Research Corner: Do Pictorial Preference Assessments Really Work?

Reading Time: 9 Minutes
Janet Lund, PhD, BCBA-D

Dr. Janet B. Lund, PhD, BCBA-D


Alternative Modality Preference Assessments, like pictorial assessments are becoming increasing popular in the literature and in practice. So we thought this would be a good research article to feature. We hope you enjoy it. If this article summary sparks your interest, please pick up your JABA, grab a cup of coffee and geek out! 

The Full Article

Assessing the efficacy of pictorial preference assessments for children with developmental disabilities
Heinicke, M. R., Carr, J. E., Pence, S. T., Zias, D. R. , Valentino, A. L., Falligant, J. M. (2016). Journal of Applied Behavior Analysis, 49, 1-21

Why was this study conducted?

In a survey conducted by Graff and Karsten (2012) 90% of behavior analysts reported using stimulus preference assessments in their practice.  Of those behavior analysts, 80% also reported that a major barrier to conducting preference assessments was lack of time.  In fact, the majority of respondents said they conduct preference assessments less than once per month.  (I don’t want to make assumptions about you or the people you work with, but I can tell you my preferences change way more frequently than that!).

Heinicke and colleagues researched a solution to this problem in the current study and evaluated the feasibility of pictorial preference assessments with children with disabilities.  They did this by completing two experiments: (1) to investigate the role of contingent access to the stimulus (that is, do participants need to be provided access to the item after seeing the picture to make the assessment effective), and (2) to evaluate the effects of schedule thinning if contingent access was required.

Experiment 1 Details (Role of Contingent Access)

Participants:  Three girls and five boys participated in the study (ranging in age from 2-11 years old).

Sessions and Materials:  Sessions were conducted 2-4 days per week and participation length ranged from 2-4 weeks.  Eight edible items were used in the study (based on caregiver and staff responses to the Reinforcer Assessment for Individuals with Severe Disabilities (RAISD) structured interview (Fisher, Piazza, Bowman, & Amari, 1996).  Caregivers were also asked to select one food item that their child was unlikely to consume (the proverbial brussel sprouts, broccoli, or liver examples for most kids), and this item was used as the control stimulus in the Reinforcer Assessments (RA).  Other items used in the sessions included laminated pictures of all food items, a timer, a data sheet (let’s hope it was electronic!), and a pen (Oh, dear!).  During the Reinforcer Assessment sessions, four containers, three pieces of different colored paper, tape, paper clips, and preferred toys were also used.

Design and Procedure:  RAs were evaluated using an alternative treatments design.  The basic experimental procedure went something like this: An RAISD was conducted to determine stimulus items, prerequisite assessments were conducted (which included a picture-to-object matching assessment, an object-to-picture matching assessment, and a pictorial mand assessment), a pictorial stimulus preference assessment (SPA) without contingent access was conducted, and then a Reinforcer Assessment was completed.  If correspondence occurred, participation was terminated.  If correspondence did not occur, Experiment 2 began, in which a pictorial SPA with contingent access was conducted followed by a Reinforcer Assessment.  Once correspondence was demonstrated, schedule thinning began.  Three schedules were used during schedule thinning, variable-ratio 3, variable-ratio 5, and extinction (i.e., no contingent access).

Some details that might interest you, but maybe not . . .

(If you do read this section you might be an ABA nerd . . .  but no judgment here—we are a company full of ABA nerds!)

Experiment 1 Pictorial Preference Assessment Details

Before the first pictorial preference assessment, participants were provided access to each edible item for 30 seconds.  SPAs were conducted using a paired-stimulus preference assessment (Fisher, Piazza, Bowman, Hagopian, Owens, & Slevin, 1992).  Participants were shown two pictures and asked to “pick one.”  After an item was selected, no access was provided and the experimenters presented the next set of pictures.  Selection percentages of each stimulus item were calculated and used for hierarchical rankings of the stimuli.

Experiment 1 Reinforcer Assessment (RA) Details

Reinforcer Assessments were used to determine if the items selected during the pictorial SPA without contingent access functioned as reinforcers and thereby determining if the SPAs were valid.  For most of the participants, a single operant RA using a paper clip transfer task with an additive progressive ratio schedule was used (Wow! That’s a mouthful!).  For two of the participants, a different task (touching a laminated color card attached to the wall) was selected due to the persistence of responding in the paper clip transfer task in the absence of social consequences.

During the paper clip task, the experiments told participants to move paper clips from one container to another if they would like access to the stimulus item associated with the condition.  Items used in the RA conditions included a high preference item determined by the SPA, a low preference item from SPA, and the control item (non-preferred food item identified by the caregiver in the pre-assessment interview).   After participants transferred 3 paper clips, they were provided with 30 seconds of access to the item associated with the condition.  The ratio then increased to VR 5.  Schedule thinning progressed until no access was provided in a 28-trial preference assessment.

