We conducted this study for two primary purposes: (1) to determine whether a direct group SPA procedure would identify a preference hierarchy for groups of students with disabilities and (2) to evaluate whether the group SPA would accurately identify reinforcers for increasing desirable behavior in the context of a group contingency. A special education teacher implemented the procedures, and participants included two classrooms of middle school students with various disabilities (e.g., ASD, SLD). The group SPA used a PS format in which electronic pictorial options (i.e., free time activities) were presented on an interactive whiteboard and students selected their preferred option using online survey software. Students participated in the SPA after receiving brief directions and sessions required about 15 min each (45 min total for 3 sessions). We aimed to validate the group SPA results by alternating group contingency sessions in which high-p. versus low-p consequences were available contingent on students being prepared for class. We also collected data on student responses to a daily social studies quiz and administered a social validity survey.

Our first research question asked whether the group SPA would result in a hierarchy of participants’ preference for potential reinforcers, and a hierarchy emerged for both classes based on the data aggregated from three sessions. In Classroom One, the high-p choice received 17 more votes than the second choice (23.86% versus 21.10% of votes) and 120 more votes than the low-p choice (23.86% versus 4.38% of votes). There was a smaller margin between the two highest-preferred options in Classroom Two (21.31% versus 20.17% votes); however, the overall data pattern was similar to Classroom One with votes for the remaining options clearly distributed across a hierarchy.

These SPA results correspond with previous research on individually administered SPAs indicating that for some individuals, immediate access to the chosen stimulus may not be necessary for identifying preferences (Brodhead et al., 2019; Daly et al., 2009), and electronic pictures can replace tangibles (Brodhead et al., 2016). To our knowledge, this is one of few studies to evaluate a direct group SPA procedure and the first to include students with disabilities. Our procedures differed from previous group SPA research in that we used a PS format rather than a multiple stimulus format. We also did not provide immediate access to the chosen option (as in Layer et al., 2008 and Radley et al., 2019), which allowed us to feasibly include activity-based reinforcers. Finally, we omitted SPAs for individual students given that research by Layer et al. and Radley et al. indicated individual and group results would correspond, and our ultimate goal was to evaluate group SPA results in the context of a group contingency.

Our second research question asked whether student behavior would differ during group contingency sessions in which high-p consequences were available compared to sessions in which low-p consequences were available. Our data suggest a functional relation for both classes, given that the mean level of students prepared for class was elevated in high-p conditions compared to low-p conditions. However, both the high-p and low-p conditions increased desirable behavior compared to baseline. Additionally, our ATD designs were relatively brief and data in the low-p condition demonstrated possible increasing trends. These data patterns suggest the low-p condition may have become more effective over time, perhaps due to its alternation with the high-p condition (i.e., multiple-treatment interference) or due to students contacting the programmed contingencies across both conditions (i.e., completing an academic task instead of putting their heads down).

To our knowledge, this is the first study to attempt validation of group SPA results in the context of a behavioral intervention. Our results suggest that even a low-p consequence may increase desirable behavior during a group contingency, which aligns with systematic reviews that identified overall positive results of group contingencies even without the use of direct SPAs (Maggin et al., 2012; Pokorski et al., 2017). Nonetheless, our results also indicate that teachers may achieve the greatest increases in desirable behavior by identifying a highly preferred consequence to include in a group contingency.

We also collected data on the percentage of correct responses to a daily video quiz in an effort to describe whether study conditions were associated with academic responding (i.e., whether being prepared for class resulted in student learning). Overall, we did not observe clear functional relations between the intervention conditions and quiz performance. It is important to note that the teacher wrote a unique quiz question for every session, which could account for the variability in these data across sessions. Although we aimed to have similar difficulty across questions, we cannot confirm whether that was the case. Student responses may have also been impacted by their level of background knowledge or interest in the daily news video. Future studies could further explore the impact of preparedness on academic responses by better controlling for these factors. Finally, we included a survey to determine whether the intervention procedures were motivating and enjoyable from students’ perspectives. These data indicated that students felt most motivated when the high-p choice was available and most enjoyed the high-p sessions compared to both baseline and low-p sessions.

