Morris et al. (2023) PDF - A Review of Methods of Assessing Social Stimuli Preference

Received: 12 October 2022 Accepted: 16 February 2023 DOI: 10.1002/jaba.981 RESEARCH ARTICLE A review of methods of assessing preference for social stimuli Samuel L. Morris 1 | Madison L. Gallagher 2 | Alva E. Allen 2 1 Department of Psychology, Lo...

Received: 12 October 2022 Accepted: 16 February 2023 DOI: 10.1002/jaba.981 RESEARCH ARTICLE A review of methods of assessing preference for social stimuli Samuel L. Morris 1 | Madison L. Gallagher 2 | Alva E. Allen 2 1 Department of Psychology, Louisiana State Abstract University, Baton Rouge, United States 2 Research on preference and reinforcer assessments has historically focused on the Department of Psychology, Southeastern Louisiana University, Hammond, United States evaluation of edible and leisure reinforcers, but the identification and use of indi- vidualized social reinforcers may be beneficial for several reasons. Recently, many Correspondence studies have evaluated methods of assessing preference for social stimuli. The pro- Samuel L. Morris, Louisiana State University, cedures and outcomes across these studies have varied greatly, and the current 236 Audubon Hall, Baton Rouge, LA, 70803, USA. Contact: state of evidence for methods of assessing preference for social stimuli remains Email: [email protected] unclear. Thus, we conducted a review to synthesize all previous research in this area, evaluate the evidence for the utility of this general methodology, and iden- Editor-in-Chief: John Borrero tify factors that may influence its utility. Our results suggest that social preference Handling Editor: Corey Stocco assessments are likely to produce results that correspond with reinforcer assess- ment hierarchies. Preference assessment modality was one factor that influenced utility; video-based preference assessments had greater correspondence with rein- forcer assessments than did other modalities. Directions for future research and implications for the use of social preference assessments in research and practices are discussed. KEYWORDS attention, preference assessment, reinforcer assessment, social interaction Stimulus preference assessments originated as a method inherently more complex because they require the stimuli of identifying preferred items or activities for individuals approached or selected to be, or become, discriminative who could not otherwise communicate their preferences for the social stimuli delivered contingent on their (e.g., Pace et al., 1985) and have developed into a tech- selection. nology for making objective, data-based decisions about The identification and use of individualized social what stimuli are likely to function as effective reinforcers reinforcers could be beneficial for several reasons (Hagopian et al., 2004). Until recently, the research liter- (e.g., DeLeon et al., 2013; Morris & Vollmer, 2019). ature has primarily focused on assessing preference for First, social reinforcers are practical to deliver and are edible and leisure items (e.g., DeLeon & Iwata, 1996; readily accessible in everyday life. Given their ubiquity, Fisher et al., 1992; Hanley et al., 2003; Roane et al., the use of social reinforcers during behavioral interven- 1998), whereas relatively few studies have evaluated tions may facilitate maintenance and generalization methods of assessing preference for social stimuli (Stokes & Baer, 1977). Second, the identification of social (e.g., Nuernberger et al., 2012). This is reflected by the reinforcers may allow for more reinforcer variability and fact that researchers have reviewed and summarized greater choice between reinforcers that may improve the methods of preference assessment in general (Hagopian social validity of intervention procedures or outcomes et al., 2004; Heinicke et al., 2019; Tullis et al., 2011), but (Hanratty & Hanley, 2021). Third, Board-Certified a corresponding review does not yet exist for methods of Behavior Analysts are ethically obligated to prioritize assessing preference for social stimuli. One potential rea- least restrictive procedures, maximize client benefits, and son for these discrepancies is that, unlike edible or leisure ensure that they do no harm to the client (Behavior Anal- items, social stimuli cannot be directly approached ysis Certification Board, 2020, Guidelines 2.15, 2.16, or selected and instead must be delivered contingent 2.17, 3.01). The use of social reinforcers may decrease on approach or selection of some other stimulus reliance on foods high in sugar or salt and sedentary lei- (e.g., picture cards; Kelly et al., 2014). As a result, sure activities that can increase the risk of related health methods of assessing preference for social stimuli are problems (e.g., Chen et al., 2016; Must et al., 2014). 416 © 2023 Society for the Experimental Analysis of Behavior (SEAB). wileyonlinelibrary.com/journal/jaba J Appl Behav Anal. 2023;56:416–427.  19383703, 2023, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/jaba.981 by Cochrane Canada Provision, Wiley Online Library on [26/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License REVIEW OF SOCIAL PAS 417 Fourth, the identification of individualized social rein- on methods of assessing preference for social stimuli and forcers may be helpful in preventing social indifference or evaluate the efficacy of existing methods using consistent avoidance (Morris & Vollmer, 2022b), which could result measures. Such a review could summarize the evidence in a loss of access to learning opportunities and socially for methods of assessing preference for social stimuli in mediated reinforcers. More specifically, methods of asses- general, demonstrate which methods of preference assess- sing preference for different types of social stimuli allow ments are most likely to produce valid and useful results, us to interact with an individual in ways that are reinfor- and identity variables that affect validity and utility. cing to them rather than expecting them to enjoy or toler- Thus, in this review we synthesized all available research ate the type of interaction we find reinforcing. evaluating methods of assessing preference for different Perhaps because of these benefits, the number of stud- types of social stimuli. Our goals were to determine the ies evaluating methods of identifying social reinforcers degree of empirical support for this general methodology has increased in recent years (Morris & Vollmer, 2020a). and to elucidate procedural characteristics that may The increase in research in this area has also brought influence its efficacy and utility. with it an increase in the variance of procedures used to assess preference for different types of social stimuli. One way that procedures have varied is the method of prefer- METHOD ence assessment used. Researchers have conducted multiple-stimulus-without-replacement preference assess- Identification of articles ments (MSWO; e.g., Neurnberger et al., 2012), paired- stimulus preference assessments (PSPA; e.g., Kelly The process of determining current evidence for methods et al., 2014), and social interaction preference assessments of assessing preference for social stimuli began with a (SIPA; e.g., Morris & Vollmer, 2019). A second way pro- three-step search for articles. First, we used EBSCO Host cedures have varied is the preference assessment modal- to search the following five online databases: Academic ity. Researchers have conducted preference assessments Search Complete, Education Resources Information in which pictures (e.g., Lang et al., 2014), videos Center, Psychology and Behavioral Sciences Collection, (e.g., Wolfe et al., 2018), people (e.g., Clay et al., 2013), PsycArticles, and PsycInfo using the following search and vocal utterances (e.g., Morris & Vollmer, 2020a) terms: preference assessment with reinforcer assessment served as the stimuli to be selected or approached, which and social interaction or attention. We also conducted a resulted in the contingent delivery of a particular type of search in Google Scholar using these same terms. Second, social interaction. A third way procedures have varied is we evaluated the Method section of articles identified in the type of reinforcer assessment conducted as a point of this initial search to determine whether they met our evaluation for the preference assessment and how many inclusion criteria: that they (a) conducted at least one stimuli from the preference assessment were evaluated. For preference assessment in which