An Evaluation of Ethnicity-Matching for Caregiver Telehealth Training in India

Abstract

Telehealth has become an increasingly viable service delivery option for function-based treatment of problem behavior (Lindgren et al., 2016). However, few applications have occurred with participants outside the United States and little research exists evaluating the role that culture plays in service delivery. In the current study, we compared functional analyses and functional communication training delivered via telehealth by ethnically matched and ethnically distinct trainers to six participants in India. We measured the effectiveness using a multiple baseline design while collecting supplemental measures of sessions to criterion, cancellations, treatment fidelity, and social validity. We directly assessed preference for the ethnically matched and ethnically distinct trainers using a concurrent chains arrangement. Sessions with both trainers were effective in reducing problem behavior and increasing functional verbal requests for participating children, and treatment fidelity was high across training modalities. There were no major differences in sessions-to-criterion or cancellations across trainers. However, all six caregivers demonstrated greater preference for sessions with the ethnically matched trainer.

For over 6 decades, behavior analysts have developed and tested procedures to reduce problem behavior and increase adaptive behavior to improve the quality of life of individuals with disabilities and their families (Fahmie & Luczynski, 2020; Ghaemmaghami et al., 2016; Hanley, 2012). As a rule, these procedures begin with a functional analysis (FA; Iwata et al., 1994) to identify the cause of problem behavior by manipulating the environmental variables that set the occasion for such behavior to occur. Treatment options matched in function are then designed to reduce the occurrence of problem behavior and increase functionally equivalent alternative behaviors. Among several function-based treatment options, functional communication training (FCT) is used widely and has been shown to reduce a variety of problem behaviors across settings (Greer et al., 2016; Tiger et al., 2008).

Furthermore, the wider availability and access to the internet has resulted in function-based assessment and treatment services being offered remotely via a telecommunication modality. For example, Ferguson et al. (2019) reviewed 28 studies that leveraged telehealth technologies to deliver ABA-based remote training and supervision and found “at least some favorable outcomes” (p. 612) regarding the efficacy of this service delivery modality. In addition, they found that 12 studies evaluated FA and FCT procedures delivered via telehealth, and 10 of these reported positive outcomes (i.e., reduction in problem behavior and increase in functional requests). However, they also noted specific limitations such as insufficient participant demographic information in the included studies that precluded telehealth from being identified as an evidence-based practice based on the criteria described by Reichow et al. (2008). The review by Ferguson et al. did not report on the ethnicity or geographical location of the participants in the studies that were included. Nevertheless, telehealth interventions are particularly relevant in situations when face-to-face interventions are not possible, such as to families in remote or rural areas with little to no access to direct services.

Despite emerging evidence of the efficacy of FA and FCT delivered via telehealth for the reduction of problem behavior, few studies have evaluated these procedures when delivered to families in regions outside of North America. Unholz-Bowden et al. (2020) reviewed 30 studies that delivered ABA-based telehealth treatments and identified only two studies (Alnemary et al., 2015; Barkaia et al., 2017) that were conducted outside of North America, and neither of these studies targeted FA and FCT procedures for their participants. Sivaraman and Fahmie (2020a) conducted a review of ABA-based telehealth interventions delivered to regions outside of the United States. They included nine studies that delivered international behavior analytic telehealth treatments, of which only one study (Tsami et al., 2019) and one unpublished doctoral dissertation (Lee, 2015) evaluated FA and FCT procedures; both studies reported reductions in problem behavior for their participants. Thus, there seems to be a paucity of data on the efficacy of FA and FCT procedures implemented via telehealth in geographical settings outside of North America. This argument was also put forth in a review published by Neely et al. (2020) on telehealth-mediated behavioral assessments and interventions in which the authors noted the need for additional research on interventions designed for behavior reduction.

With the increasing demand for behavior analytic services across the globe, there is a need for a robust and thorough understanding of the efficacy and social validity of these procedures with culturally diverse families outside of the United States. Sivaraman and Fahmie (2020b) reported the efficacy and social validity of a manualized and culturally adapted caregiver training program in India. In this study, participating caregivers learned to conduct FA and FCT procedures in a small-group training format. All caregivers displayed improvements in the skills pertaining to function identification and functional communication training. The authors used the framework described by Bernal et al. (1995) for reporting cultural adaptations across eight key dimensions including language, the ethnicity of the person delivering the intervention, setting of the training, and content of the training. However, the authors did not compare the efficacy of any aspect of the culturally adapted training with standard training.

By contrast, multiple studies have been published comparing culturally adapted treatment with standard treatment in psychotherapy for culturally diverse children and youth (e.g., Flicker et al., 2008; Miranda et al., 2003). One aspect of culturally adapted treatment that has received attention is whether racial/ethnic matching between clients and therapists improves overall outcomes. Sue and Sue (2016) described the importance of several factors in ethnicity matching between clients and therapists including the age of the client, the client’s adherence to cultural values, and the specific intervention involved. For example, Kim and Atkinson (2002) noted that Asian American clients who were more traditional rated ethnically matched therapists as more empathetic, whereas the less traditional clients rated ethnically distinct therapists are more empathetic. Sue et al. (1991) observed that ethnicity-matching had a positive impact on Mexican American clients’ treatment outcomes and further noted that ethnic and language match was a predictor of better treatment outcomes in clients who did not speak English as their primary language. Likewise, African American clients reported greater preference for racially matched therapists (Flaherty & Adams, 1998) and ethnicity-matching was associated with lower dropout rates in Asian clients (Flaskerud & Liu, 1991).

