Evaluation of Statement Accuracy on Ethical Decision-Making

Abstract

The purpose of this study was to examine how different ethical scenarios and different likelihood of statement accuracy influenced recommendations to seek more information or report an ethical violation. Twenty participants were recruited to participate in a pre-workshop survey where they were presented with five hypothetical ethical scenarios that each corresponded to one of five probabilities that scenario was accurate (100%, 60%, 40%, 20%, and 10%). We found when given a scenario that was 100% accurate, 17 of 20 participants indicated they would advise the individual to report the ethical violation. As report accuracy decreased, the proportion of participants that advised to report also decreased. Future directions and implications for ethical decision making in research and practice are discussed.

Introduction

Ethics refers to rules that make claims about right or wrong behavior (e.g., do no harm) as relevant to a group of people connected socially (e.g., profession, religion; Boone, 2017). Ethics are especially important to the field of applied behavior analysis (ABA) because behavior analysts often conduct research or provide direct treatment with vulnerable or at-risk populations (Bailey & Burch, 2022; Brodhead et al., 2022). Applied behavior analysts have, therefore, published many ethics-based resources to guide research and practice-related decisions with these vulnerable or at-risk populations. For example, textbooks have focused on broad ethical issues pertaining to individuals with developmental disabilities (Hayes et al., 1994), punishment and aversive control (Sidman, 2000), general practice (Bailey & Burch, 2022), and research ethics (Cox et al., 2022; Normand & Donohue, 2023). Additional examples include position papers addressing restraint and seclusion (Vollmer et al., 2011), a client’s right to effective behavioral treatment (Van Houten et al., 1988), conversion therapy (Conine et al., 2022), and contingent electric skin shock (Association for Behavior Analysis International, 2022).

Ethics literature in behavior analysis has focused broadly on two areas. One area has focused on theoretically extending basic principles of operant behavior to behavior labeled as ethical (Baum, 2005; Hocutt, 2013; Skinner, 1953, 1971). This area of scholarship has attempted to broadly outline how ethical behavior can be established, maintained, and changed based on principles of operant and respondent conditioning. Authors have argued that ethical behavior is behavior and therefore subject to the same laws and principles as all other behavior (see also Newman et al., 1996).

A second area has focused on ethical issues relating to the delivery of behavior-analytic services. However, only recently have data-based publications begun to emerge in this area. Often these studies are conducted in survey format (e.g., Brown & Brodhead, 2023; Quigley et al., 2019; Witts et al., 2020) or in a translational choice context (e.g., Cox, 2020). Data have also been published about ethical violations reported to the Behavior Analyst Certification Board ([BACB], 2018). These data include information about 219 notices the BACB filed to different individuals during the calendar years 2016 and 2017. To provide context, there was an average of 57,591 Board Certified Behavior Analysts (BCBAs), Board Certified Assistant Behavior Analysts (BCaBAs), or Registered Behavior Technicians (RBTs) during 2016 and 2017. If individuals credentialed by the BACB make an average of only one ethical decision per day then there were 42,041,430 ethical decisions in the same time span. This means we have data on only 0.0005% of all ethical decisions made during that period.

The BACB violations database is important for two reasons. First, it provides the practitioners of ABA information about areas they can focus on to improve. For example, the 2018 report suggested that supervision was the most reported actionable item. As a result, a handful of articles and resources were subsequently published directly addressing supervision (e.g., Lindbald, 2021; Kazemi et al., 2022). If the BACB database will be used to help grow and advance the field of ethical implementation of ABA services, then the quality of the database is important. Second, scholars have suggested that codes of ethics without supporting contingencies of enforcement often fail to influence behavior (e.g., Epstein & Hanson, 2021). For BCBAs, those contingencies are likely maintained primarily through the BACB and the BACB database provides a window into those contingencies in action. Therefore, an open question is what variables control whether someone reports a potential ethical violation to the BACB?

Studying the variables that control reporting ethical violations presents several challenges. First, the traditional approach to studying behavior involves repeated measures of behavior at stability (Sidman, 1960). But the decision to report a violation to the BACB is a one-shot choice within a defined context (i.e., you either choose to report or not). If one chooses not to report, they may seek more information and, once more information is obtained, they are now in a new decision context and again have to make a single choice to report or not. Thus, studying the variables that influence decisions to report an observed behavior to the BACB should involve situations similar to how the behavior occurs in nonlaboratory environments.