the relative occurrence of example, reinforcer assessments have involved concurrent- approach or selection responses that produced at least operants assessments (e.g., Morris & Vollmer, 2019), two different types of social stimuli was evaluated. progressive-ratio assessments (Harper et al., 2021), and An additional requirement was that articles (b) included evaluations of skill acquisition (those including measures a reinforcer assessment in which the occurrence of a dif- of acquisition as the dependent variable, Morris & ferent response (i.e., not selection or approach) was eval- Vollmer, 2020d). Researchers have included as few as one uated across conditions in which responding produced at and as many as seven stimuli from the preference assess- least two of the different social stimuli previously evalu- ment in the reinforcer assessment and included as little as ated or across conditions in which it produced no pro- 12.5% and as much as 100% of all preference assessment grammed consequences (i.e., control or baseline) and at stimuli in the reinforcer assessment. least one of the different stimuli previously evaluated. We Despite this variance, most studies have demonstrated evaluated each of the first 500 articles (sorted by rele- one or more preference assessment methods to be effec- vance) returned by the initial search to determine whether tive in identifying individualized social reinforcers for they met the inclusion criteria. We did not search beyond most participants. However, the degree to which existing the first 500 search results because all studies meeting methods of preference assessment discriminate relative inclusion criteria were identified within the first reinforcer efficacy and consistently identify the most effi- 100 search results provided. Thus, after reviewing cacious reinforcers remains unclear. Moreover, the differ- 400 search results without any study meeting inclusion ences in procedures have not been systematically varied criteria, the search was terminated. Next, the third step within or across most studies. As a result, it remains entailed screening all articles included in the reference unclear how these differences in procedures may affect section and subsequent citing research of the articles whether a preference assessment permits one to identify identified through steps one and two. Taken together, the reinforcers or discriminate differences in relative rein- entire search process yielded 20 studies that met inclusion forcer efficacy. One way to elucidate these points of criteria. All were peer-reviewed articles, dissertations, or uncertainty is to review and synthesize existing research theses written in English and available before September  19383703, 2023, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/jaba.981 by Cochrane Canada Provision, Wiley Online Library on [26/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 418 MORRIS ET AL. 21, 2022. We included unpublished dissertations and the- characteristics were not consistently reported across the ses with the intention of reducing the influence of publi- articles, which made their interpretation or use in further cation bias on our synthesis of research in this area data analysis untenable. (Paez, 2017). Assessment-level measures Dependent measures For each assessment conducted with each participant, we Article-level measures coded the number of stimuli included in the assessment (including control options with no programmed conse- For each article, we recorded the author(s), title, year quences) and the hierarchy of preference or relative rein- published, total number of participants, type(s) of prefer- forcer efficacy produced by the preference and reinforcer ence assessment used, and type(s) of reinforcer assess- assessments. These measures were based on the authors’ ment used. For each preference assessment conducted, descriptions of the results and graphs included in the we coded the type and modality. Assessment type articles and served to summarize the results of each referred to the name or description given by the authors. assessment for subsequent analyses. For preference The categories coded for preference assessment type were assessments, the percentage of trials on which each stimu- PSPA, MSWO, and SIPA. Additional preference lus was selected or the order in which they were removed assessment types would have been coded separately from the array produced the hierarchy. For single- and (e.g., response-restriction preference assessment, Hanley concurrent-operants reinforcer assessments, the relative et al., 2003), but none were conducted in the included rate of responding across sessions produced the hierar- studies. Preference assessment modality referred to the chy. For progressive-ratio reinforcer assessments, the rel- stimuli that were approached or selected during the ative cumulative number of responses or breakpoint assessment. The categories coded for preference assess- schedule value produced the hierarchy. For skill acquisi- ment modality were picture cards, videos, people, and tion, the relative percentage correct, trials to mastery, or vocal utterances. We also coded the type of each rein- skills mastered produced the hierarchy. For all cases in forcer assessment by the name or description provided by which the authors’ description or depiction of the result the authors. The categories coded for reinforcer assess- indicated a tie, we assigned all stimuli for which a tie was ment type were single operant, concurrent operants, pro- obtained a rank equal to the mean of the ranks remaining gressive ratio, and skill acquisition (i.e., those including (e.g., if a tie was obtained for the last two stimuli out of percentage correct, trials to mastery, or other measures of an array of five, then they were both assigned a rank of acquisition as the dependent variable). One exception to 4.5). For cases in which nonsocial stimuli were also eval- this categorization was Lang et al. (2014), who evaluated uated (e.g., Butler & Graff, 2021; Clay et al., 2018), the effects of high- and low-preference social stimuli on we only analyzed the preference and reinforcer assess- the occurrence of correct responses and challenging ment results pertaining to social stimuli. An additional behavior during discrete trial training. Due to the similar- assessment-level measure that we attempted to code was ity of these procedures to those described as skill- the amount of time required to complete each assessment acquisition reinforcer assessments, Lang et al. (2014) was (i.e., assessment duration). However, these data were not included in that category. consistently provided by the authors, and in some cases, procedural detail was insufficient to estimate the maxi- mum time an assessment could have required. Thus, Participant-level measures these data were also excluded from subsequent analyses. Several measures were coded at the level of each partici- pant. These measures included authors’ descriptions of Data analysis basic demographics, including sex, age, diagnosis, and communication modality. Communication modality was We evaluated correspondence between preference and coded into two categories, vocal and alternative or aug- reinforcer assessment results using three different ana- mentative communication. More specific categorization lyses ranging from more comprehensive and complex to of communication modality (e.g., sign, picture exchange) practical and straightforward. First, for each comparison was not coded because it was not consistently available of the results between preference and reinforcer assess- across articles. We also attempted to code the level of ments, we calculated the percentage of comparison with communication (e.g., one-word mands or full sentences) agreement. This served as a comprehensive, relatively and the results of specific (e.g., matching identical stringent measure of correspondence between the prefer- pictures of social interactions) or general (e.g., Verbal ence and reinforcer assessments and how effectively pref- Behavior Milestones Assessment and Placement Program, erence assessments differentiated the relative reinforcer Sundberg, 2014) skills assessments, but these participant efficacy of social stimuli. Percentage of comparison with  19383703, 2023, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/jaba.981 by Cochrane Canada Provision, Wiley Online Library on [26/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License REVIEW OF SOCIAL PAS 419 agreement was calculated by