However, it should be noted that not all previous studies on ethnicity-matching have reported positive outcomes. Cabral and Smith (2011) conducted a meta-analytic review on the impact of racial/ethnic matching in psychotherapy to evaluate its effect on clients’ preference for a therapist and overall therapy outcomes. Although they found no effect on treatment outcomes, clients indicated a moderately strong preference (d = 0.63) for a therapist of one’s own race/ethnicity. Researchers have argued that racial/ethnic matching may improve client outcomes due to a shared cultural history between the client and the therapist and by reducing client concerns about being misunderstood or mistreated (e.g., Kohatsu et al., 2000).

Furthermore, the recent review on the use of cultural adaptations in telehealth ABA services by Sivaraman and Fahmie (2020a) indicated that five out of nine studies involved some form of ethnicity or nativity matching. For example, in the study by Barkaia et al. (2017), the first author who delivered the intervention to participants with autism in Georgia was a native of Georgia who could speak fluently in both English and Georgian. Likewise, in the study by Neely et al. (2020) in which intervention was provided to participants in Japan, the coach was a native of Japan who could speak fluently in both English and Japanese. Tsami et al. (2019) included participants in Greece, Turkey, Mexico, and Ukraine and described using an interpreter for translations and matching the nativity of the trainer/interpreter with the participants. Although the trainer was ethnically matched with the participants in these and other studies included in the review by Sivaraman and Fahmie, none of the studies tested the impact of such matching.

Given these findings, it is important to conduct additional research comparing the effects of behavior analytic treatments delivered by ethnically distinct and ethnically matched trainers. Behavior analysts’ use of direct measures of behavior change of the intervention recipients (e.g., reduction of problem behavior in children with autism) and stakeholders (e.g., caregiver fidelity and preference) may lead to an enhanced understanding of the function of ethnicity matching on the adoption and success of services (see Sivaraman & Fahmie, 2020a, for a discussion), and contribute important insights in this general area of interest that cuts across disciplines. Thus, the purpose of the present study was to compare FA and FCT delivered via telehealth by ethnically matched and ethnically distinct trainers using (1) adult participant outcomes such as fidelity of implementation; (2) child behavior outcomes including measures of challenging behavior and functional requests; (3) sessions to complete assessment/treatment and number of session cancellations; and (4) participant preference for the two trainers.

Method

Participants and Settings

Six children in South India diagnosed with an autism spectrum disorder (ASD) and their mothers participated in the study. All participants were Indian. The mothers (henceforth referred to as the caregivers) of the children conducted all the sessions. Participants were recruited through flyers distributed on the first author’s social media accounts and through referrals from speech and occupational therapists in the region (i.e., the states of Tamil Nadu and Andhra Pradesh in India). To be eligible for the study, the child had to be diagnosed with ASD, as reported by the family and confirmed through documentation (if available). They also had to engage in problem behavior. All caregivers could speak English and the local language of the family (Tamil). Although Akhil’s mother spoke English and conducted the sessions herself, she requested that his father be present during the sessions with the ethnically distinct trainer. In addition, all caregivers had to have access to the internet and to a smartphone, laptop, or desktop computer to participate in the study. Although not a requirement for participation, the internet speeds of all participants were at least 10 Mb/s.

Akhil was an 8-year-old boy who attended an inclusive education school in Chennai in India. His target behaviors included screaming and self-injury, and he communicated using vocal speech. His score on the Childhood Autism Rating Scale (CARS; Schopler et al., 1980) was 32, based on administration 4 years before the study. Ruben was a 6-year-old boy who attended a special education school in Chennai. His problem behaviors included screaming and aggression. He could speak in short phrases. His score on the Childhood Autism Rating Scale (CARS; Schopler et al., 1980) was 35, based on administration 2 years before the study. Hari was a 7-year-old boy in Chennai who attended an inclusive school. His target behaviors were screaming and flopping. He communicated using short phrases and sentences. Dev was a 6-year-old boy from Madurai in India who studied in a special education school. He communicated in short phrases and his target behaviors included aggression and screaming. Both Hari and Dev were reported by the caregivers to be diagnosed with autism but could not produce documentation.

Hema was a 6-year-old girl from Hyderabad. She was not attending school during the study due to COVID-related school closures. Her score on the Childhood Autism Rating Scale (CARS; Schopler et al., 1980) was 33, based on administration 2 years before the study. She communicated using vocal speech and her problem behaviors included screaming and aggression. Tarun was a 6.5-year-old boy in Chennai, and he was not attending school during the course of the study due to COVID-related school closures. His problem behaviors included screaming and aggression. He required “substantial support” as per the DSM level of support scale, completed 2 years before the study. See Table 1 for a complete description of participant sociodemographic information.