A challenge for studying the variables that may control reporting to the BACB is access to real-time, nonlaboratory ethical decision making. Observing a professional’s ethical decision making during a certain length of time (e.g., a day) brings tremendous challenges due to ethical considerations of manipulating real-life ethical variables, and also raises questions about the internal validity of the experiment itself (where the presence of an observer or knowledge of study participation may differentially occasion different behavior). The BACB’s Ethics Code for Behavior Analysts (hereafter referred to as the BACB Code) states that “Behavior analysts are expected to be knowledgeable about and comply with the Code and Code-Enforcement Procedures” (BACB, 2020, p. 5; see also the BACB’s Code-Enforcement Procedures, BACB, 2022). Without access to the BACB’s internal data or the ability to conduct real-time observations, it is difficult to measure the ethical actions of practicing behavior analysts. As a result, most research on ethical decision making presents the participant with different vignettes and scenarios where they indicate what they would do in different situations.

The purpose of this pilot study was to examine how different ethical scenarios and different likelihood of statement accuracy influenced recommendations to seek more information or report a potential violation to the BACB. We chose statement accuracy as an independent variable for this pilot study because reporting an ethical violation to a credentialing body, such as the BACB, requires documented evidence of that violation. The quality of that evidence may then be used by the credentialling body to evaluate the extent to which that evidence aligns with the alleged violation. Further, if a professional raises concern about an ethical violation to a colleague about a third party, the accuracy of that evidence could influence how that colleague advises that professional how to proceed (e.g., “You have presented clear evidence of wrongdoing” vs. “You may believe this is happening, but the facts you have provided do not seem to substantiate your concern”).

A preliminary evaluation of the ethical decision making of behavior analysts, such as individuals credentialed by the BACB, may serve as a starting point for a broad evaluation of ethical behavior and contribute to the growing body of literature that uses translational research to understand applied problems (e.g., Call et al., 2015; Cox & Brodhead, 2021; White-Cascarilla et al., 2023). Individuals credentialed by the BACB are bound to ethical codes associated with those credentials, which are designed to inform ethical behavior and ethical decision making. Though preliminary, we encourage readers to review our method, results, and discussion in the broader context of behavioral research on ethical decision making and consider how our report may serve as one potential framework for refined systematic and direct evaluation of ethical decision making in the future. By identifying variables that may affect how individuals credentialed by the BACB respond to ethical violations, behavior analysts may begin to understand behavioral determinants of ethical decision making and establish a framework for broader, more wide-scale empirical evaluation of ethical decision making.

Method

Participants and Setting

Twenty participants were recruited from a workshop on ethics at a regional conference in Fall 2019 (additional participant and workshop information is purposefully omitted to protect participant anonymity). Prior to starting the workshop, participants were provided with an option to complete a pre-workshop survey. Informed consent was obtained from each participant prior to completing the pre-workshop survey. The mean participant age was 36 years (range: 25–53). Nineteen participants self-identified as female; one self-identified as male. Nineteen participants identified as not Hispanic or Latinx; one identified as Hispanic or Latinx. Nineteen participants reported having a BCBA credential; one participant reported having a BCaBA credential. The mean length of time participants reported having either credential was 4.75 years (range: 8 months–13 years).

Materials

Participants completed a survey on their own personal computer or electronic handheld device (e.g., a phone or tablet). The survey was programmed in Qualtrics (2019) and measured participant decisions when provided with a scenario that described a violation of the BACB’s Professional and Ethical Compliance Code for Behavior Analysts1 (2014). We first identified scenarios in recently published books on ethics and behavior analysis (Bailey & Burch, 2016; Sush & Najdowski, 2019) that we thought clearly described an ethical violation. Each scenario depicted a violation of a different BACB Code Guideline (1.01, 1.05, 1.07, 2.13, and 8.01). Scenarios were revised for clarity and brevity. See Table 1 for each of the five scenarios, which includes a cross reference for the 2014 and 2020 versions of the BACB Code.

Table 1.