evaluating the relative ranks identification of the most reinforcing stimulus. for all stimuli included in both the preference and rein- In contrast, instances in which the highest preferred stim- forcer assessment. Each rank for which the same social ulus was at least one rank lower or shown not to function stimuli were indicated by both assessments was scored as as a reinforcer in the reinforcer assessment constituted a an agreement, whereas any rank for which two different lack of correspondence in the identification of the most social stimuli were indicated was scored as a disagree- reinforcing stimulus. Larger differences in ranks were ment. Then, we divided the number of ranks for which indicative of less correspondence. In the example given there was an agreement by the total number of compari- above, the highest preferred stimulus according to the sons (i.e., total number of stimuli included in both the preference assessment (i.e., dancing) was not the same preference and reinforcer assessment) and converted that rank in the reinforcer assessment, but only one rank off. value to the percentage of comparison with agreement by Although it constituted a more lenient, rudimentary test, multiplying by 100. For example, consider a hypothetical instances in which only the highest preferred stimulus data set in which the preference assessment included five was evaluated during the reinforcer assessment were stimuli and yielded the following hierarchy (from highest coded as corresponding in their indication of the most to lowest): dancing, hugs, high fives, praise, and control. reinforcing type of social stimulus if it was demonstrated Then, the reinforcer assessment included three of those to function as a reinforcer. same stimuli and yielded the following hierarchy: high The most reinforcing social stimulus may support fas- fives, dancing, and control. To calculate the percentage ter skill acquisition (Morris & Vollmer, 2020d) and a of comparison with agreement, we focus only on the greater number of responses (Harper et al., 2021; three stimuli included in both assessments and assign Morris & Vollmer, 2020a) relative to, for example, the them new ranks after excluding the other stimuli (e.g., in second-most reinforcing social stimulus. However, in the preference assessment, high fives would be ranked 2/3 some cases “any reinforcer will do” and it may not be rather than 3/5 for the purpose of this analysis). Then, necessary to ensure the most reinforcing stimulus has each rank is compared and scored as an agreement or dis- been identified (e.g., when trying to support low-effort, agreement. The same stimulus is indicated in rank established skills). Therefore, third, we analyzed whether three for both assessments, but they indicated different the stimuli indicated as highest preferred by the prefer- stimuli for ranks one and two. Thus, there would be ence assessment were found to function as reinforcers to agreement for 1/3 ranks, which would result in a percent- any extent by evaluating the percentage of data sets for age of comparison with agreement of 33%. Instances in which the first-, second-, and third-most preferred stimu- which only one social stimulus from the preference lus of each preference assessment maintained more assessment (e.g., the highest preferred) was evaluated in responding relative to baseline or a control condition the reinforcer assessment were scored as an agreement if during the reinforcer assessment. We evaluated the top that stimulus was shown to function as a reinforcer rela- three social stimuli because almost all data sets included tive to a control (i.e., no programmed consequences) or at least three types of social stimuli. Additionally, evalu- baseline condition. Instances in which one assessment ating multiple stimuli allowed us to ascertain whether indicated a tie and the other did not were scored as dis- only the highest preferred stimuli were likely to function agreements; however, if both assessments indicated a tie as reinforcers or whether multiple highly preferred stimuli for the same stimuli at the same rank, then an agreement were likely to function as reinforcers. In the example was scored. We chose to use this unconventional, strin- above, the highest ranked stimulus (i.e., dancing) func- gent measure instead of Spearman rank-order correlation tioned as a reinforcer; the second highest (i.e., hugs) was coefficients to ensure a valid, meaningful measure even in not evaluated in the reinforcer assessment, and therefore cases for which few stimuli were evaluated, the number it would be excluded from this analysis; and the third of stimuli compared varied across data sets, or assump- highest (i.e., high fives) functioned as a reinforcer. This tions for conducting the correlation coefficients were not served as the most practical and lenient measure of corre- met (e.g., monotonicity). spondence between the preference and reinforcer assess- Although hierarchies of relative reinforcer efficacy ment results. may be useful (e.g., when implementing differential rein- forcement by quality), researchers and clinicians may often only be interested in identifying the most reinfor- Interrater agreement cing types of social stimuli. Thus, in the second analysis, we evaluated whether the preference assessment identified The review of the database search results was com- the most reinforcing stimuli by analyzing the reinforcer pleted by one rater, but a second rater independently assessment rank of stimuli indicated as the highest pre- reviewed the first 100 articles of the database search ferred by the preference assessment. Instances in which results for the purposes of establishing interrater the highest preferred stimulus in the preference assess- agreement for the application of the inclusion criteria. ment was the highest ranked (i.e., most reinforcing) in the Agreement was obtained for 96 out of 100 of the arti- reinforcer assessment constituted correspondence in cles reviewed by both raters. Both raters reviewed  19383703, 2023, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/jaba.981 by Cochrane Canada Provision, Wiley Online Library on [26/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 420 MORRIS ET AL. and discussed the disagreements and ultimately TABLE 1 Breakdown of articles reached a consensus about whether they met inclusion Article characteristics Articles Participants criteria. Disagreements primarily pertained to studies Journal that evaluated social interaction in general (e.g., Kanaman et al., 2022) but did not evaluate dif- Behavior Interventions 2 4 ferent types of social stimuli. All data were coded Behavior Modification 1 3 from the Method and Results sections of each study Education and Training in Autism and 1 3 by the second author. To evaluate interrater agree- Developmental Disabilities ment, a second rater independently coded all vari- European Journal of Behavior Analysis 1 3 ables for 7/20 articles (35%), accounting for 71/100 Journal of Applied Behavior Analysis 8 65 (71%) of the total participants. Interrater agreement Journal of Behavioral Education 2 7 was calculated by comparing observers’ data across Learning and Motivation 1 2 each dependent measure (i.e., article, participant, and Research in Autism Spectrum Disorders 1 2 assessment-level measures) for each assessment con- Research in Developmental Disabilities 1 5 ducted with each participant in each study. A second rater reviewed the first out of every three articles Nonpublished 2 6 when they were organized in alphabetical order by Year published/created first author. For each variable, an agreement was 2010–2014 5 18 scored if both raters entered the exact same value. 