Human service practitioners from varying fields make ethical decisions daily. At some point during their careers, many behavior analysts may face ethical decisions outside the range of their previous education, training, and professional experiences. To help practitioners make better decisions, researchers have published ethical decision-making models; however, it is unknown the extent to which published models recommend similar behaviors. Thus, we systematically reviewed and analyzed ethical decision-making models from published peer-reviewed articles in behavior analysis and related allied health professions. We identified 55 ethical decision-making models across 60 peer-reviewed articles, seven primary professions (e.g., medicine, psychology), and 22 subfields (e.g., dentistry, family medicine). Through consensus-based analysis, we identified nine behaviors commonly recommended across the set of reviewed ethical decision-making models with almost all (n = 52) models arranging the recommended behaviors sequentially and less than half (n = 23) including a problem-solving approach. All nine ethical decision-making steps clustered around the ethical decision-making steps in the Ethics Code for Behavior Analysts published by the Behavior Analyst Certification Board (2020) suggesting broad professional consensus for the behaviors likely involved in ethical decision making.
Keywords: behavior analysis, choice, decision making, ethical behaviorEthical decision making is operant behavior involving a behavior chain of complex responses (Marya et al., 2022). As behavior analysts, we make difficult ethical decisions daily. Behavior analysts are typically taught to respond to ethical scenarios via vignettes or descriptions of real-world ethical dilemmas (e.g., Bailey & Burch, 2016; Sush & Najdowski, 2019). However, the variability in ethical dilemmas that behavior analysts contact can be extensive and often contains contextual information not included in past training. Such contextual variables (e.g., impact of and on stakeholders, organizational variables, perspective of the funding source) might alter one’s course of action. Ethical decision-making models can equip behavior analysts with the needed tools to navigate varied and complex dilemmas. Thus, behavior analysts can benefit from models that allow an analysis of contextual variables because those variables often impact solutions.
Ethical conduct of board certified behavior analysts is governed by the Behavior Analyst Certification Board (BACB) ethical codes. Since its inception, the BACB has disseminated three major codes—Guidelines for Responsible Conduct for Behavior Analysts (BACB, 2004, 2010), the Professional and Ethical Compliance Code for Behavior Analysts (BACB, 2014), and most recently the Ethics Code for Behavior Analysts (BACB, 2020). Although versions prior to 2020 outlined specific ethical obligations and provided a framework and reference for considering paths of action when confronted with ethical challenges, no ethical decision-making tool was embedded until the most recent Code iteration.
Within applied behavior analysis (ABA), several ethical decision-making models have been published to guide behavior analysts to make optimal decisions (BACB, 2020; Bailey & Burch, 2013, 2022; Brodhead, 2015; Brodhead, Quigley, & Wilczynski, 2018; Newhouse-Oisten et al., 2017; Rosenberg & Schwartz, 2019; Sush & Najdowski, 2019). These models unanimously share the common goal of providing readers with a systematic approach to ethical decision making, yet include unique elements that provide varying contextual recommendations. Some models offer a generalizable approach affording wider applicability to a variety of ethical situations (BACB, 2020; Bailey & Burch, 2013, 2016, 2022; Brodhead et al., 2018; Rosenberg & Schwartz, 2019; Sush & Najdowski, 2019), and other models provide guidance to navigate specific ethical situations (Brodhead, 2015; Newhouse-Oisten et al., 2017). Moreover, some models incorporate a problem-solving approach wherein multiple behaviors are considered along with their possible outcomes to aid decision making in ethical contexts (Rosenberg & Schwartz, 2019).
Existing models within the behavior analytic literature have all emerged in the last 7 years and offer a discipline-specific approach. However, many other allied disciplines (e.g., medicine, psychology) have published literature offering models for ethical decision making for a longer period than the field of behavior analysis. Recently, there have been calls to action where behavior analysts have been looking to and learning from related professions (LaFrance et al., 2019; Miller et al., 2019; Pritchett et al., 2021; Taylor et al., 2019; Wright, 2019). Learning from other disciplines may help the field of behavior analysis rule out ineffective approaches or derive novel effective solutions more quickly.
The purpose of this systematic literature review was to conduct a descriptive analysis of ethical decision-making models across behavior analysis and allied disciplines. This literature review aimed to identify similarities and differences in approaches to ethical decision making that could inform future ethical decision-making models and aid the development of ethical decision-making skills in behavior analysts.