Table 1.

Participant sociodemographic information

	Gender (age)	Caregiver (age)	Residence	Autism diagnosis		Treatment allocation
				CARS	DSM
Akhil	Male (8)	Mother (38)	Chennai, India	32 (30–37)		CA-FA/S-FCT
Ruben	Male (6)	Mother (36)	Chennai, India	35 (30–37)		S-FA/CA-FCT
Hari	Male (7)	Mother (38)	Chennai, India			S-FA/CA-FCT
Dev	Male (6.5)	Mother (33)	Madurai, India			CA-FA/S-FCT
Hema	Female (6)	Mother (34)	Hyderabad, India	33 (30–37)		S-FA/CA-FCT
Tarun	Male (6.5)	Mother (40)	Chennai, India		2 (1–3)	CA-FA/S-FCT

S Standard, CA Culturally Adapted, FA Functional Analysis, FCT Functional Communication Training, CARS Childhood Autism Rating Scale, DSM Diagnostic and Statistical Manual

Three children (Akhil, Dev, and Hari) could speak and respond to instructions in both English and Tamil, and the other three (Ruben, Hema, and Tarun) could speak and respond to instructions in English alone. Sessions were conducted via telehealth with the experimenters present in their home or office, and the participants present in their respective homes. Caregivers conducted sessions in the living room (Akhil, Ruben), bedroom (Dev, Tarun), or child’s study/playroom (Hema, Hari). Participant assent was gained before each session by asking the child if they were ready to begin playing.

Four trainers who were all master’s level behavior analysts were involved in the study. One trainer was associated with culturally adapted training (henceforth, ethnically matched trainer) whereas the other three trainers were associated with standard training (henceforth, ethnically distinct trainer). The ethnically matched trainer was Indian, native to the region of Chennai and was matched in cultural identity with the participants. She had 7 years of direct experience in behavioral interventions with children with ASD, and 3 years of experience providing telehealth interventions for children with ASD and their families. She could speak in English and Tamil with the participants and was based in Belgium during the study. She did not know any of the participants prior to the study. The three ethnically distinct trainers had approximately 4 years of experience providing direct behavioral services and 2 years of experience providing telehealth interventions. They spoke in English with the caregivers. The ethnicities of the ethnically distinct trainers were Mexican American, Middle Eastern American, and Caucasian, and they had never lived in India. They were all located in California during the course of the study. The ethnically distinct trainers were all blind to the purpose of the study.

Dependent Measures and Response Definitions

Problem behavior included screaming, defined as vocalizations above conversation level; aggression, defined as the hands or other body parts forcefully (e.g., to leave a mark or make an audible sound) coming into contact with the parent’s face and body, pinching others, or pushing others; self-injury, defined as the child’s body forcefully (e.g., to leave a mark or make an audible sound) coming into contact with objects or surfaces; and flopping, defined as throwing the body on the floor from a standing or sitting position. Functional requests were defined as emitting an appropriate vocal request without prompts. We used 10-s partial interval recording for problem behavior and frequency recording for functional requests. Data on problem behavior were converted to percentage of intervals scored, and data on requests were converted to percentage of opportunities in which an independent response occurred.

We also collected data on caregivers’ procedural integrity during the FA and FCT sessions using procedures similar to those described by Suess et al. (2014) and Tsami et al. (2019). The data were reported as a percentage by dividing the number of correctly implemented steps by the total number of steps and multiplying by 100. The steps in the FA were the delivery of the relevant antecedent (i.e., removal of an item, presentation of a demand, removal of attention) and relevant consequence (i.e., returning an item, or removing a demand, or providing attention contingent on problem behavior). The steps in FCT were (1) delivering the antecedent (e.g., removal of tangible item); (2) prompting the functional request, if necessary; (3) delivering the consequence contingent on functional request; and (4) maintaining the antecedent if problem behavior occurred. The trainer prompted removal of reinforcers every 30 s during the tangible sessions of the FA and the FCT sessions. These particular prompted responses were removed in our measure of fidelity because there was no opportunity for these responses to be emitted independently. However, all other steps (e.g., delivery of the reinforcer contingent on the functional request during FCT) were only counted correct when the caregiver completed them independently without any coaching from the trainer.

In addition, we measured the number of sessions required for the child to exhibit differentiated responding during the FA and to meet the predetermined mastery criteria for the functional request during FCT. We considered differentiation achieved in the FA if problem behavior was higher in one or more test conditions compared to the control condition across at least three sessions. However, we extended the FA by a session or two past the point of differentiation for a few participants to accommodate our multiple baseline design. We considered mastery criteria met when the child exhibited 100% independent functional requests with no problem behavior across two consecutive sessions. We did not include the delay training during FCT (see below) sessions because the terminal delay duration and thinning schedule varied across participants. We also measured the total number of appointments and cancellations for each participant. Finally, we measured the caregivers’ selection responses during the preference assessment conducted at the end of the study by recording their written response to an email request.

