational plan was added to the standard BIP as a motivational strategy (IMP; Lila et al., 2018; Romero-Martínez et al., 2019). The IMP is based on several approaches aimed at increasing intervention compliance in BIPs that have proved their effectiveness in different settings: motivational interviewing (Miller & Rollnick, 2002), stages of change approach (Prochaska & DiClemente, 1982; Prochaska et al., 1992), solution-focused brief therapy (De Shazer & Berg, 1997), and the Good Lives model (Langlands et al., 2009; Ward, 2002). The main components of the IMP are: (1) five individual motivational interviews, of which three were conducted during the evaluation phase to reduce participants’ resistance to the intervention and establish participants’ personal goals, one was conducted in the middle of the program to monitor their progress, and the last one was conducted around the end of the program to assess their achievements; (2) three group sessions, where participants share their personal goals, explain their progress to the group, and receive feedback, support, and advice from facilitators and other group participants (these group sessions are conducted at the beginning, middle, and end of the program); (3) facilitators follow-up and reinforcement of participants’ goals in every weekly group session throughout the intervention; and (4) retention techniques, such as phone calls when participants miss a group session. The IMP also implies that facilitators adopt an empathic and motivational attitude throughout the intervention, creating a climate of acceptance and using confrontation only when it becomes absolutely necessary.

As far as we know, the only previous experimental study analyzing whether motivational strategies increase working alliance and protherapeutic behaviors in BIP participants was conducted by Musser et al. (2008). In their study, groups receiving two motivational interviewing sessions as a pretreatment intervention were compared to control groups (i.e., intake as usual, without motivational interviewing). These researchers observed in the experimental group an increase in protherapeutic behaviors (i.e., verbalizations of responsibility for abusive behavior and group intervention usefulness) and higher working alliance late in intervention rated by therapists.

The present study pulls ahead previous research in two relevant issues, one related to the implementation of motivational strategies and the other related to the assessment of both protherapeutic behaviors and working alliance. First, in our study the motivational strategy was delivered throughout the BIP and not only as pretreatment intervention. We expected that the higher the exposure to motivational strategies the more positive the effect on the working alliance and protherapeutic behaviors. Second, we used a systematic observational methodology conducted by external observers to assess both protherapeutic behaviors and the working alliance. This methodological approach addressed impression management and social desirability issues when using offenders’ self-reports, as well as potential biases in facilitators’ self-reports or observations (Bennett, 2007; Gracia et al., 2015; Juarros-Basterrechea et al., 2018; Santirso et al., 2018; Weber et al., 2019).

The sample consisted of 153 males convicted of intimate partner violence against women and court-mandated to a community-based BIP. Offenders had been sentenced to less than a two-year term in prison, did not have previous criminal records, and their sentence was suspended on the condition that they attended the intervention. Eligibility criteria for this study were the following: (a) men over 18 years of age, (b) who had no severe psychological disorder, (c) had no severe substance abuse problems, and (d) had signed an informed consent form. Mean age was 40.73 years (SD = 11.99, range: 18-78). The sample consisted primarily of Spanish (71.7%, n = 110), Latin American (n = 17, 11.2%), European (other than Spanish, n = 13, 8.6%), African (n = 11, 7.2%), and Asian (n = 2, 1.3%) males. Regarding educational level, 6.4% had no education, 51% had primary education, 31.5% had secondary education, and 11.1% had university education. As for marital status, 33.3% were single (n = 51), 32.7% were divorced (n = 50), 24.2% were married or in a relationship (n = 37), and 9.8% were separated (n = 15). On average, annual family household income was between €6,000 and €;12,000. About half of the participants were unemployed 45.1% (n = 69) at the time of the initial assessment.