Articles included in this systematic review met the following three criteria: published in peer-reviewed journals through June 2020, written in English, and the title or abstract included keywords from the search (described below). We began the review in July 2020 and completed it in August 2021.
We conducted a systematic review of the literature on ethical decision-making models for the fields of applied behavior analysis, education, medicine, occupational therapy, psychology, social work, and speech language pathology using the Preferred Reporting Item for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Moher, Liberati, Tetzlaff, Altman, & Prisma Group, 2009). We chose these fields because of their similarities to behavior analysis’ mission in serving vulnerable populations. The following procedures were completed in accordance with the PRISMA guidelines: (1) potential articles meeting inclusion criteria were identified; (2) the identified articles were comprehensively screened; (3) the eligibility of each article was evaluated across dependent measures; and (4) the included articles were analyzed.
The first and second authors completed primary database searches using PsycINFO and PubMed. The keywords used to identify potential articles to be included in this analysis were: applied behavior analysis, clinical psychology, counseling psychology, decision mak*, educat*, ethic*, model, medicine, nursing, occupational therapy, speech and language*, and social work. In particular, the key words “ethic*”, “decision mak*”, and “model” were used in combination with the terms “applied behavior analysis,” or “clinical psychology,” or “counseling psychology,” or “medicine,” or “nursing,” or “occupational therapy,” or “speech,” or “language.”
The initial PsycINFO and PubMed searches yielded 635 articles. Of these, 46 were duplicates. The titles and abstracts of the remaining 589 articles were read by the first and second authors to evaluate the inclusion of keywords. Full-text articles were retrieved for studies that included the words ethics or ethical, decision making, or model in their abstracts or titles (n = 249). Of these, a total of 173 articles were selected for full-text review.
The articles selected for full-text review (n = 173) were read in their entirety to evaluate whether they met these criteria: (1) included humans as the population of interest; (2) mentioned decision making; (3) mentioned ethics; (4) provided at least three identifiable steps to be followed as a part of a model in either a text or figure format; and (5) the provided model addressed how to respond to ethical dilemmas. The first and second authors scored each of the 173 articles across the aforementioned criteria to determine whether they would be included in the final analysis. Articles (n = 27) for which it was unclear whether they met any of the criteria were coded as needing additional review, and the third and fourth authors completed an additional full-text review to determine whether they would be included in the final analysis. A total of 126 articles were removed for not meeting all five of the criteria. Thus, 47 articles remained to be included in the analysis.
Next, the first and second authors conducted a manual search (i.e., identification through other sources) of the references (n = 1,354) for the remaining 47 articles. The screening criteria for this search was identical to the initial screening in which the title and abstract were searched for the inclusion of the words ethics or ethical, decision making, and model. Seventy-nine additional articles were identified through this process. Of these 79 articles, 16 were identified as duplicates from the initial PsycINFO and PubMed searches. Twelve articles were inaccessible to us online or through available library loans and were thus excluded. A list of these articles is not included in this article but is available upon request. Upon reviewing the full text of the remaining 51 articles, 26 additional articles met eligibility to be included in the analysis. In sum, a total of 60 articles met all inclusion criteria and were included.
Interrater reliability was scored using a consensus-based approach. In particular, all four authors collaboratively scored each of the models across the various measures described in the section below. If there was disagreement on scoring at any point, the authors collaboratively reviewed the model using figures provided within the article and any available text describing the model until consensus in scoring was reached.
Articles that met criteria for inclusion were evaluated across four dependent measures. First, we evaluated the steps included within the models from each article. Second, we categorized the model by the professional discipline or field of study. Third, we evaluated whether the model author presented the model in a specific order or sequence (i.e., linear or sequential model). Lastly, we scored whether the model included a problem-solving approach. We provide greater detail on each of these dependent measures below.
The models from each article were evaluated across nine steps (Table (Table1). 1 ). These steps were developed during the process of data synthesis. We read the included articles and identified common themes based on their prevalence in the examined literature. Next, we began classifying articles by the inclusion of these steps, indicating whether each article contained each of the identified steps. Then, we began tracking additional steps that appeared in articles. If those steps appeared in multiple articles, we added them as official steps in the analysis. When this was done, all previously coded articles were recoded for these additional steps. For the purpose of the current review, we identified the following nine components of ethical decision making: (1) ethical radar; (2) urgent detour; (3) pinpoint the problem; (4) information gathering; (5) available options/behaviors; (6) ranking and weighing; (7) analysis; (8) implementation; and (9) follow-up. Details on scoring criteria for each of these steps can be found in Appendix Table Table4. 4 . We scored models included in each article as either including or not including the steps listed above. This was done by using the text description of the model, if provided, or the figure representation of the model if descriptive text was not included.