Hypothetical Scenarios and Corresponding BACB Code Guidelines and Standards Used In The Study

Scenario	BACB Code
	2014	2020
Anthony is a BCBA who works in the home of a child whose parents strongly believe in FC, or facilitated communication (AKA “supported typing”). Anthony uses a typing machine and physical prompting to guide a child (who is in the third grade and nonverbal) to generate language-related responses. Anthony justifies his work by saying, “I don't look at the screen." A BCBA at the school has tested the student and has demonstrated that he is clearly prompt-dependent and produces no intelligible responses on the device on his own.	1.01	2.01
I work as a consulting BCBA and I provide home-based services for a school district. It has been reported to me (by clients and colleagues) that the school district's BCBA "has a serious drug problem." I interact with him rarely and do not have any evidence of this, though I am concerned that illicit drug use or perceived drug use by consumers and colleagues-is a violation of our ethical guidelines, in that it is (1) in violation of state laws and (2) does not represent our field well.	1.05	1.10
I work at a center where none of the program coordinators who do behavioral treatment are BCBAs or BCaBAs. Nonetheless, they are required to send their paperwork to a BCBA. The BCBA has never seen any of the clients except for an hour or so at intake, knows nothing about their history or behaviors, and never reviews what the program coordinators report. However, the BCBA signs the forms as though he did the work.	1.07	1.13
My daughter was receiving ABA therapy from a BCBA. The therapist was billing our insurance company for services my daughter never received. The BCBA is an in-service provider, and by law, all ABA therapy is covered by the insurance company. I have asked the BCBA for explanations and clarifications, but she has refused me any recourse, telling me to take my daughter home.	2.13	2.06
Zara has a BCBA and also manages her autism agency's marketing and promotional efforts. Because she maintains the agency's public relations image, she often gives interviews to media outlets as a representative of the agency. In her most recent interview, Zara is quoted as comparing ASD and other DDs to diseases such as cancer and endorsing ABA-based treatment strategies as an assured cure.	8.01	5.03

Procedure

Participants were randomly assigned to one of five different conditions. Each condition randomly combined one of five hypothetical ethical scenarios to one of five probabilities that scenario was accurate (100%, 60%, 40%, 20%, and 10%). The probabilities were chosen based on common probabilities published in the decision-making literature (e.g., McKerchar & Renda, 2012). Given the pilot nature of this study, these standard probabilities were chosen in order to maintain congruence with established conventions in probability discounting research. Each scenario included a single decision trial. Participants were exposed to five trials—one for each scenario-probability combination.

After consenting to participate in the study, participants were provided with the following instructions:

For this task, you will be presented with a series of ethical scenarios that depict violations of the Behavior Analyst Certification Board’s Professional and Ethical Compliance Code for Behavior Analysts. On each page, you will be provided with information as it is being reported to you by a colleague, who is a Board Certified Behavior Analyst. You will be asked to make a choice between advising your colleague who provided you with this information to report the individual to the Behavior Analyst Certification Board, or to request that the individual obtain additional information before you can make a determination. All scenarios and choices in this task are hypothetical. But please respond as though this were an actual case and you were making this decision. It is also important to note that there are no right or wrong answers. We are interested in what choice you would make given only the available information on the screen.

Following the instructions, the individual scenarios were presented. Written below each scenario was a statement that indicated the probability that the scenario was accurate (e.g., “There is a 10% probability this report is accurate.”). Participants selected one of two possible responses: (1) advise your colleague to report the individual to the BACB; or (2) request to obtain additional information. Based on our collective experiences in practice and ethical decision making, the two options reflect possible decisions when faced with a report of an ethical violation. Additional options may exist in nonresearch settings, however, we restricted choice to two options for ease of analysis.

Following the participants’ choice with one scenario, the next scenario was presented with a different probability of report accuracy, and participants were asked to choose between the two options. This process continued until each participant completed five ethical scenario tasks with five different percentages of scenario accuracy. Finally, participants completed a short demographic survey.

Data Analysis

The primary dependent variable (DV) was the proportion of participants who advised the colleague to report the individual in the scenario to the BACB. We conducted two analyses to determine how each independent variable (report accuracy, ethical scenario) independently influenced the DV. First, we determined how the probability of report accuracy influenced the DV. This was accomplished by fitting the hyperboloid probability discounting equation to the DV (Rachlin, 2006). Second, we determined how the ethical scenarios influenced the DV. This was accomplished via chi-squared tests for independence followed by follow-up pairwise tests of paired proportions. Finally, we were interested in whether report accuracy and ethical scenario interacted to influence the DV. This analysis was accomplished using multinomial logistic regression. All data analyses were conducted using Python version 3.7.2.

Results

When given a scenario that was 100% accurate, 17 of 20 participants indicated they would advise the individual to make a report to the BACB. As report accuracy decreased, the proportion of participants that advised to report also decreased. In particular, across all scenarios, the proportion of participants that advised their colleague to report the individual to the BACB decreased as the probability decreased that the report was accurate (Fig. 1). The hyperboloid probability discounting equation (Rachlin, 2006) described well the influence of report accuracy on choice (R² = .98; h = 12.86; s = 0.27) and report accuracy was a significant predictor within the multinomial logistic regression analysis (χ² (4, N = 100) = 23.61, p < .001). This indicates that report accuracy may influence ethical decision making and is likely to do so hyperbolically.