Readers should note several limitations of this study. First, our group contingency package included several potentially effective components (e.g., BST for the target behavior, multiple daily reminders of expectations, assignment of an instructional activity to students who did not meet expectations), and we cannot be sure which of the components were necessary and sufficient to increase the percentage of students prepared for class. Second, our dependent variable was a brief, discrete behavior. We do not know how the group contingency would have impacted a more effortful, continuous behavior (e.g., engagement throughout the lesson). We chose this dependent variable because it was a common classroom concern and the teacher and instructional assistant could feasibly collect data during instruction; however, researchers may want to include continuous behaviors in future studies. Third, because we did not collect IOA data for Classroom Two during baseline, we cannot ensure the accuracy of data within that phase. Notably, IOA was 100% for nearly all sessions in which we did collect secondary data and we collected sufficient data during the experimental ATD phase for Class Two. Fourth, we did not measure the social validity of the group SPA procedures from students’ perspective (i.e., our survey only included questions related to the group contingency). Future research may investigate whether group SPA procedures like the ones in this study are enjoyable to students. Last, we were unable to complete an ABAB design for Classroom Two due to multiple interruptions (e.g., winter break) during the study timeline, and thus our evaluation of the group contingency in the second class is not as strong as our evaluation in the first class.

Several variables related to the development of group SPAs warrant future investigation. First, researchers should identify the number of group SPA sessions necessary to identify consistent results and whether the number of sessions should vary based on participant characteristics. Although standard PS procedures include only one session (Fisher et al., 1992), we included three sessions administered across three days, which we hypothesized would allow us to identify accurate and stable preferences for extended use during the group contingency. However, our session-by-session analyses suggest that all three sessions may not have been necessary. Notably, we generally identified the same high-p and low-p stimuli across all three sessions, suggesting that the group contingency would have used the same reinforcers if a single session had been conducted. This finding stands in contrast to similar analyses of individually administered MSWO assessments (Conine et al., 2021), which suggest that an aggregate measure of multiple SPA sessions may identify different highest-preferred stimuli than single SPA sessions. Such findings may be related to the differences between group and individual SPA formats, between PS and MSWO formats, or both. These finding should be replicated in future research.

Relatedly, we used results of the group SPA conducted at the beginning of the study during all sessions; future research could compare results of a pre-intervention SPA to a brief SPA included at the beginning of group contingency sessions (as Radley et al. (2019) also suggested). Additionally, researchers should identify whether other direct SPA formats can be effectively adapted for groups of students with disabilities. We chose a PS format because it limited the number of options available at once and allowed the teacher to create a presentation with all slides before the study (i.e., it was feasible). However, an electronic MSWO format is potentially more efficient if the implementor could create one slide with all options and then delete chosen options during the assessment (similar to Brodhead et al., 2016).

Researchers should also continue investigating methods for students to choose an option during group SPA sessions. We used Google Forms to create an anonymous voting method in alignment with previous group SPA research (Layer et al., 2008; Radley et al. 2019; Resetar Volz & Cook, 2009). We hypothesized that a public voting method (e.g., hand raising) would result in some students’ opinion influencing other students. However, we do not know whether this was necessary, and using the Google Form required time to analyze results outside of the sessions. Future research could investigate whether there are differential results during group SPA sessions that use public versus private responses.

This study also adds to the currently limited body of research on independent group contingencies, and our findings point to several avenues of future research on this intervention. For example, we suggest that future researchers consider whether some students need a unique reinforcer during independent group contingency sessions (Chow & Gilmour, 2015; Radley et al., 2019). In this study, 100% of students in Classroom One were prepared during the final high-p sessions, which suggests the same consequence was a reinforcer for all students’ behavior in that class. However, there was no group contingency session in Classroom Two during which 100% of students were prepared. Future researchers may choose to conduct an individual SPA with non-responders, evaluate correspondence of individual SPA results with the group SPA, and evaluate the effectiveness and social validity of including differentiated reinforcers within independent group contingency sessions. Additional research comparing the effectiveness of independent contingencies to interdependent and dependent contingencies is also warranted. Although some evidence suggests they can be equally effective (Groves & Austin, 2017), it is possible that an interdependent contingency would have yielded better outcomes in Classroom Two.

Finally, in future empirical evaluations, we encourage researchers to make procedural considerations to allow teachers to feasibly conduct group SPAs. In this study, a teacher was the primary implementer and data collector; thus, we prioritized her ability to implement procedures when designing them. For example, our use of the PS format aligned with her preference to create one set of materials and analyze results after administration, and the use of Chromebooks and Google Forms further increased the feasibility of data collection. From both an implementation and data collection standpoint, procedures that are feasible for teachers can increase the probability of accurate and sustained implementation in natural settings. Future research should also assess social validity from the implementer’s perspective.