Steps from the Decision-Making Model from the Ethics Code for Behavior Analysts (2020) and from the Current Literature Review
4. Information gathering (Overall information related to the case that should be considered in making a decision. For example, is the person a minor?)
4a. Affected parties (Any language that mentions people involved or how actions would impact them).
4b. Reference professional code of ethics.
4c. Reference other codes of ethics (personal, religious, organizational).
4d. Case specific information (Catch all for all other information).
*Step 4 of the BACB model aligns with components from Step 6 of current literature review.
This step was coded when the model author(s) recommended gathering contextually relevant information that would be needed to make an ethical decision. The information collected was further divided into the following subcategories where appropriate:
a. Affected parties: This step was coded if the model author(s) included any language that mentioned different people involved in the situation or how the situation might impact different parties. For example, if parents, teachers, or other affected individuals are relevant to the ethical dilemma or decision.
b. Reference to the relevant professional code(s) of ethics: This step was coded if the model author(s) guided the model users to follow their professional code of ethics.
c. Reference to other codes of ethics (personal, religious, organizational): This step was coded if the model author(s) guided the model users to follow other codes of ethics that differ from the code of ethics from their professional affiliation(s). For example, if the practitioner is prompted to refer to the rules and regulations specific to their organization, or a reference is made to their religious or personal values.
d. Case-specific information: This step was coded if the model author(s) referenced any other information that might be specific to the situation but was not captured in the other subcategories listed above. For example, issues of client preferences, quality of life, contexts and settings, and assessment of the practitioners’ understanding of the circumstances all fell into this category.
The field of study of each article was recorded (e.g., psychology). Where possible, we also included a secondary field of study (e.g., school psychology). The primary field of study of the article was determined based on the journal that it was published in and the intended audience of the article. Secondary fields of study were coded to further gather information about the specific subfield. For example, if the article was published in a psychology journal and the audience of the article was specifically school psychologists.
Models within each article were scored as including a problem-solving component or approach if the model author(s) guided the model users to identify two or more possible solutions and likely outcomes or consequences to the possible solutions. Models that did not include more than one possible solution and did not anticipate outcomes to solutions were scored as not including a problem-solving component.
We coded whether the proposed model was linear or sequential in nature. That is, the model author(s) indicated that steps in the model followed a certain order or sequence wherein each preceding step in the model was to be considered prior to moving on to subsequent steps. If a model was not linear or sequential, this was also recorded.
A total of 55 ethical decision-making models across 60 peer-reviewed journal articles were analyzed. Models included in more than one article were counted as duplicates, and papers that included more than one model resulted in each unique model being coded.
Table Table2 2 shows the number of models that included each of the nine steps. None of the steps were present in all models and the step that was included in the greatest number of models was ranking and weighing information (n = 51; 93%). After ranking and weighing information, the steps found in the most-to-least number of models were: affected parties and available options/behaviors (n = 49; 89%); reference other codes of ethics (e.g., personal, religious, organizational; n = 44; 80%); analysis (n = 43; 78%), reference of professional codes (n = 40; 73%); case specific information (n = 38; 69%); implementation and pinpoint the problem (29 models each; 52%); follow up (n = 26; 47%); ethical radar (n = 21; 38%); urgent detour (n = 16; 29%); and, information gathering (n = 11; 20%).