2015–2019 7 21 A disagreement was scored if there was any discrep- 2020–2022 8 61 ancy in the value recorded. The number of agree- Totals 20 100 ments was divided by the total comparisons and then that quotient was multiplied by 100 to obtain the interrater agreement score. Following this initial eval- TABLE 2 Breakdown of participant characteristics uation, interrater agreement was 94%. The disagree- ments identified in the process of evaluating Participant characteristics Articles Participants interrater agreement were systematic. One set of dis- Sex agreements pertained to the raters using two different Male 19 69 versions of the same study (i.e., a dissertation that Female 12 31 was published during data analysis process), and Diagnosis another set of disagreements pertained to a study with multiple types of preference assessment, which No diagnosis 2 33 were organized in a different order by the raters Autism spectrum disorder 19 59 (i.e., one coder recorded data for the PSPA first, Other 4 8 whereas another coded data for the MSWO first). Age rangea This pattern of disagreements indicated no problems 1–6 13 58 with the definitions used to code the dependent mea- 7–12 11 26 sures, so the authors reached consensus on all dis- 13–18 6 10 agreements and ensured related concerns were not 19–24 4 4 present in the data for which interrater agreement was not evaluated. Communication modalitya Vocal 13 71 AAC 8 21 RESULTS Note. AAC = alternative or augmentative communication. a Indicates participant characteristics that were not specified for all participants in Characteristics of the studies included in the review, their some articles. participants, and the assessments that they evaluated are summarized in Tables 1, 2, and 3. Rather than describe all data included in the tables, we will highlight a few and under 6 years of age, and they often communicated notable characteristics of each. Table 1 shows that most vocally. In reviewing Table 3, note that in many cases of this research has been published in the Journal of multiple types of assessment or multiple assessments of a Applied Behavior Analysis but has also appeared in eight given type were conducted with each participant other journals. The years 2020 to 2022 account for 40% (e.g., Morris & Vollmer, 2020a), so we also show the of the studies in which 61% of the participants with total number of data sets (i.e., individual preference whom methods of assessing preference for social stimuli assessments that were compared with a reinforcer assess- were evaluated. Table 2 shows that participants were ment). Table 3 shows that PSPAs and picture-based often male, diagnosed with autism spectrum disorder, assessments have been evaluated with the largest numbers  19383703, 2023, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/jaba.981 by Cochrane Canada Provision, Wiley Online Library on [26/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License REVIEW OF SOCIAL PAS 421 of participants and data sets. Additionally, concurrent reinforcer assessment results, the percentage of each com- operants and progressive-ratio reinforcer assessments parison with agreement. Specifically, for each of the have served as the point of comparison most frequently. 270 comparisons of preference and reinforcer assessment Next, we will review the results of our analyses across all results, we determined the percentage of stimuli that was data sets. However, see the online Supporting Informa- included in both the preference and reinforcer assessment tion for a table summarizing the results of all analyses (i.e., were compared) for which the same relative rank across each type of preference and reinforcer assessment (i.e., an agreement) was obtained. The left panel depicts evaluated by each study included in the review. all data, with each data point representing a single data Figure 1 depicts our first and most comprehensive set or comparison (i.e., a single preference assessment for evaluation of correspondence between preference and a given participant compared with their reinforcer assess- ment). The height of the gray bar indicates the mean per- centage of comparison with agreement across all data TABLE 3 Breakdown of assessment characteristics sets, M = 71%, 95% CI [67, 76]. However, a high degree Assessment characteristics Articles Participants Data sets of variability across data sets was observed. Although PA type many resulted in 100% agreement, many others resulted in 0% agreement. The right panel of Figure 1 serves to PSPA 15 90 145 further clarify this picture by showing the percentage of MSWO 4 16 78 data sets for which each level of agreement (e.g., ≥81%) SIPA 3 18 47 was obtained. For 60% of all data sets, a percentage PA modalitya agreement greater than or equal to 81% was obtained. Pictures 13 77 209 However, the second-most likely level of agreement was Video 6 18 18 much lower; for 19% of all data sets, a percentage agree- Person 3 10 10 ment less than 20% was obtained. Vocal 2 6 30 Although hierarchies of relative reinforcer efficacy may be useful, this level of precision is not needed for RA type some purposes. Figure 2 demonstrates the degree to Single operant 5 11 56 which preference assessments identified the most effica- Concurrent operants 10 30 68 cious reinforcers (left panel) or indicated highly preferred Progressive ratio 3 42 88 stimuli that functioned as reinforcers to any extent (right Skill acquisition 5 25 58 panel), which correspond to our second and third ana- Note. PA = preference assessment, PSPA = paired-stimulus preference assessment, lyses of the data, respectively. The left panel shows that MSWO = multiple stimulus without replacement preference assessment, SIPA = for 76% of data sets, the social stimulus identified as social interaction preference assessment, and RA = reinforcer assessment. highest preferred by the preference assessment was also a Indicates an assessment characteristic that was not specified by all articles. Article count is greater than 20 due to some articles conducting multiple types of found to be the most reinforcing during the reinforcer assessments across multiple modalities of assessments. assessment. For cases in which the highest preferred F I G U R E 1 Percentage of comparison with agreement for all data sets. The left panel depicts the percentage of comparison with agreement across all data sets, each represented by a single data point. The height of the gray bar indicates the mean percentage of comparison with agreement. The right panel depicts the percentage of data sets that fell within each range of percentage of comparison with agreement, to more clearly depict the distribution of values.  19383703, 2023, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/jaba.981 by Cochrane Canada Provision, Wiley Online Library on [26/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 422 MORRIS ET AL. F I G U R E 2 Percentage of data sets for which the highest preferred were the most reinforcing or reinforcing to any extent. The left panel depicts the percentage of data sets for which the highest preferred stimulus stayed the same or changed in rank from the preference to the reinforcer assessment. The right panel depicts the percentage of data sets for which the one-, two-, and three-most-preferred stimuli were found to function as reinforcers. HP = highest preferred. social stimulus was not found to be the most reinforcing, TABLE 4 Assessment characteristics influence on correspondence it was observed to be the second-most reinforcing social Percentage of Percentage for which stimulus evaluated (i.e., a change of one rank) for 17% of comparison with HP was the most data sets. Discrepancies of more than one rank were agreement reinforcing much less likely, in part because few reinforcer assess- Assessment ments evaluated several social stimuli that were included characteristics Mean 95% CI Percentage 95% CI in the preference assessment. The right panel shows that PA type the social stimuli indicated as first-, second-, and third- PSPA 71 (65–78) 77 (70–84) most preferred stimuli were observed to function as rein- MSWO 79 (71–86) 79 (69–88) forcers for 96, 87, and 82% of data sets, respectively, rela- tive to a baseline or control condition in which no SIPA 63 (54–73) 68 (55–82) programmed consequences were included or relative to PA modality other types of social stimuli evaluated. Picture/card 68 (63–74) 73 (67–79) One purpose of this review was to evaluate how Video 87 (79–95) 94 (84–105) different procedural characteristics may influence how well Person 90 (70–110) 90 (70–110) a preference assessment can identify social reinforcers and Vocal 82 (70–94) 87 (74–100) delineate their relative utility. Accordingly, we evaluated RA type how preference-assessment type, preference-assessment Single-operant 69 (56–81) 74 (62–86) modality, and reinforcer-assessment type influenced the percentage of comparison with agreement and the percent- Concurrent- 77 (69–86) 80 (71–90) operants age of data sets for which the preference assessment Progressive-ratio 53 (46–60) 60 (49–69) identified the most reinforcing stimulus. Table 4 summa- rizes the results of this analysis. These two measures of Skill-acquisition 97 (93–101) 100 (N/A) correspondence varied across preference assessment type Note. HP = highest preferred, PA = preference assessment, PSPA = paired- and modality; however, the differences were often small stimulus preference assessment, MSWO = multiple stimulus without replacement preference assessment, SIPA = social interaction preference assessment, and there were few data sets in some categories RA = reinforcer assessment, CI = confidence interval, and N/A = not applicable. (i.e., person), so the corresponding 95% CIs frequently overlapped. One notable exception was that video-based preference assessments had a higher percentage of compar- assessment type, but the differences were often small ison with agreement and a higher percentage of data sets and the 95% CIs often overlapped. However, one that identified the most reinforcing stimulus than did exception was that preference assessments often had picture-based preference assessments. These two mea- higher correspondence with the results of concurrent- sures of correspondence also varied across reinforcer- operants and skill-acquisition reinforcer assessments  19383703, 2023, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/jaba.981 by Cochrane Canada Provision, Wiley Online Library on [26/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License REVIEW OF SOCIAL PAS 423 which yielded a moderate negative correlation of 0.70, 95% CI [ 0.85, 0.46]. DISCUSSION This review of the literature serves to synthesize the avail- able evidence for methods of assessing preference for social stimuli. The results of this synthesis allow several important conclusions. The results of our analysis suggest that preference assessments for social stimuli are likely to accurately differentiate relative reinforcer efficacy as defined by reinforcer assessment outcomes. Specifically, the percentage of stimuli compared for which agreement F I G U R E 3 The relation between proportion of the hierarchy in rank was obtained was 71%, on average, and agree- compared and percentage of comparison with agreement. Each data ment greater than 81% was obtained for 60% of all point depicts the mean percentage of comparison with agreement and 270 data sets. Given this high level of correspondence proportion of hierarchy compared for a given type of preference to with reinforcer assessments about relative reinforcer effi- reinforcer assessment comparison within each study. The dashed line is cacy, preference assessments for social stimuli can be use- a trend line fit to the data via simple linear regression. CI = 95% confidence interval; r = Pearson correlation coefficient. ful in cases for which a hierarchy of differentially reinforcing social stimuli is needed, such as differential reinforcement by quality for independent versus than they did with progressive-ratio reinforcer prompted responses (e.g., Campanaro et al., 2020; assessments. Cividini-Motta & Ahearn, 2013; Johnson et al., 2017) or Another procedural characteristic that may influence appropriate versus problem behavior (e.g., Briggs correspondence is the number of stimuli included in the et al., 2019; Kunnavatana et al., 2018). preference assessment that were subsequently evaluated However, researchers and clinicians may often be during the reinforcer assessment (i.e., the proportion of more interested in identifying effective reinforcers than in the total preference hierarchy that was compared). Essen- differentiating relative reinforcer efficacy. Thus, a more tially, this entails a difference in the rigor with which the practical question is whether preference assessments can preference assessment was evaluated. For example, identify the most efficacious reinforcers—or at least stim- instances in which all of the stimuli included in the prefer- uli that are likely to function as reinforcers to any extent. ence assessment were evaluated in the reinforcer assess- The results of our review provide an answer: preference ment constitute a more rigorous or thorough evaluation assessments are likely to identify the most efficacious relative to those including a smaller proportion of the social reinforcers (76% of data sets) and to indicate highly preference hierarchy (i.e., 0.4 or two out of five stimuli). preferred social stimuli that function as reinforcers Figure 3 shows how the proportion of the total prefer- (96% of data sets). Preference assessments for social stim- ence hierarchy compared influenced the percentage of uli produced useful results for a clear majority of data comparison with agreement across studies. Each data sets and, therefore, may be suitable for routine reinforcer point represents the mean proportion of the hierarchy identification in the context of research and practice. compared and mean percentage of comparison with They almost always indicated highly preferred stimuli agreement for a given method of preference assessment that functioned as effective social reinforcers and were (e.g., video-based PSPA) evaluated by a given study likely to identify the most efficacious social reinforcers, (e.g., Davis et al., 2022). In general, higher levels of which may support much more responding (Harper agreement were obtained for the studies that evaluated a et al., 2021; Morris & Vollmer, 2020a) and facilitate fas- smaller proportion of the total preference hierarchy. ter skill acquisition (Morris & Vollmer, 2020d). Additionally, most studies for which 100% agreement Despite the general finding that preference assess- was obtained for all participants evaluated less than half ments for social stimuli can be useful in identifying effec- of the preference-assessment hierarchy. In contrast, for tive social reinforcers and discriminating their relative the studies that evaluated the entire preference assessment efficacy, they did not do so for many data sets included hierarchy more variability in the mean percentage com- in this review. This variance in preference assessment parison with agreement was observed. Some studies still utility indicates that additional research is needed to obtained high levels of agreement, whereas others identify and evaluate variables that may contribute to obtained moderate or low agreement. We quantified the this variance. In this review, we hoped to highlight how relationship between the proportion of the hierarchy differences in procedural characteristics may influence compared and the percentage of comparison with agree- preference assessment utility. Though the unsystematic ment by calculating a Pearson correlation coefficient, variation in procedural and participant characteristics  19383703, 2023, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/jaba.981 by Cochrane Canada Provision, Wiley Online Library on [26/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 424 MORRIS ET AL. inhibits us from drawing clear conclusions, our results skills of individuals with whom they are conducted. As suggest that video-based preference assessments may noted in the Method section, we intended to record and yield results that have greater correspondence with rein- analyze the effect of these variables as a part of the cur- forcer assessments. This finding seems intuitive given that rent review but could not to do so due to inconsistency videos or graphic-interchange-format images (i.e., GIFs) in procedure or reporting across studies. Thus, one sim- may increase both the discriminability of stimuli to be ple direction for future research is to include assessment approached or selected and their correspondence with the results that generally characterize participants’ reper- in vivo delivery of social interaction. As a result, video- toires or to evaluate specific skills related to discriminat- based preference assessments may help participants to ing and identifying social interactions. For example, respond in a way that is indicative of their preference scores on the visual/perceptual, matching-to-sample, for and the relative reinforcing efficacy of different and listener portions of the Verbal Behavior Milestones social stimuli. Our results also suggest that preference Assessment and Placement Program (Sundberg, 2014); assessments may have higher correspondence with skill- outcomes of assessments of sociability (Morris & acquisition and concurrent-operants reinforcer assess- Vollmer, 2020c, 2021, 2022c); accuracy in identification ments than with progressive-ratio reinforcer assessments. of pictures of social interaction as a speaker and listener One explanation for their higher degree of correspon- (Morris & Vollmer, 2020a); and accuracy in matching dence is that skill-acquisition and concurrent-operants pictures and videos of