Immediately after the conclusion of training, the caregivers completed a modified version of the Treatment Acceptability Rating Form (Reimers & Wacker, 1988) and the social validity survey used by Tsami et al. (2019). The survey evaluated the caregivers’ acceptance of the procedures and the use of telehealth as a modality to receive services. The survey consisted of 10 questions rated on a 7-point Likert scale. See Table 2 for a complete list of survey questions and caregiver responses.

Table 2.

Social validity survey and mean caregiver responses

Items	CA-FA/ S-FCT (Mean)	Range	S-FA/ CA-FCT (Mean)	Range
How acceptable do you find the functional analysis assessment procedure? (1 = unacceptable; 7 = very acceptable)	6.8	6–7	6.8	6–7
How acceptable do you find the treatment to be regarding your concerns about your child? (1 = unacceptable; 7 = very acceptable)	7	7	6.8	6–7
How likely is this treatment to make permanent improvements in your child’s behavior? (1 = unlikely; 7 = very likely)	6	5–7	6.3	6–7
How willing are you to carry out this treatment? (1 = not at all willing; 7 = very willing)	6.8	6–7	7	7
How much time will be needed each day for you to carry out this treatment? (1 = much time; 7 = little time)	6.8	6–7	6.8	6–7
How confident are you that the treatment will be effective? (1 = not confident; 7 = very confident)	6.3	6–7	6.3	6–7
How willing would you be to change your family routine to carry out this treatment? (1 = not at all willing; 7 = very willing)	6	5–7	6.3	6–7
How disruptive will it be to your family (in general) to carry out this treatment? (1 = very disruptive; 7 = not at all disruptive)	6.8	6–7	6.2	6–7
How effective is this treatment likely to be for your child? (1 = not at all effective; 7 = very effective)	6.2	6–7	6.8	6–7
How well will carrying out this treatment fit into your family routine? (1 = not at all well; 7 = very well)	6.2	6–7	6.2	6–7

S Standard, CA Culturally Adapted, FA Functional Analysis, FCT Functional Communication Training

Interobserver Agreement

Secondary observers independently scored approximately 35% of the sessions in each condition. The observers were fluent in English and the ethnically matched trainer created transcripts for the sessions that were conducted in Tamil. To calculate interobserver agreement, sessions were divided into consecutive 10-s intervals. Interobserver agreement data on problem behavior, functional requests, and caregiver responses were calculated by dividing the number of intervals for which both observers agreed on the number of responses (for functional requests) or the occurrence or nonoccurrence of the response (for problem behavior and procedural integrity steps) by the total number of observation intervals and converting to a percentage. Interobserver agreement data for problem behavior during the FA and FCT sessions averaged 99% (range: 83%–100%) and 98% (range: 93%–100%), respectively, across all six participants. Interobserver agreement data for functional requests during the FCT sessions averaged 98% (range: 90%–100%) across all participants. Interobserver agreement data for caregivers’ procedural integrity during the FA and FCT sessions averaged 95% (range: 87%–100%) and 98% (range: 83%–100%), respectively, across all participants.

Cultural Adaptations

Sessions conducted by the ethnically matched and ethnically distinct trainers were conducted exactly in the manner described by Wacker, Lee, Padilla Dalmau, et al. (2013a); Wacker, Lee, Dalmau, et al. (2013b). The ethnically matched trainer offered caregivers the choice to conduct the sessions in either Tamil, English, or a combination of both based on their preference. Hari’s caregiver opted for sessions with the ethnically matched therapist to be in Tamil; all other caregivers chose to use both languages. These sessions were bilingual and the ethnically matched trainer took the caregiver’s lead and responded in the same language as the caregiver during interactions.

In addition, we employed some overarching adaptations across participants that seemed necessary to ensure successful participation in the study. In particular, all participants were recruited through the ethnically matched trainer’s social media account or the referrals of allied rehabilitation professionals in the region. All caregivers met with the ethnically matched trainer during the intake interview when they were offered details about the study.

Experimental Design

A multiple baseline design across participants was used in the study. The ethnically matched trainer was randomly assigned during the FA for half of the participants. For these same participants, the ethnically distinct trainer conducted the FCT that followed. For the other half of the participants, the sequence of sessions with the ethnically matched and ethnically distinct trainer was reversed, with the ethically distinct trainer for the FA and the ethnically matched trainer for FCT. We counterbalanced trainer type across conditions due to the potential for sequencing (e.g., one trainer type always was experienced first/last or always was associated with a particular condition type) to affect outcomes. A concurrent chains arrangement was used to assess preference following completion of FA and FCT training.

Procedure

Intake Interview

The ethnically matched trainer completed a 1-hr appointment before the start of the FA with each caregiver. First, the caregiver completed and signed the informed consent forms to participate in the study. Then, the trainer and caregiver identified topographies of problem behavior, developed operational definitions, and discussed the conditions under which the targeted problem behavior occurred. In addition, the trainer explained the purpose of the FA and the FCT to the caregiver. The caregivers were informed that they would be assigned two trainers, one for the FA and another for the FCT, and that this was being done in order for the treatment team to maintain an optimum caseload.