Steps Included in Each Model
| Steps | No. of models (%) | Models |
|---|---|---|
| Ethical radar (Identify that something doesn't feel right) | 21 (38%) | Boccio, 2021; Bommer et al., 1987; Cassells et al., 2003; Cassells & Gaul, 1998; Christensen, 1988; DeWolf, 1989; Duff & Passmore, 2010; Ehrich et al., 2011; Fan, 2003; Forester-Miller & Davis, 1996; Grundstein-Amado, 1991; Hayes, 1986; Heyler et al., 2016; Hill et al., 1998; Hough, 2008; Kaldjian et al., 2005; Kanoti, 1986; Kirsch, 2009; Macpherson et al., 2020; Ponterotto & Reynolds, 2017; Zeni et al., 2016 |
| Urgent detour (Legal or reporting issue that needs to be addressed immediately) | 16 (29%) | Boccio, 2021; Bolmsjö, Sandman, & Andersson., 2006b; Bommer et al., 1987; Candee & Puka, 1984 (Deontology); Cassells et al., 2003; Cassells & Gaul, 1998; DeWolf, 1989; Ehrich et al., 2011; Fan, 2003; Forester-Miller & Davis, 1996; Greipp, 1997; Hill et al., 1998; Hughes & Dvorak, 1997; Sileo & Kopala, 1993; Soskolne, 1991; Tymchuk, 1986 |
| Pinpoint the problem (Specify the exact issue) | 29 (53%) | Boccio, 2021; Bolmsjö et al., 2006b; Bommer et al., 1987; Christensen, 1988; Fan, 2003; Green & Walker, 2009; Grundstein-Amado, 1991; Haddad, 1996; Harasym et al., 2013; Hill et al., 1998; Hough, 2008; Johnsen et al., 2020; Johnson et al., 2017; Jones, 1991; Kaldjian et al., 2005; Kanoti, 1986; Kirsch, 2009; Laletas, 2018; Liang et al., 2017; Marco et al., 2011; Murphy & Murphy, 1976; Park, 2012; Phillips, 2006; Shahidullah et al., 2019; Soskolne, 1991; Sullivan & Brown, 1991; Toren & Wagner, 2010; Tsai & Harasym, 2010; Zeni et al., 2016 |
| Information gathering | 11 (20%) | Cassells et al., 2003; DeWolf, 1989; Ehrich et al., 2011; Harasym et al., 2013; Hayes, 1986; Hough, 2008; Hughes & Dvorak, 1997; Jones, 1991; Sileo & Kopala, 1993; Tsai & Harasym, 2010; Tymchuk, 1986 |
| Affected parties (Any language that mentions people involved or how actions would impact them). | 49 (89%) | Boccio, 2021; Bolmsjö et al., 2006b; Bommer et al., 1987; Candee & Puka, 1984 (Deontology); Candee & Puka, 1984 (Utilitarian); Cassells et al., 2003; Cassells & Gaul, 1998; Christensen, 1988; Cottone, 2001; du Preez & Goedeke, 2013; Duff & Passmore, 2010; Fan, 2003; Ferrell et al., 1991; Forester-Miller & Davis, 1996; Green & Walker, 2009; Greipp, 1997; Grundstein-Amado, 1991; Haddad, 1996; Harasym et al., 2013; Hayes, 1986; Heyler et al., 2016; Hill et al., 1998; Hough, 2008; Hughes & Dvorak, 1997; Hundert, 2003; Johnsen et al., 2020; Johnson et al., 2017; Jones, 1991; Kaldjian et al., 2005; Kanoti, 1986; Kirsch, 2009; Laletas, 2018; Liang et al., 2017; Macpherson et al., 2020; Murphy & Murphy, 1976; Nekhlyudov et al., 2009; Phillips, 2006; Park, 2012; Ponterotto & Reynolds, 2017; Schaffer et al., 2000; Schneider & Snell, 2000; Siegler, 1982; Shahidullah et al., 2019; Sileo & Kopala, 1993; Soskolne, 1991; Sullivan & Brown, 1991; Tsai & Harasym, 2010; Tunzi & Ventres, 2018; Tymchuk, 1986; |
| Reference professional code of ethics | 40 (73%) | Boccio, 2021; Bolmsjö et al., 2006b; Cassells et al., 2003; Cassells & Gaul, 1998; Christensen, 1988; Cottone, 2001; DeWolf, 1989; du Preez & Goedeke, 2013; Duff & Passmore, 2010; Ehrich et al., 2011; Fan, 2003; Forester-Miller & Davis, 1996; Green & Walker, 2009; Greipp, 1997; Haddad, 1996; Harasym et al., 2013; Hayes, 1986; Heyler et al., 2016; Hill et al., 1998; Hough, 2008; Hughes & Dvorak, 1997; Johnsen et al., 2020; Kaldjian et al., 2005; Kirsch, 2009; Laletas, 2018; Liang et al., 2017; Macpherson et al., 2020; Marco et al., 2011; Park, 2012; Phillips, 2006; Ponterotto & Reynolds, 2017; Schaffer et al., 2000; Schneider & Snell, 2000; Shahidullah et al., 2019; Siegler, 1982; Sileo & Kopala, 1993; Soskolne, 1991; Sullivan & Brown, 1991; Toren & Wagner, 2010; Tsai & Harasym, 2010 |