Fig. 1.

Proportion of Respondents Who Indicated to Report to the BACB across Statement Accuracies

Across all data accuracies, the proportion of participants that advised their colleague to report the individual to the BACB differed depending on the ethical scenario (χ² (4, N = 100) = 20.90, p < 0.001; Fig. 2). Follow-up pairwise comparisons indicated statistically significant differences in the DV between the scenarios: “guideline 1.05” vs. “guideline 1.07” (χ² (1, n = 40) = 3.68, p = 0.05); “guideline 1.05” vs. “guideline 2.13” (χ² (4, n=40) = 6.42, p = 0.01), and “guideline 8.01” vs. “guideline 2.13” (χ² (4, n = 40) = 3.61, p = 0.05).

Fig. 2.

Proportion of Respondents Who Indicated to Report to the BACB across Hypothetical Scenarios

Finally, the interaction term in the multinomial logistic regression model was statistically significant (χ² (11, N = 100) = 23.45, p = 0.015). This indicates that the influence of report accuracy on the DV depended on the scenario and vice versa.

Discussion

The results of our study suggest that report accuracy affects ethical decision making at the group level. When a scenario was 100% accurate, 85% of participants (n = 17) advised to report to the BACB. As report accuracy decreased, the proportion of participants that advised to report also decreased. Report accuracy may be one variable that affects the ethical decision making of individuals credentialed by the BACB when faced with an ethical violation.

We found that report accuracy and scenario interacted to influence ethical decision-making. For example, when provided with a scenario that described a false or deceptive statement (i.e., Guideline 8.01), all participants indicated to “report” when the statement was 100% accurate (n = 20), though no participants made this selection at lower accuracies (see Fig. 3). However, when provided with a scenario that described a violation of insurance fraud (i.e., Guideline 2.13), all participants advised to “report” at 100% and 60% accuracy, and the proportion of participants who chose to “report” were .60 and .63 at 40% and 10% accuracies, respectively. Further, more participants advised reporting an individual to the BACB with scenarios involving activities that may cause potential harm to clients or the profession of ABA (e.g., Guideline 2.13) compared to personal issues a credentialed behavior analyst might be experiencing (e.g., Guideline 1.05).

Fig. 3.

Group Responding within Each Hypothetical Scenario

It makes logical sense that the details of an ethical scenario would influence ethical decision making. However, aside from statement accuracy, what those additional details are and the extent to which those details affect ethical decision making are unclear and present ripe areas for future research. For example, future research could systematically manipulate probabilities of client harm in the context of various ethical scenarios to evaluate how those changes in probabilities affect ethical decision making. Researchers may systematically manipulate the form of harm and/or who is being harmed (e.g., physical harm to the client vs. minor harm where the victim is unclear) to evaluate the extent to which changes in harm and the recipients of that harm affect ethical decision making. In addition, variables such client age and disability severity, length of service in the profession, reported delay to resolution, and knowledge about ethics are also critical variables to consider exploring in order to identify what factors may affect ethical decision making.

The probabilities we chose to examine (100%, 60%, 40%, 20%, and 10%) are similar to those reported in decision-making research (e.g., Cox & Brodhead, 2021; McKerchar & Renda, 2012). Given the pilot nature of this study, it made sense to follow established precedent in probability discounting. That said, probability of statement accuracy likely exists on a continuum, ranging from 0 to 100% accurate. Therefore, it would be important for future research to use additional probabilities, especially between 60% and 100% in order to further explore and understand ethical decision making.

We used a single paragraph of text to present each ethical scenario. With this limited information, we were surprised at the proportion of participants who advised to report to the BACB rather than request more information. These data suggest an important area for future research is what amount and type of information are needed before individuals credentialed by the BACB feel comfortable reporting information to the BACB. Related to this, though not advisable, it is common for people to request (and receive) advice on ethical issues on social media platforms where only a single paragraph of information is given2 (see O’Leary et al., 2017). Future research could examine how certified behavior analysts make ethical decisions in social media contexts, as well as in more traditional (e.g., face-to-face) contexts.