Functional Analysis

Conditions included attention, tangible, escape, and play, alternated in a multielement design. All sessions were 5 min; all test sessions were initiated with the presentation of an establishing operation (attention removal, tangible removal, or demand delivery) and included a continuous schedule of reinforcement for problem behavior. Procedures were identical to those described by Wacker, Lee, Dalmau, et al. (2013b), with one exception. Dev’s problem behavior occurred during mealtimes and he refused to eat until he accessed his iPad. The tangible condition of his FA was conducted in the presence of a snack or a meal, and access to the iPad was removed at 30-s intervals. Caregivers conducted the sessions with coaching from the trainer in a manner similar to that described by Wacker et al. For example, if the parent did not deliver the putative reinforcer contingent on targeted problem behavior, the trainer vocally prompted the caregiver to do so.

Functional Communication Training

Procedures were similar to those described by Wacker, Lee, Padilla Dalmau, et al. (2013a). All sessions were 5 min. Multiple sessions were conducted during each appointment with scheduled breaks in between sessions. The experimenters (i.e., the ethnically matched and ethnically distinct trainers) designed the FCT based on the results of the FA and in consultation with the caregivers. If results of the FA suggested multiple functions, the trainer described the FCT procedures for each function to the caregiver, advised the caregiver to prioritize the most difficult situation(s) they experienced with their child at home to target first, and based the initial treatment on the caregiver’s preference. Hema’s problem behavior was found to be multiply controlled, and the caregiver chose to target only the tangible function because this was a priority for the family.

The trainer and caregiver determined the form of the functional request before FCT. Table 3 displays the targeted function(s) of problem behavior and the selected request for each participant. For Tarun, his problem behavior occurred when the caregiver took away his toys (usually Lego blocks, or other construction toys) and modified the way they were constructed. So, we chose the request “Please don’t touch my toys,” and the caregiver returned the toy without modifying it. The trainer explained the FCT procedure to the parent before the beginning of each session.

Table 3.

Functions of problem behavior and functional requests chosen for each participant

Participant	Function	Functional request
Akhil	Access to tangible	“My way with Lego blocks”
Ruben	Access to tangible	“I want the blocks”
Hari	Access to tangible	“Please give me the phone”
Dev	Access to tangible	“Give me the iPad”
Hema	Access to tangible and escape	“Give me the phone, please”
Tarun	Access to tangible	“Please don’t touch my toys”

Because all six participants’ problem behaviors were maintained by access to tangible items, the trainer instructed the parent to remove the preferred item at the beginning of each FCT session, to immediately vocally prompt the functional request, to provide praise and access to the tangible for 30 s contingent on a correct request, and to withhold access to the tangible item contingent on problem behavior. The trainer always prompted the removal of reinforcers every 30 s during the tangible sessions of the FA and the FCT sessions. We chose to continue prompting the schedule of delivery of antecedents so that the caregiver could focus on delivering accurate topographical responses to problem behavior. During the initial sessions, the trainer vocally prompted the caregiver to implement the other components of FCT (e.g., “You can return the iPad now”) as well and provided immediate feedback for correct and incorrect responses (e.g., “Nice job returning the iPad”). The trainer then gradually delayed the prompts to permit the caregiver to respond independently and waited until the end of the session to deliver detailed feedback. The trainer smiled, nodded, used gestures, or said “great job” intermittently during the session for correct implementation of the FCT. There was no set schedule for fading the prompts and each trainer delayed the prompts based on the caregiver’s progress.

Once the children made 100% independent functional requests across two consecutive sessions with no problem behavior, the trainer commenced delay training during which the children were required to wait for a few seconds or minutes before being provided access to the item. The terminal delay duration was determined by the caregiver in consultation with the trainer conducting FCT sessions. During delay training, the session was conducted exactly like the FCT condition described above until the child made an independent functional request. Once the child made a request, the caregiver said “Great job asking nicely. You have to wait” and then proceeded to deliver an instruction pertaining to the activity the child was to engage in during the delay, if applicable (e.g., “Draw a circle”; “Eat one bite of food”). The delay training began with an initial duration of 5 s and increased if there were two consecutive sessions with no problem behavior and at least 75% independent functional requests. For Tarun alone, delay durations began with 3 s because it took longer for him to emit the functional request without engaging in problem behavior. During the delay periods, Dev engaged in eating bites of food, Hema and Ruben completed academic activities, and the remaining participants watched the caregiver engage with the reinforcer (e.g., blocks). In Tarun’s case, during delay training, the caregiver responded to his request with “Great job saying nicely ‘It is my turn now’” and modified the construction toy or added components to it while he played with other toys. At the end of the delay period, the modified toy was returned to him.

Preference Assessment

After the conclusion of the training, we offered three additional consultation sessions to each caregiver. Prior to each session, caregivers were given a choice between the ethnically matched and ethnically distinct trainers. The caregiver was sent an email with the following text:

We are available for additional consultation sessions. You can use them for advice on any challenges you are facing, tips for teaching new skills (ex: social skills with peers, group skills, classroom behaviours etc.) or use them as direct sessions with the child being present.