Participants were told they could quit at any time.  For one participant, a concurrent-operants RA was used.  During this assessment, three colored containers were placed on the table and each container was associated with one of the three stimuli (High Preference, Low Preference, Control).  When the participant placed a paper clip in one of the containers, the researchers delivered the stimulus item associated with that colored container.  No PR schedule was used, instead items were delivered on a continuous schedule.  Sessions lasted 5 minutes in the concurrent-operants RA.  The dependent variable used during the RAs was the break point (i.e., the highest or last schedule value completed).

What were the results of Experiment 1?

Three of the eight participants (37.5%) had high correspondence between the pictorial SPAs without contingent access and the subsequent Reinforcer Assessments, thereby suggesting the SPAs for these participants produced valid results by confirming the predictions of the pictorial SPA without contingent access.

Five of the eight participants (62.5%) had low correspondence between the SPAs without contingent access and subsequent RAs, suggesting the predictions of the pictorial preference assessment without contingent access were not valid.  The researchers therefore conducted a pictorial SPA WITH contingent access for these participants, and correspondence was then demonstrated.  The problem with this situation is that by requiring access to the stimuli to demonstrate correspondence, several of the main benefits of conducting pictorial SPAs were eliminated (brevity and ease of administration).

The purpose of Experiment 2 was to investigate the role of contingent access to the reinforcer for those participants who did not demonstrate correspondence between the pictorial SPA without access and the subsequent RA.  The authors evaluated the effects of schedule thinning to determine if the pictorial SPA could be made more practical for these participants.

Experiment 2 (Role of Schedule Thinning)

Participants:  One girl and four boys participated in the second experiment.  These participants were the ones whose SPA without access did not demonstrate correspondence to the subsequent RA in experiment 1.

Sessions and Materials:  The materials used and session arrangements were consistent with experiment 1.  Sessions were conducted 2 to 4 days a week and participation length ranged from 4 to 8 weeks.

Design and Procedure:  An alternating treatments design was used to during the RAs and a nonconcurrent multiple baseline design across participants was used during schedule thinning.  The trials for the pictorial SPA with contingent access were conducted in the same manner as the pictorial SPA without contingent access, except that participants were given access to a small amount of the edible stimulus after selection.  A single-operant RA with progressive ratio was conducted after the SPA with contingent access using the same three items (high preference, low preference, and control).  Schedule thinning was conducted with participants who had high levels of correspondence between the results of the pictorial SPA with contingent access and the RA in order to determine if conditional reinforcement properties could be used to create more practical pictorial SPAs

What did Schedule Thinning look like?

Three SPAs with contingent access were conducted using a variable-ratio (VR) 3 schedule.  Experimenters looked for a clear gradient (defined as each SPA producing selection percentages between 14-86%).  They also looked for consistent gradients which were defined as all three SPAs identifying the same item as the high preference item and the low preference item.  If break points in the RA were highest for the high preference item, the schedule was then thinned to VR 5 in the next 3 pictorial SPAs with contingent access.  Schedule thinning continued until no access was provided in the 28-trial SPA (i.e., extinction).

What were the results of Experiment 2?

The researchers found that the SPAs without contingent access were shorter (mean of 6 minutes and 52 seconds), as compared to SPAs with access (mean of 13 minutes and 59 seconds).   Schedule thinning with the SPAs with contingent access produced valid assessment results but obviously increased the time involved and thereby decreased the utility of alternative modality SPAs.  Two of the 5 participants (40%) showed great success with schedule thinning and only required one set of SPAs at each of the 3 schedule values (i.e., VR 3, VR 5, and extinction) before they were able to make meaningful choices in pictorial SPAs without access.  One participant (20%) required a second set of SPAs at the VR 3 value and one participant (20%) required a second set of SPAs at the VR 5 value.  The final participant (20%) did not have success at the VR 3 schedule so the experimenters implemented a denser VR 2 schedule.  He also displayed inconsistent results during the VR 5 schedule.

Bottom Line

The authors concluded pictorial preference assessments without contingent access were successful for fewer than half of the participants.  They also found that providing contingent access created an effective SPA however they noted that by doing this it negated the main benefit of alternative modality preference assessment.   They also concluded that schedule thinning was a successful strategy and viable option for those of us who have clients who require contingent access during SPAs.  The authors concluded by encouraging future research into pictorial preference assessments and encouraged addition investigation into prerequisite skills assessments and their relationship with alternative modality preference assessments.

ReferencesFisher, W. W. , Piazza, C. C. , Bowman, L. G. , & Amari, A. (1996). Integrating caregiver report with a systematic choice assessment to enhance reinforcer identification. American Journal on Mental Retardation,101, 15–25.

Fisher, W. W. , Piazza, C. C. , Bowman, L. G. , Hagopian, L. P. , Owens, J. C. , & Slevin, I. (1992). A comparison of two approaches for identifying reinforcers for persons with severe and profound disabilities. Journal of Applied Behavior Analysis25, 491–498. doi: 10.1901/jaba.1992.25-491

Graff, R. B. , & Karsten, A. M. (2012). Assessing preferences of individuals with developmental disabilities: A survey of current practices. Behavior Analysis in Practice,5, 37–48.



Related Posts