| Reference other codes of ethics (personal, religious, organizational). | 44 (80%) | Boccio, 2021; Bolmsjö et al., 2006b; Bommer et al., 1987; Candee & Puka, 1984 (Deontology); Cassells et al., 2003; Cassells & Gaul, 1998; Christensen, 1988; Cottone, 2001; du Preez & Goedeke, 2013; Duff & Passmore, 2010; Ehrich et al., 2011; Fan, 2003; Ferrell et al., 1991; Forester-Miller & Davis, 1996; Garfat & Ricks, 1995; Green & Walker, 2009; Greipp, 1997; Haddad, 1996; Harasym et al., 2013; Hayes, 1986; Heyler et al., 2016; Hill et al., 1998; Hough, 2008; Hundert, 2003; Johnson et al., 2017; Jones, 1991; Kaldjian et al., 2005; Kirsch, 2009; Laletas, 2018; Liang et al., 2017; Macpherson et al., 2020; Marco et al., 2011; Nekhlyudov et al., 2009; Park, 2012; Phillips, 2006; Schaffer et al., 2000; Schneider & Snell, 2000; Shahidullah et al., 2019; Sileo & Kopala, 1993; Sullivan & Brown, 1991; Toren & Wagner, 2010; Tsai & Harasym, 2010; Tymchuk, 1986; Zeni et al., 2016; |
| Case specific information (catch all for all other information) | 38 (69%) | Bommer et al., 1987; Candee & Puka, 1984 (Deontology); Cassells et al., 2003; Cassells & Gaul, 1998; Christensen, 1988; Cottone, 2001; DeWolf, 1989; Ehrich et al., 2011; Ferrell et al., 1991; Forester-Miller & Davis, 1996; Greipp, 1997; Grundstein-Amado, 1991; Haddad, 1996; Harasym et al., 2013; Hayes, 1986; Hughes & Dvorak, 1997; Hundert, 2003; Johnsen et al., 2020; Johnson et al., 2017; Jones, 1991; Kaldjian et al., 2005; Kanoti, 1986; Laletas, 2018; Liang et al., 2017; Murphy & Murphy, 1976; Nekhlyudov et al., 2009; Park, 2012; Phillips, 2006; Ponterotto & Reynolds, 2017; Schneider & Snell, 2000; Shahidullah et al., 2019; Siegler, 1982; Sileo & Kopala, 1993; Soskolne, 1991; Sullivan & Brown, 1991; Tsai & Harasym, 2010; Tunzi & Ventres, 2018; Zeni et al., 2016 |
| Available options / behaviors (anything that places constraints of behavioral alternatives) | 49 (89%) | Boccio, 2021; Bolsmjö et al., 2006b; Candee & Puka, 1984 (Deontology); Candee & Puka, 1984 (Utilitarian); Cassells et al., 2003; Cassells & Gaul, 1998; Christensen, 1988; Cottone, 2001; DeWolf, 1989; du Preez & Goedeke, 2013; Duff & Passmore, 2010; Fan, 2003; Ferrell et al., 1991; Forester-Miller & Davis, 1996; Garfat & Ricks, 1995; Greipp, 1997; Grundstein-Amado, 1991; Harasym et al., 2013; Hayes, 1986; Heyler et al., 2016; Hill et al., 1998; Hough, 2008; Hughes & Dvorak, 1997; Hundert, 2003; Johnsen et al., 2020; Johnson et al., 2017; Jones, 1991; Kaldjian et al., 2005; Kanoti, 1986; Kirsch, 2009; Laletas, 2018; Liang et al., 2017; Macpherson et al., 2020; Marco et al., 2011; Murphy & Murphy, 1976; Nekhlyudov et al., 2009; Park, 2012; Phillips, 2006; Ponterotto & Reynolds, 2017; Schaffer et al., 2000; Schneider & Snell, 2000; Shahidullah et al., 2019; Siegler, 1982; Sileo & Kopala, 1993; Soskolne, 1991; Toren & Wagner, 2010; Tsai & Harasym, 2010; Tunzi & Ventres, 2018; Tymchuk, 1986 |