Do let us know if you are interested in this. Kindly indicate whether you want the session with [ethnically matched trainer’s name], or with [ethnically distinct trainer’s name].

The email was signed by both trainers and sent from the email address associated with the project so that it did not appear that it was coming from a particular trainer. The order in which the trainers’ names appeared was counterbalanced across trials. The caregiver was blind to the purpose of the study and their responses were recorded as selection responses in the preference assessment. If the caregiver indicated that they did not have a preference (which occurred for only one participant), they were sent the following email signed by both trainers:

Thank you for indicating that you would like to continue with the consultation sessions. We are more than happy to work with you and [child’s name]. We prefer that you make a choice (between [ethnically distinct trainer’s name] and [ethnically matched trainer’s name]) because we believe you might be a better judge of what works best with your child and yourself. For the moment, both therapists are available and happy to help!

No data were recorded during the consultation sessions.

Results

Data on problem behavior and functional requests for all six participants are displayed in Figs. 1 and 2. Problem behavior during the FA served as a baseline. The results of the FA indicated that problem behavior was maintained by access to tangible items for five of the six participants. For the sixth participant (Hema), problem behavior was maintained by both access to items and escape. All of the participants’ problem behaviors decreased to zero levels during FCT and remained low when the caregiver delayed access to the reinforcer by a few seconds or few minutes. In addition, all participants emitted independent functional requests by the end of the training sessions. All caregivers conducted the FA and FCT sessions with an average fidelity of at least 80%.

Fig. 1.

Problem behavior and functional requests measured during the culturally adapted functional analysis (FA) and standard functional communication training (FCT) for Akhil, Dev, and Tarun

Fig. 2.

Problem behavior and functional requests measured during the standard functional analysis (FA) and culturally adapted functional communication training (FCT) for Ruben, Hema and Hari

Akhil, Dev, and Tarun completed the FA with the ethnically matched trainer and FCT with the ethnically distinct trainer. Akhil required 12 FA sessions to show differentiation across conditions. His problem behavior was maintained by access to toys (e.g., Lego blocks). At the end of the training, his functional requests were at 100%, and he could tolerate delays of up to 3 min before his requests were reinforced. Dev required 11 FA sessions to achieve a differentiated outcome and his problem behavior was maintained by access to the iPad. He emitted no functional requests during baseline. During FCT, his functional requests increased to 100% and were maintained when the parent gradually required him to eat two bites of food before he could request for the iPad. Tarun required 14 FA sessions and his problem behavior was maintained by access to toys. During FCT, his functional requests increased to 100% and he tolerated delays of up to 30 s, during which his caregiver played with the toys maintaining problem behavior while he played with other toys. Akhil, Dev, and Tarun took 4, 6, and 33 sessions (M: 14) to meet mastery criterion for the functional request during FCT respectively.

Ruben, Hema, and Hari completed the FA with the ethnically distinct trainer and FCT with the ethnically matched trainer. Ruben required 12 FA sessions for a differentiated outcome and his problem behavior was maintained by access to toys. He could independently emit functional requests by the end of the training and tolerate delays of up to 2 min. Hema’s FA required 15 sessions to show that problem behavior was maintained by access to the smartphone and escape from academic tasks. The caregiver chose to address the tangible function because it was associated with the highest levels of problem behavior. During FCT, her functional requests for the smartphone increased to 100% and she could wait for up to 2 min, during which she completed academic tasks, before getting access to the phone. Hari’s FA took 19 sessions and his problem behavior was maintained by access to the smartphone. His functional requests increased steadily with the introduction of FCT, and he could tolerate delays of up to 2.5 min before his requests were reinforced by the end of the training. Ruben, Hema, and Hari took 15, 3, and 7 sessions (M: 8) to meet mastery criterion for the functional request during FCT, respectively.

The caregivers’ mean ratings on the social validity survey are displayed in Table 2. Overall, caregivers in both groups (i.e., those who received the FA with the ethnically distinct trainer and FCT with the ethnically matched trainer, and vice-versa) indicated that they found the training acceptable, that the procedures were not time-consuming, and that they could implement the procedures with their children. They also indicated that they found the functional analysis acceptable and that they believed the training would produce permanent behavior change. In addition, there were minimal differences between the mean ratings of caregivers of both groups.

The total number of appointments and cancellations for each participant is displayed in Table 4. On average, the FCT with the ethnically distinct trainer took 11 appointments (4, 8, and 21 appointments each for Akhil, Dev and Tarun) and FCT with the ethnically matched trainer took 5 appointments (5, 4, and 5 appointments each for Ruben, Hema and Hari). The difference in the mean number of FCT appointments can primarily be attributed to Tarun who underwent FCT with the ethnically distinct trainer and required 21 appointments. Across all 26 appointments scheduled with the ethnically matched trainer (both FA and FCT), 1 appointment (with Hari) was cancelled, whereas a total of 17 appointments out of 54 scheduled appointments with the ethnically distinct trainers were cancelled. Again, most of these cancellations occurred with Tarun’s caregivers during his extended time in the FCT phase. The reasons for cancellation reported by the caregivers included misunderstanding of the scheduled time, sudden travel plans, and other commitments. Based on parent choice and trainer availability, sessions were predominantly conducted in the mornings (participants’ local time) and this meant that ethnically distinct trainers conducted sessions late at night and the ethnically matched trainers conducted sessions early in the morning in their local times.