Standard batterer intervention program (SBIP). This condition consisted of 35 weekly group sessions of a standard cognitive-behavioral intervention. It was divided into six modules with the following main aims: a) first module: to build a climate of trust and establish norms for the group to function; b) second module: to introduce the IPV basic concepts and address attribution of responsibility; c) third module: to train in cognitive emotion management techniques and cognitive restructuring; d) fourth module: to develop awareness of IPV consequences on victims, empathy, and positive communication skills in intimate relationships; e) fifth module: to discuss sexist attitudes, gender roles, and gender equality; and f) sixth module: to consolidate learning objectives and prevent relapse. Several techniques were applied during the SBIP (i.e., group dynamics, role-playing, monitored exercises, and training in cognitive restructuring or emotion management skills). Closed-ended groups in both conditions ranged from 10 to 12 men per group and were led by two facilitators.

Standard batterer intervention program plus individualized motivational plan (SBIP + IMP). The experimental condition consisted of the same standard cognitive-behavioral intervention with the addition of the IMP (see description above).

Facilitators training and intervention adherence. Facilitators were psychologists with at least two years’ experience in BIPs. They received approximately 25 hours of training in their respective intervention condition. Facilitators were blind to the intervention condition. Each pair of facilitators intervened exclusively on one intervention condition. Facilitators for each condition were supervised independently once every two weeks. Supervision sessions focused on adherence to treatment protocol, group management, participants’ progress, and preparation of future sessions. To ensure the content of and adherence to the protocol, written intervention manuals for each condition were used (Lila et al., 2018).

Randomization. Participants assessed for eligibility came through the penitentiary system (N = 181). Twenty-eight men were excluded, mainly for not attending the first meeting (see Figure 1). A random number generator was used to allocate participants to the SBIP or the SBIP + IMP condition. Fourteen intervention groups were established, seven for the SBIP + IMP condition (n = 74) and seven for the SBIP condition (n = 79). Figure 1 provides the description of participant flow from recruitment to study completion.

Participants were clearly told that refusing to participate in the study would not affect their legal situation and would have no legal consequences. Confidentiality was assured, with the sole exception of situations that could pose a risk to participants or other people. Participants who agreed to take part in this study completed a written consent form and were randomly assigned to the SBIP + IMP or SBIP condition. Four independent trained graduate research assistants, who were blind to the objectives and hypotheses of the study, coded videotaped sessions. Working alliance and protherapeutic group behaviors were assessed twice: early in intervention (sessions 3-7) and late in intervention (sessions 24-28). Raters previously underwent training in which they assessed the same recorded session separately until they reached an acceptable level of agreement (i.e., not differing by more than one point on each assessed item). Two-hour recorded intervention sessions were divided into 24 five-minute intervals. Each participant received an average rating across session intervals on working alliance and protherapeutic group behaviors.

Working Alliance Inventory-Observer short version (WAI-O-S; Tichenor & Hill, 1989; Spanish version by Santirso et al., 2018). This observational scale assesses both general working alliance, and two components of working alliance (i.e., agreement and bond). The WAI-O-S contains 12 items (e.g., “There is a mutual liking between participant and facilitator”, “There is agreement on what is important for the participant to work on”). Raters responded on a 7-point Likert-type scale ranging from 0 (conclusive evidence against) to 7 (conclusive evidence in favor). The scale had adequate internal consistency in this study with Cronbach’s alpha equal to .97 and .92 for early and late in intervention measures, respectively. In a previous study with a sample of male IPV offenders, results showed an excellent level of inter-rater agreement and significant correlations with other indicators of intervention effectiveness (e.g., stage of change, motivation to change, and protherapeutic group behaviors; see Santirso et al., 2018).