social interaction to correspond- reinforcer assessments provide access to social stimuli on ing in vivo social interactions (Morris & a schedule similar to that used for preference assessments Vollmer, 2020b; Wolfe et al., 2018) may all covary with (i.e., a fixed-ratio 1 schedule), whereas progressive-ratio the validity and utility of different forms of preference reinforcer assessments provide access to social stimuli on assessment. These data may help to account for variance a very different schedule than those used during prefer- in the efficacy of different assessments for a given partic- ence assessments. Notably, progressive-ratio reinforcer ipant, across participants, and across studies, as evi- assessments may be useful for evaluating the utility of denced by recent research on social preference preference assessment results in contexts with leaner assessment (Morris & Vollmer, 2020a) and other topics schedules of reinforcement (e.g., during schedule thin- (e.g., Pizarro et al., 2021). If future research can consis- ning). Investigating this potential utility is an important tently incorporate these measures, then it may become direction for future research. possible to differentiate participants for whom more Another procedural characteristic that we found to intensive methods of assessing preference for social stim- account for some variance in correspondence between pref- uli (e.g., SIPA; Morris & Vollmer, 2019) are necessary erence and reinforcer assessments was the proportion of the from participants for whom more efficient methods preference hierarchy evaluated. Studies that included a (e.g., vocal PSPA or single-session picture-based more comprehensive or rigorous analysis of the preference MSWO; Morris & Vollmer, 2020a) may suffice. Ulti- assessment also permitted the identification of discrepancies mately, this could allow clinicians and researchers to that were not otherwise apparent. Thus, it may be beneficial minimize time and resource requirements and maximize for future research that seeks to improve methods of social validity and utility on an individual basis. preference assessment to focus on video-based assessments Relatedly, future research pursuant of these goals and to evaluate their validity and utility by incorporating should not lose sight of one of the most fundamental all or most of the same stimuli in subsequent reinforcer functions of preference assessment: to individualize pro- assessments. Future research should also continue to sys- cedures and identify preferred stimuli or activities. This tematically evaluate procedural characteristics of the assess- function of preference assessment is especially important ment, such as methods of preassessment exposure to the for individuals who cannot communicate their prefer- contingencies; modality (e.g., vocal versus picture versus ences in other ways. Despite the convenience and clarity video) and assessment type (e.g., PSPA versus MSWO); such exclusion may afford, individuals with less devel- and characteristics of how the assessment is implemented, oped communication and discrimination skills should such as the implementor or assessor (e.g., Huntington & not be excluded from participation in this research. In Schwartz, 2022) and other contextual variables fact, they should be the focus of it; they are the individ- (e.g., location, history with assessor, preceding activities). uals who stand to benefit the most from the continued Some procedural characteristics that were not included in development of methods of assessing preference for our review, such as the type of social stimuli evaluated, social stimuli. For example, the validity of a video- reinforcer-assessment response requirements, and time that based preference assessment for individuals who cannot elapsed between preference and reinforcer assessments, limit respond to questions like, “Do you want hugs or chase?” the conclusions that can be drawn and will be important to and cannot match videos or pictures to corresponding include and evaluate in future research. in vivo interactions is more meaningful and useful to Another direction by which future research may consider than whether the same assessment is effective account for variance in preference assessment utility is for individuals who can consistently display those skills by systematically evaluating the current repertoire and (see Morris & Vollmer, 2020a, 2020b, for related  19383703, 2023, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/jaba.981 by Cochrane Canada Provision, Wiley Online Library on [26/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License REVIEW OF SOCIAL PAS 425 discussions). Moreover, the continued improvement of for general reinforcer identification, they are more likely methods of assessing preference for social stimuli could to use indirect (i.e., observations, asking parent) or infor- complement and facilitate improvement in our use of mal (i.e., ask the client, provide choice) methods than participatory practices (Pritchett et al., 2021), obtaining they are to use preference assessments due to time and participant assent (Morris et al., 2021), and obtaining training constraints (Graff & Karsten, 2012). Although it objective, participant-focused measures of social valid- remains unclear how this may vary across different types ity (Hanley, 2010). of reinforcers (i.e., leisure versus social stimuli), it is Our inclusion criteria for this review were restrictive important to consider these alternative methods of identi- in some ways and lenient in others, which limited the fying individualized social reinforcers. Recent research breadth and clarity of our conclusions, respectively. We has found that relative to several methods of preference required that the research involved a preference assess- assessment, clinician-reported hierarchies of the relative ment that evaluated at least two different types of social reinforcer efficacy of social stimuli are less strongly corre- stimuli, as well as a reinforcer assessment in which at lated with reinforcer-assessment hierarchies and less least one of those same social stimuli were evaluated. likely to identify the most efficacious reinforcers This was important to allow for consistent methods of (Morris & Vollmer, 2022a; see their Figure 5). More evaluation and comparison across studies but excluded research on indirect or informal methods of identifying and some methods of identifying individualized social rein- evaluating social reinforcers is needed; however, as of now forcers. One method that was excluded was the indepen- it seems such methods may not be a suitable replacement dent use of reinforcer assessments (Piazza et al., 1999; for social preference assessments. Moreover, although we Smaby et al., 2007), which represented some of the earli- could not systematically evaluate preference assessment est evaluations of methods that were effective at identi- duration in the current review, recently developed methods fying individualized social reinforcers and have been found to take fewer than 10 min to discriminate differentiating their relative efficacy. Although such relative reinforcer efficacy (e.g., Morris & Vollmer, 2020a) studies were seminal in the development of methods of and fewer than 5 min to consistently identify effective social assessing preference for social stimuli, they are not as reinforcers (e.g., Morris et al., 2023) and, therefore, may be straightforward or efficient as most current methods of feasible for routine use. social preference assessment are. Excluded were those In summary, our review of existing evidence suggests evaluating preference for or interest in different conver- that social preference assessments are an effective tech- sation topics (Hood et al., 2022; Kronfli et al., 2022; nology for identifying individualized social reinforcers Stocco et al., 2021). To date, researchers have only used and discriminating their relative reinforcer efficacy. Still, indirect assessments of topic preferences via caregiver additional research is needed to continue improving report, which did not meet our inclusion criteria. How- methods of assessing preference for social stimuli and ever, preferred conversation topics represent a particu- accounting for variance in their utility. Given the many larly ubiquitous and useful form of individualized social potential benefits of individualized social reinforcers, an reinforcement. Thus, an important direction for future important next step will be to evaluate whether and how