Table 4.

Sessions to differentiation or mastery, total appointments, cancellations and treatment fidelity during FA and FCT for each participant

Participant	FA				FCT
	Appts.	Canc.	TF Mean (Range)	Sessions to Diff.	Appts.	Canc.	TF Mean (Range)	Sessions to Mastery
	Culturally adapted FA				Standard FCT
Akhil	3	0	90% (66–100)	12	4	2	93% (75–100)	4
Dev	5	0	94 (75–100)	11	8	1	91% (61–100)	6
Tarun	4	0	96 (80–100)	14	21	12	96% (71–100)	33
	Standard FA				Culturally adapted FCT
Ruben	5	1	96% (75–100)	12	5	0	91% (66–100)	15
Hema	8	0	89% (66–100)	15	4	0	82% (75–100)	3
Hari	8	1	95% (75–100)	19	5	1	94% (73–100)	7

Appts. Planned appointments, Canc. Cancellations, TF Treatment Fidelity, FA Functional Analysis, FCT Functional Communication Training, Diff. Differentiation

During the preference assessment conducted after FCT, Akhil, Dev, Hari, and Ruben’s caregivers completed three trials each. They all chose sessions with the ethnically matched trainer each time. Hema’s caregiver completed two preference assessment trials and opted for sessions with the ethnically matched trainer both times. The third trial could not be completed because the family was relocating and did not have access to internet services. Tarun’s caregiver completed two trials and initially indicated that she had no preference between the two trainers. However, when the experimenter further prompted her to make a choice (see above), she opted for a session with the ethnically matched trainer.

Discussion

We compared the efficacy, fidelity, and preference for telehealth behavioral services delivered by ethnically matched and ethnically distinct trainers to six families in India. The FA and FCT services delivered by both types of trainers were equally effective in the reduction of problem behavior and acquisition of appropriate alternative behavior, respectively, for all child participants. We did not observe major differences in the sessions to meet our mastery criterion or caregiver fidelity across both training modalities. All caregivers reported that the FA and FCT procedures (delivered by both types of trainers) were acceptable, and that it was likely that treatment would result in permanent behavior change in their children. In addition, we measured two secondary variables to evaluate differences between the training provided by the two types of trainers. First, we found no systematic differences in the number of session cancellations across both types of trainers. Second, all six caregivers demonstrated exclusive preference for sessions with the ethnically matched trainer.

To the best of our knowledge, this is the first study to directly compare behavior analytic training delivered by ethnically matched and ethnically distinct trainers. Our findings suggest that both training modalities were effective at identifying the maintaining variables and subsequently producing reductions in problem behavior and increases in functional requests. These findings lend additional support to the efficacy of function-based behavior analytic treatments for problem behavior. The individualized nature of the assessment and treatment may have contributed to this outcome and addressed cultural differences between the participants and trainers.

However, it should be noted that certain overarching adaptations were made to both training modalities in order to carry out the study. For example, our recruitment materials were distributed to rehabilitation professionals in the targeted region, and these professionals recommended our training to some families under their care. This was important to obtain the participants necessary for carrying out such a study. In addition, all caregivers could afford their own internet, could speak in both English and Tamil, and lived in a large city. Thus, one could argue that the current study did not include a comprehensive set of demographics that would benefit from this type of culturally adapted training (i.e., families from rural areas with low socioeconomic status, education, and acculturation levels). Including such families presented specific experimental and practical challenges to address our research questions at this preliminary stage of investigation. Further, given that behavior analytic practitioners were made accessible to the families in the current study, they may have been more likely to seek services as opposed to other families who do not regularly have access to such services (either from local region or elsewhere).

It is encouraging to note that we found no differences in the primary dependent variables (i.e., child problem behavior, functional requests) across the training offered by the two types of trainers, and the goal of producing socially significant behavior change was met in both conditions. An intriguing finding of the current study was that sessions with the ethnically matched trainer were chosen by all of our participants during the preference assessment. Indeed, a critical question for future research pertains to why sessions with the ethnically matched trainer were preferred. The behavioral processes at action could have been that the ethnically matched trainer engaged in behavior that served as a positive consequence for the caregiver’s verbal behaviors or that the trainer identified culturally relevant discriminative stimuli during interactions with the caregiver (e.g., a frown, or a shake of the head as disapproval) and modified their own behavior accordingly. Sue and Sue (2016) also note that matching per se may not be as important as a therapist’s willingness to engage with sociocultural issues. Although our study was not designed to answer this question, conducting closer function-based investigations on caregiver-trainer interactions could be an area for future research (see Larsen et al., 2022, for a recent study focusing on this line of research).