We acknowledge that reporting an ethical violation to the BACB is often not the only, or best, way to respond to an observed ethical situation. At the time this study was conducted, the BACB Code obligated that BCBAs work to resolve the situation with the person alleged to have made a violation. In addition, the BACB required that additional steps be taken prior to submitting a Notice of Alleged Violation (e.g., attempting to resolve the situation if possible, collecting supporting documentation, and the involvement of legal authorities where appropriate). These forms of resolution still hold today with the BACB’s (2020) most recent iteration of their ethics code.

We used a binary choice format of “report” or “seek more information” as a starting point for research into ethical decision making in ABA. It is worth emphasizing the choice of “report” or “seek information,” identified in part due to their salient differences and hypothesized likelihood of establishing control of participant behavior in this pilot study, should not be interpreted as the only response options for BCBAs facing ethical issues. We urge caution in interpreting our experimental preparation as a model for real-world decision making. Instead, our results provide preliminary support that probabilities, such as statement accuracy, may affect ethical decision-making.

Limitations

This pilot study resembles a stark departure from traditional probability discounting research, and therefore there are several limitations worth highlighting. First, we evaluated decisions to report to the BACB with only five ethical scenarios. These scenarios do not capture the many daily ethical scenarios practicing behavior analysts experience. Future research could evaluate ethical decision making with frequently reported ethical scenarios that result in disciplinary action by the BACB (see BACB, 2018, 2019). Related to this, our study evaluated scenarios that were developed in the context of the BACB’s Professional and Ethical Compliance Code (2014) that has been replaced with an updated version (see BACB, 2021). Though the text of the BACB’s 2021 ethics code (i.e., Ethics Code for Behavior Analysts) has now been updated, the ethical values espoused in that updated version are as relevant now as the time this study was conducted (see BACB, 2020).

Second, the hypothetical nature of the task may have influenced participants’ responding. Participant choice was restricted to two options and choice may have differed if participants were provided with more or different choice options. Also, participants were told that each scenario depicted a true violation (whereas only the BACB, in the case of our study, could identify whether a true violation occurred). The results may have differed had participants been asked to first evaluate the scenario and identify if they believed a true violation had occurred. Future research should also consider using a panel of experts to validate or identify scenarios that are likely ethical violations to reduce experimenter bias and improve external validity of findings. Finally, participants may have responded differently if they were confronted with these scenarios while working with clients or other practitioners. Additional factors such as knowledge of ethics, length of time in the profession, client age, and risk of harm are also factors that likely affect decision making. Each limitation presents an important area for future research.

Third, we analyzed the influence of ethical scenario and statement accuracy on decisions to report using proportions of a group. This was done for practical and logical reasons. It’s unclear that presenting the same scenario repeatedly to participants would have led to different responding within a short time-frame. In addition, specific situations with ethical implications are unlikely to occur more than once. Thus, whereas repeated responding at steady-state is the typical approach used to analyze choice in behavior analysis, this approach fails to capture one-shot decisions common in everyday life. People have been shown to make different decisions with probabilistic outcomes depending on whether they can make repeated choices or get only one-shot (e.g., Cox & Dallery, 2018). Future research on ethical decision making may have more face validity by using one-shot choices and proportions as the dependent variable. Nevertheless, analyzing the proportion of a group does not isolate the behavioral processes of influence (Sidman, 1960).

Fourth, we did not control for the severity of ethical scenarios. The relative severity of ethical violations may account for differences in responding across questions. One potential solution would be to present a survey to BCBAs and ask them to indicate the level of severity of each violation on a Likert scale. This initial measure of severity could then be used within- or between-subjects to research ethical decision-making across varying severity of scenarios.

Finally, as is the case with any scientific article, we hope readers will consume our experimental preparation with a healthy dose of skepticism and consider the extent to which our pilot experiment may or may not appreciate the complexities of every-day ethical decision making. Of course, ethical decision making is more complex than to advise a person to either file a report or to request additional information! We encourage readers to not miss the forest for the trees, and we encourage readers to appreciate how evaluating behavior in a decision context can provide a lens through which behavioral researchers can understand behavioral determinants of operant behavior, including ethical behavior.

Data Availability

All data generated or analyzed during this study are available from the first author upon request.

Declarations

Ethical Approval

Informed consent was obtained for all participants, in accordance with local, state, and federal human subject projection laws and ethical standards.

Conflict of Interest

The first and third authors receive speaker fees to present on the topic of ethics and behavior analysis, and also receive financial compensation for the sale of textbooks they have authored or edited that are cited in this article.