research is developing and evaluating methods of they are identified and used in clinical practice and in directly assessing an individual’s conversation topic research. preferences. Our inclusion criteria also resulted in the inclusion A C K N O W L E D G M EN T S of a few studies with goals beyond identifying and eval- We would like to thank Sarah Lucia, Ian Hekmatdoost, uating individualized social reinforcers. For example, Adelaide Pizzuto, and Sara Powell for their comments Butler and Graff (2021) used social preference assess- and suggestions. ments as one component of their evaluation of the sta- bility of preference across time. Similarly, Clay et al. C O N F L I C T O F I N T E R E S T S TA T E M E N T (2018) and Leaf et al. (2014) used social preference The authors declare no conflict of interest. assessments as one component of their evaluation of the relative efficacy of social and nonsocial reinforcers. Due ETHICS AP PR OV A L to these differences in purpose, such studies did not No human or animal subjects were used for the produc- include as comprehensive of an evaluation of social tion of this article. preference assessment results as did studies that solely sought to improve preference-assessment methodology. OR CID Nevertheless, they still provide important points of evi- Samuel L. Morris https://orcid.org/0000-0003-0724- dence for the utility of preference assessments in identi- 6954 fying effective social reinforcers. In the current review, we focused on summarizing the REF ER ENCE S utility of preference assessments in identifying individual- References marked with an asterisk indicate studies included in the ized social reinforcers. However, clinicians report that, review.  19383703, 2023, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/jaba.981 by Cochrane Canada Provision, Wiley Online Library on [26/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 426 MORRIS ET AL. Behavior Analysis Certification Board. (2020). Ethics code. Retrieved *Gray, C. L. (2019). The effects of reinforcer magnitude using social rein- from https://bacb.com/ethics-code/ forcers on skill acquisition for children diagnosed with autism Briggs, A. M., Dozier, C. L., Lessor, A. N., Kamana, B. U., & [Unpublished doctoral dissertation]. The Chicago School of Pro- Jess, R. L. (2019). Further investigation of differential reinforce- fessional Psychology. ment of alternative behavior without extinction for escape- Hagopian, L. P., Long, E. S., & Rush, K. S. (2004). Preference assess- maintained destructive behavior. Journal of Applied Behavior ment procedures for individuals with developmental disabilities. Analysis, 52(4), 956–973. https://doi.org/10.1002/jaba.648 Behavior Modification, 28(5), 668–677. https://doi:10.1177/ *Butler, C., & Graff, R. B. (2021). Stability of preference and reinfor- 0145445503259836 cing efficacy of edible, leisure, and social attention stimuli. Journal Hanley, G. P. (2010). Toward effective and preferred programming: of Applied Behavior Analysis, 54(2), 684–699. https://doi.org/10. A case for the objective measurement of social validity with recipi- 1002/jaba.807 ents of behavior-change programs. Behavior Analysis in Practice, Campanaro, A. M., Vladescu, J. C., Kodak, T., DeBar, R. M., & 3(1), 13–21. https://doi.org/10.1007/BF03391754 Nippes, K. C. (2020). Comparing skill acquisition under varying Hanley, G. P., Iwata, B. A., Lindberg, J. S., & Conners, J. (2003). onsets of differential reinforcement: A preliminary analysis. Jour- Response-restriction analysis: I. Assessment of activity preferences. nal of Applied Behavior Analysis, 53(2), 690–706. https://doi.org/ Journal of Applied Behavior Analysis, 36(1), 47–58. https://doi.org/ 10.1002/jaba.615 10.1901/jaba.2003.36-47 Chen, M. H., Lan, W. H., Hsu, J. W., Huang, K. L., Su, T. P., Hanratty, L. A., & Hanley, G. P. (2021). A preference analysis of rein- Li, C. T., Lin, W. C., Tsai, C. F., Tsai, S. J., Lee, Y. C., forcer variation and choice. Journal of Applied Behavior Analysis, Chen, Y. S., Pan, T. L., Chang, W. H., Chen, T. J., & Bai, Y. M. 54(3), 1062–1074. https://doi:10.1002/jaba.835 (2016). Risk of developing type 2 diabetes in adolescents and *Harper, A. M., Dozier, C. L., Briggs, A. M., de Villegas, S. D., young adults with autism spectrum disorder: A nationwide longi- Ackerlund Brandt, J. A., & Jowett Hirst, E. S. (2021). Preference tudinal study. Diabetes Care, 39(5), 788–793. https://doi.org/10. for and reinforcing efficacy of different types of attention in pre- 2337/dc15-1807 school children. Journal of Applied Behavior Analysis, 54(3), Cividini-Motta, C., & Ahearn, W. H. (2013). Effects of two variations 882–902. https://doi.org/10.1002/jaba.814 of differential reinforcement on prompt dependency. Journal of Heinicke, M. R., Carr, J. E., & Copsey, C. J. (2019). Assessing prefer- Applied Behavior Analysis, 46(3), 640–650. https://doi.org/10.1002/ ences of individuals with developmental disabilities using alterna- jaba.67 tive stimulus modalities: A systematic review. Journal of Applied *Clay, C. J., Bloom, S. E., Slocum, T. A., Samaha, A. L., & Behavior Analysis, 52(3), 847–869. https://doi.org/10.1002/jaba.565 Callard, C. H. (2020). Language preference and reinforcing efficacy Hood, S. A., Gopez, J. M., Fallon, M. J., Byczynski, F. A., of praise in bilingual children with disabilities. Journal of Applied Aquino, S. C., & Monroy, S. (2022). The beginning of a friend- Behavior Analysis, 53(1), 536–544. https://doi.org/10.1002/jaba.609 ship: Teaching individuals with autism to identify shared interests. *Clay, C. J., Samaha, A. L., Bloom, S. E., Bogoev, B. K., & Journal of Applied Behavior Analysis, 55(4), 1030–1058. https:// Boyle, M. A. (2013). Assessing preference for social interactions. doi.org/10.1002/jaba.951 Research in Developmental Disabilities, 34(1), 362–371. https://doi. *Huntington, R., & Schwartz, I. (2022). A preliminary examination of org/10.1016/j.ridd.2012.07.028 social preference across assessors. Behavioral Interventions, 37(2), *Clay, C. J., Samaha, A. L., & Bogoev, B. K. (2018). Assessing prefer- 556–566. https://doi.org/10.1002/bin.1858 ence for and reinforcing efficacy of components of social interac- *Huntington, R. N., & Higbee, T. S. (2018). The effectiveness of a video tion in individuals with autism spectrum disorder. Learning and based preference assessment in identifying social reinforcers. Motivation, 62, 4–14. https://doi.org/10.1016/j.lmot.2017.03.008 European Journal of Behavior Analysis, 19(1), 48–61. https://doi. *Davis, T., Weston, R., Hodges, A., & Gerow, S. (2022). Comparison org/10.1080/15021149.2017.1404397 of picture-and video-presentation preference assessments for social Johnson, K. A., Vladescu, J. C., Kodak, T., & Sidener, T. M. (2017). interactions. Journal of Behavioral Education, 31(2), 367–387. An assessment of differential reinforcement procedures for learners https://doi.org/10.1007/s10864-020-09402-5 with autism spectrum disorder. Journal of Applied Behavior Analy- *Davis, T. N., Hodges, A., Weston, R., Hogan, E., & sis, 50(2), 290–303. https://doi.org/10.1002/jaba.372 Padilla-Mainor, K. (2017). Correspondence between preference Kanaman, N. A., Hubbs, A. L., Dozier, C. L., Jones, B. A., assessment outcomes and stimulus reinforcer value for social inter- Foley, E., & Brandt, J. A. (2022). Evaluating the effects of social actions. Journal of Behavioral Education, 26(3), 238–249. https:// interaction on the results of preference assessments for leisure doi.org/10.1007/s10864-017-9271-x items. Journal of Applied Behavior Analysis, 55(2), 430–450. *DeLeon, I. G., Bullock, C. E., & Catania, A. C. (2013). Arranging https://doi.org/10.1002/jaba.897 reinforcement contingencies in applied settings: Fundamentals and *Kelly, M. A., Roscoe, E. M., Hanley, G. P., & Schlichenmeyer, K. implications of recent basic and applied research. In G. J. Madden, (2014). Evaluation of assessment methods for identifying social W. V. Dube, G. Hanley, T. Hackenberg, & K. A. Lattal (Eds.), reinforcers. Journal of Applied Behavior Analysis, 47(1), 113–135. APA Handbook of Behavior Analysis (Vol. 2, pp. 47–75). https://doi. https://doi.org/10.1002/jaba.107 org/10.1037/13938-033 Kronfli, F. R., Vollmer, T. R., Parks, M. E., & Hack, G. O. (2022). DeLeon, I. G., & Iwata, B. A. (1996). Evaluation of a multiple-stimulus A brief assessment to identify sensitivity to a conversational part- presentation format for assessing reinforcer preferences. Journal ner’s interest. Behavior Analysis in Practice, 15, 838–844. https:// of