However, the apparent impact of ethnicity-matching on preference must be interpreted with caution for a few reasons. First, the ethnically matched trainer had an average of 3 years more experience delivering behavior analytic services than the ethnically distinct trainers. In addition, all families met with the ethnically matched trainer during the intake interview and may have had local knowledge of the trainer prior to the study. These factors could have influenced preference to some degree; for example, it could be argued that the ethnically matched trainer had an opportunity to establish rapport prior to the experimental comparison. Future research could remediate this by matching the trainers on years of experience and opportunities for interaction with participants prior to the investigation. In addition, future research should incorporate a variety of additional ethnically matched and ethnically distinct trainers to establish the generality of this finding.

It should be noted that in the current study we used a single-subject experimental design to compare the effects of ethnicity-matching. Previous research investigating or comparing such training for psychotherapy commonly employs ratings of hypothetical future therapists (Cabral & Smith, 2011) or group designs (i.e., randomized clinical trials or crossover designs; see Hall et al., 2016, for a review). Our design allowed us to directly evaluate efficacy without requiring a waitlist or withholding training sessions with the ethnically matched trainer altogether, as would be standard in a group design on this question. Further, counterbalancing the order in which training was received (i.e., first receiving sessions with the ethnically matched or ethnically distinct trainer) allowed us to detect potential biases related to the recency of interactions with a particular therapist or with a particular training context (e.g., preference for the trainer associated with FCT simply due to that trainer having the most recent interactions). Thus, the design presented here could serve as a preliminary model for future research on this important topic.

A significant difficulty in comparing the effects of ethnicity-matching is attempting to isolate and control all possible variables in the trainers’ behavior. Whether evaluated in a single subject or group design, research on this topic is challenged by isolating specific cultural adaptations while remaining sensitive to the unique needs of families. Although a new researcher in this area may be inclined to suggest matching the trainers on all behaviors except a few select “adaptations,” it is not possible to do so while providing the personalized attention that is required by the ethical guidelines published by the Behavior Analytic Certification Board (see BACB, 2021, ethics code 1.05c). In addition, there are subtle behaviors that naturally vary across cultures and that would be difficult to program or manualize across trainers. One example encountered in this line of research was a particular side-to-side head nod, that was interpreted as “no” by a trainer from the United States but that is a sign of agreement for people native to Chennai. In the current study, the ethnically matched trainer likely introduced cultural adaptations unsystematically and the ethnically distinct trainers may have introduced similar adaptations incidentally. For example, the ethnically matched trainer suggested that the caregiver use phrases that were commonly used in the region during interactions (e.g., “I will play now” instead of “My turn” during the tangible condition of the FA). It seems unlikely that the ethnically distinct trainers introduced the same adaptations as those used by the ethnically matched trainer, given that the ethnically distinct trainers were blind to the purpose of the study, were unfamiliar with people from this region of India, and reported anecdotally that they did not introduce such adaptations. Nevertheless, researchers may more carefully manualize both training protocols in the future so that differences are clearer and more consistent.

Despite these limitations, we feel our findings lay an important foundation for future research. Investigations aimed at identifying the factors that contribute to differences in caregiver preference such as ours seem crucial to understand how culture and ethnicity impact intervention. The next step in this line of research is to offer a more refined analysis of the features of interaction that differed between the two types of trainers. For example, videos of sessions may be used to code caregiver–therapist interactions, and these interactions may be analyzed for various reactions (e.g., positive, negative, confused; see Larsen et al., 2022). These reactions themselves must be carefully defined based on local cultural norms. Another avenue for future research could be to obtain additional measures that may intersect with preference for culturally adapted intervention, such as sustained interest in services or the likelihood of referral. Measures of maintenance of targeted skills in the children and their caregivers may also serve as a supplement to the subjective measure of social validity used in the current study (Kennedy, 2002).

Our findings should not be misconstrued to suggest that families in India should only be served by Indian therapists. On the contrary, assuming that certain cultural adaptations are found to reliably affect behavioral services in a positive way, the next question is whether those adaptations can be trained for use in therapists who are not matched in background with the families they serve. That is, can U.S.-based therapists obtain training in the cultural sensitivity needed to serve diverse families in a manner similar to therapists matched in ethnicity? This line of research is particularly important given the relatively low percentage of certified behavior analysts who are representative of those cultures found in under-resourced areas outside of the United States (BACB, 2021). Although the international migration that has resulted in cultural diversity within countries or social groups is more recent, the diversity across social groups in different parts of the world is by no means new and has been documented extensively. Diverse populations may require culturally adapted approaches to meet their goals, and behavior analysts must evolve to suit the needs of their clients. Paul (1967) highlighted the importance of studying “What treatment, by whom, is most effective for this individual with that specific problem, and under which set of circumstances,” and half a century later, this is still relevant to professionals delivering behavioral services to culturally dissimilar individuals. We hope that the current study will encourage more researchers to investigate the role that culture plays in interventions, and eventually advance the precision, scope, and depth of behavior analysis.

Declarations

Conflict of Interest

The authors declare no conflict of interest.

Ethics Statement

The study was approved by the University of Nebraska Medical Centre’s Institutional Review Board. We obtained informed consent from all participants.