Observational coding of protherapeutic group behavior (Semiatin et al., 2013). This observational tool assesses protherapeutic behaviors of group participants through their verbalizations. It is a 3-item measure of the following protherapeutic behaviors: (a) responsibility for abuse—participants’ verbalizations related to assuming vs. denying responsibility for their abusive actions, consequences of these actions, and the need for a personal change to avoid committing abusive acts in the future; (b) participant role behavior—interpersonal behaviors within the group that promote or hinder change by other participants; the coding system addressed four types of participant role behavior along two axes, confirmation vs. confrontation and negative progress vs. positive progress; and (c) group value—participant verbalizations related to the perceived value of the group and the intervention in general. Raters assessed each of the protherapeutic behaviors on a 5-point Likert-type scale ranging from 1 (conclusive evidence against) to 5 (conclusive evidence in favor). The effect of different raters assessing the same participants was evaluated. Pooled reliability correlation (r) across the three coded variables for averaged session ratings was .53, n = 30, p = .011.

To analyze whether participants who received SBIP + IMP and participants who received SBIP were equivalent at the time of allocation, chi-square and independent t-tests were conducted for categorical and continuous variables, respectively. To assess ‘time’ and ‘group’ differences in working alliance and protherapeutic behavior, we carried out repeated measures ANOVAs, with working alliance and protherapeutic group behaviors (early and late in intervention) as between-subject factors, and ‘intervention group’ (SBIP + IMP or SBIP) as the within-subject factor. When a factor was significant in previous ANOVAs, Bonferroni tests were performed. All statistical analyses were conducted using SPSS 26.0, and two-tailed tests with p set to .05 were considered as significant.

Table 1 contains information on sociodemographic characteristics by treatment condition. The results showed that randomization was satisfactory. No pretreatment characteristics significantly distinguished SBIP + IMP and SBIP groups.

There was a significant effect of ‘time’ on general working alliance, and on the agreement and bond subscales, F(1, 102) = 15.30, p = .0001, η² = .13; F(1, 102) = 15.65, p = .0001, η² = .13; F(1,102) = 12.49, p = .001, η² = .11, respectively, with higher scores later in intervention than early in intervention in the total sample (see Table 2). A significant effect of ‘group’ was also found on general working alliance, and on agreement and bond subscales, F(1, 102) = 12.50, p = .001, η² = .11; F(1, 102) = 9.77, p = .002, η² = .09; F(1, 102) = 16.63, p = .0001, η² = .14, respectively. Size effects were moderate to large (Cohen, 1988). Participants who received SBIP + IMP intervention showed higher general working alliance than those who received SBIP intervention, regardless of intervention moment (see Table 2 and Figure 2). No other significant effects were found on working alliance.

Additionally, a significant effect of ‘time’ was found on responsibility for abuse, participant role behavior, and group value scores, F(1, 102) = 12.81, p = .001, η² = .11; F(1, 102) = 7.00, p = .009, η² = .06; F(1, 102) = 13.20, p = .0001, η² = .12, respectively, with higher scores late in intervention compared to early in intervention in the total sample (see Table 2). Moreover, a significant ‘group’ effect was found on responsibility for abuse, participant role behavior and group value scores, F(1, 102) = 13.92, p = .0001, η² = .12; F(1, 102) = 10.63, p = .002, η² = .09, F(1, 102) = 20.85, p = .0001, η² = .17, respectively, with participants who received SBIP + IMP having higher scores (see Figure 3). Size effects were moderate to large (Cohen, 1988). Furthermore, a significant ‘time * group’ effect was found on responsibility for abuse and participant role behavior scores, F(1, 102) = 5.23, p = .02, η² = .05; F(1, 102) = 7.31, p = .008, η² = .07, respectively (see Table 2 and Figure 3). Specifically, participants who received SBIP + IMP intervention showed higher responsibility for abuse early (p = .001) and late in intervention (p = .013) than those who received SBIP. Additionally, participants who received SBIP + IMP condition had significantly higher participant role behavior scores (p = .002) and tended to show higher participant role behaviors scores late in treatment than those who received SBIP only (p = .067). Finally, responsibility for abuse and participant role behaviors significantly increased throughout intervention in participants who received SBIP (for all, p < .0001), but not in those in SBIP + IMP condition (for all, p > .35).