Applied Behavior Analysis, 29(4), 519–533. https://doi:10.1901/ doi.org/10.1007/s40617-021-00668-3 jaba.1996.29-519 Kunnavatana, S. S., Bloom, S. E., Samaha, A. L., Slocum, T. A., & Fisher, W., Piazza, C. C., Bowman, L. G., Hagopian, L. P., Clay, C. J. (2018). Manipulating parameters of reinforcement to Owens, J. C., & Slevin, I. (1992). A comparison of two approaches reduce problem behavior without extinction. Journal of Applied for identifying reinforcers for persons with severe and profound Behavior Analysis, 51(2), 283–302. https://doi.org/10.1002/jaba.443 disabilities. Journal of Applied Behavior Analysis, 25(2), 491–498. *Lang, R., van der Werff, M., Verbeek, K., Didden, R., Davenport, K., https://doi:10.1901/jaba.1992.25-491 Moore, M., Lee, A., Rispoli, M., Machalicek, W., O’Reilly, M., Graff, R. B., & Karsten, A. M. (2012). Assessing preferences of individ- Sigafoos, J., & Lancioni, G. (2014). Comparison of high and low uals with developmental disabilities: A survey of current practices. preferred topographies of contingent attention during discrete trial Behavior Analysis in Practice, 5(2), 37–48. https://doi:10.1007/ training. Research in Autism Spectrum Disorders, 8(10), bf03391822 1279–1286. https://doi.org/10.1016/j.rasd.2014.06.012  19383703, 2023, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/jaba.981 by Cochrane Canada Provision, Wiley Online Library on [26/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License REVIEW OF SOCIAL PAS 427 *Leaf, J. B., Dale, S., Kassardjian, A., Tsuji, K. H., Taubman, M., Piazza, C. C., Bowman, L. G., Contrucci, S. A., Delia, M. D., McEachin, J. J., Leaf, R. B., & Oppenheim-Leaf, M. L. (2014). Adelinis, J. D., & Goh, H. L. (1999). An evaluation of the proper- Comparing different classes of reinforcement to increase expressive ties of attention as reinforcement for destructive and appropriate language for individuals with autism. Education and Training in behavior. Journal of Applied Behavior Analysis, 32(4), 437–449. Autism and Developmental Disabilities, 49(4), 533–546. https:// https://doi.org/10.1901/jaba.1999.32-437 www.jstor.org/stable/24582349 *Pierce, T. (2019). An evaluation of the consistency and accuracy of Morris, C., Detrick, J. J., & Peterson, S. M. (2021). Participant assent children preference for and reinforcing efficacy of different in behavior analytic research: Considerations for participants with types of attention across different adults [Unpublished doc- autism and developmental disabilities. Journal of Applied Behavior toral dissertation]. The Chicago School of Professional Analysis, 54(4), 1300–1316. https://doi.org/10.1002/jaba.859 Psychology. Morris, S. L., Allen, A. E., & Gallagher, M. L. (2023). Evaluating the Pizarro, E. M., Vollmer, T. R., & Morris, S. L. (2021). Evaluating skills number of sessions in multiple stimulus without replacement pref- correlated with discriminated responding in multiple schedule erence assessments for social interaction. Behavior Analysis: arrangements. Journal of Applied Behavior Analysis, 54(1), Research and Practice. Advance online publication. https://doi.org/ 334–345. https://doi.org/10.1002/jaba.759 10.1037/bar0000264 Pritchett, M., Ala’i-Rosales, S., Cruz, A. R., & Cihon, T. M. (2021). *Morris, S. L., & Vollmer, T. R. (2019). Assessing preference for differ- Social justice is the spirit and aim of an applied science of human ent types of social interaction. Journal of Applied Behavior Analy- behavior: Moving from colonial to participatory research prac- sis, 52(4), 1064–1075. https://doi.org/10.1002/jaba.597 tices. Behavior Analysis in Practice, 15(4), 1–19. https://doi.org/10. *Morris, S. L., & Vollmer, T. R. (2020a). A comparison of methods for 1007/s40617-021-00591-7 assessing preference for social interactions. Journal of Applied Roane, H. S., Vollmer, T. R., Ringdahl, J. E., & Marcus, B. A. (1998). Behavior Analysis, 53(2), 918–937. https://doi.org/10.1002/jaba.692 Evaluation of a brief stimulus preference assessment. Journal of *Morris, S. L., & Vollmer, T. R. (2020b). A comparison of picture and Applied Behavior Analysis, 31(4), 605–620. https://doi.org/10.1901/ GIF-based preference assessments for social interaction. Journal of jaba.1998.31-605 Applied Behavior Analysis, 53(3), 1452–1465. https://doi.org/10. Smaby, K., MacDonald, R. P. F., Ahearn, W. H., & Dube, W. V. 1002/jaba.680 (2007). Assessment protocol for identifying preferred social conse- Morris, S. L., & Vollmer, T. R. (2020c). Evaluating the function of quences. Behavioral Interventions, 22(4), 311–318. https://doi.org/ social interaction using time allocation as a dependent measure: 10.1002/bin.242 A replication and extension. Journal of Applied Behavior Analysis, Stocco, C. S., Saavedra, I., Fakharzadeh, S., Patel, M. R., & 53(4), 2405–2420. https://doi.org/10.1002/jaba.750 Thompson, R. H. (2021). A comparison of intervention for prob- *Morris, S. L., & Vollmer, T. R. (2020d). Evaluating the stability, valid- lematic speech using reinforcement with and without preferred ity, and utility of hierarchies produced by the Social Interaction topics. Journal of Applied Behavior Analysis, 54(1), 217–230. Preference Assessment. Journal of Applied Behavior Analysis, https://doi.org/10.1002/jaba.770 53(1), 522–535. https://doi.org/10.1002/jaba.610 Stokes, T. F., & Baer, D. M. (1977). An implicit technology of generali- Morris, S. L., & Vollmer, T. R. (2021). Evaluating the function of social zation. Journal of Applied Behavior Analysis, 10(2), 349–367. interaction for children with autism. Journal of Applied Behavior https://doi.org/10.1901/jaba.1977.10-349 Analysis, 54(4), 1456–1467. https://doi.org/10.1002/jaba.850 Sundberg, M. L. (2014). VB-MAPP Verbal Behavior Milestones Assess- Morris, S. L., & Vollmer, T. R. (2022a). Comparing clinician-reported ment and Placement Program: A language and social skills assess- hierarchies of relative reinforcer efficacy to reinforcer assessment hier- ment program for children with autism or other intellectual archies. Behavior Analysis: Research and Practice, 22(4), 354–367. disabilities. AVB Press. https://doi.org/10.1037/bar0000257 Tullis, C. A., Cannella-Malone, H. I., Basbigill, A. R., Yeager, A., Morris, S. L., & Vollmer, T. R. (2022b). Increasing social time alloca- Fleming, C. V., Payne, D., & Wu, P. F. (2011). Review of the choice tion and concomitant effects on mands, item engagement, and and preference assessment literature for individuals with severe to pro- rigid or repetitive behaviors. Journal of Applied Behavior Analysis, found disabilities. Education and Training in Autism and Developmental 55(3), 814–831. https://doi.org/10.1002/jaba.919 Disabilities, 46(4), 576–595. http://www.jstor.org/stable/24232368 Morris, S. L., & Vollmer, T. R. (2022c). The matching law provides a *Wolfe, K., Kunnavatana, S. S., & Shoemaker, A. M. (2018). An inves- quantitative description of social time allocation in children with tigation of a video-based preference assessment of social interac- autism. Journal of Applied Behavior Analysis, 55(3), 934–957. tions. Behavior Modification, 42(5), 729–746. https://doi.org/10. https://doi.org/10.1002/jaba.934 1177/0145445517731062 Must, A., Phillips, S. M., Curtin, C., Anderson, S. E., Maslin, M., Lividini, K., & Bandini, L. G. (2014). Comparison of sedentary behaviors between children with autism spectrum disorders and SU P P O R T I N G I N FO R M A T I O N typically developing children. Autism, 18(4), 376–384. https://doi. Additional supporting information can be found online org/10.1177/1362361313479039 *Nuernberger, J. E., Smith, C. A., Czapar, K. N., & Klatt, K. P. in the Supporting Information section at the end of this (2012). Assessing preference for social interaction in children diag- article. nosed with autism. Behavioral Interventions, 27(1), 33–44. https:// doi.org/10.1002/bin.1336 Pace, G. M., Ivancic, M. T., Edwards, G. L., Iwata, B. A., & Page, T. J. How to cite this article: Morris, S. L., Gallagher, (1985). Assessment of stimulus preference and reinforcer value with M. L., & Allen, A. E. (2023). A review of methods profoundly retarded individuals. Journal of Applied Behavior Anal- of assessing preference for social stimuli. Journal of ysis, 18(3), 249–255. https://doi.org/10.1901/jaba.1985.18-249 Paez, A. (2017). Gray literature: An important resource in systematic Applied Behavior Analysis, 56(2), 416–427. https:// reviews. Journal of Evidence-Based Medicine, 10(3), 233–240. doi.org/10.1002/jaba.981 https://doi.org/10.1111/jebm.12265