| Ranking / weighing of information (If they are talking about clarifying value and identifying and applying guidelines; risk-benefit analysis; any inclusion of learned history and its influence.) | 51 (93%) | Boccio, 2021; Bolsmjö et al., 2006b; Bommer et al., 1987; Candee & Puka, 1984 (Deontology); Candee & Puka, 1984 (Utilitarian); Cassells et al., 2003; Cassells & Gaul, 1998; Christensen, 1988; Cottone, 2001; du Preez & Goedeke, 2013; Duff & Passmore, 2010; Ehrich et al., 2011; Fan, 2003; Ferrell et al., 1991; Forester-Miller & Davis, 1996; Garfat & Ricks, 1995; Green & Walker, 2009; Greipp, 1997; Grundstein-Amado, 1991; Haddad, 1996; Harasym et al., 2013; Hayes, 1986; Heyler et al., 2016; Hill et al., 1998; Hughes & Dvorak, 1997; Hundert, 2003; Johnsen et al., 2020; Johnson et al., 2017; Jones, 1991; Kaldjian et al., 2005; Kanoti, 1986; Kirsch, 2009; Laletas, 2018; Liang et al., 2017; Macpherson et al., 2020; Marco et al., 2011; Murphy & Murphy, 1976; Nekhlyudov et al., 2009; Park, 2012; Phillips, 2006; Ponterotto & Reynolds, 2017; Schaffer et al., 2000; Schneider & Snell, 2000; Shahidullah et al., 2019; Siegler, 1982; Soskolne, 1991; Sullivan & Brown, 1991; Tsai & Harasym, 2010; Tunzi & Ventres, 2018; Tymchuk, 1986; Zeni et al., 2016 |
| Analysis (synthesizing the steps from above to make a decision) | 43 (78%) | Bolsmjö et al., 2006b; Bommer et al., 1987; Candee & Puka, 1984 (Utilitarian); Cassells et al., 2003; Cassells & Gaul, 1998; Christensen, 1988; Cottone, 2001; du Preez & Goedeke, 2013; Duff & Passmore, 2010; Ehrich et al., 2011; Fan, 2003; Ferrell et al., 1991; Forester-Miller & Davis, 1996; Green & Walker, 2009; Grundstein-Amado, 1991; Haddad, 1996; Harasym et al., 2013; Heyler et al., 2016; Hill et al., 1998; Hughes & Dvorak, 1997; Hundert, 2003; Johnsen et al., 2020; Johnson et al., 2017; Jones, 1991; Kaldjian et al., 2005; Kanoti, 1986; Kirsch, 2009; Laletas, 2018; Macpherson et al., 2020; Murphy & Murphy, 1976; Nekhlyudov et al., 2009; Park, 2012; Phillips, 2006; Ponterotto & Reynolds, 2017; Schaffer et al., 2000; Shahidullah et al., 2019; Soskolne, 1991; Sullivan & Brown, 1991; Toren & Wagner, 2010; Tsai & Harasym, 2010; Tunzi & Ventres, 2018; Tymchuk, 1986; Zeni et al., 2016 |
| Implementation (Carry out the solution; action) | 29 (53%) | Bolsmjö et al., 2006b; Cassells & Gaul, 1998; Christensen, 1988; DeWolf, 1989; du Preez & Goedeke, 2013; Duff & Passmore, 2010; Ehrich et al., 2011; Ferrell et al., 1991; Forester-Miller & Davis, 1996; Garfat & Ricks, 1995; Haddad, 1996; Harasym et al., 2013; Heyler et al., 2016; Hill et al., 1998; Hough, 2008; Jones, 1991; Kanoti, 1986; Kirsch, 2009; Laletas, 2018; Macpherson et al., 2020; Murphy & Murphy, 1976; Park, 2012; Phillips, 2006; Ponterotto & Reynolds, 2017; Soskolne, 1991; Sullivan & Brown, 1991; Toren & Wagner, 2010; Tsai & Harasym, 2010; Tymchuk, 1986 |
| Follow up (post-implementation evaluation) | 26 (47%) | Bolsmjö et al., 2006b; Bommer et al., 1987; Cassells & Gaul, 1998; Christensen, 1988; DeWolf, 1989; du Preez & Goedeke, 2013; Ferrell et al., 1991; Forester-Miller & Davis, 1996; Garfat & Ricks, 1995; Harasym et al., 2013; Heyler et al., 2016; Hill et al., 1998; Hough, 2008; Johnsen et al., 2020; Kanoti, 1986; Kirsch, 2009; Liang et al., 2017; Macpherson et al., 2020; Murphy & Murphy, 1976; Park, 2012; Phillips, 2006; Ponterotto & Reynolds, 2017; Soskolne, 1991; Sullivan & Brown, 1991; Toren & Wagner, 2010; Tymchuk, 1986 |
Figure Figure1 1 shows a stacked bar chart of the primary and secondary fields of the ethical decision-making models. Medicine dominated the resulting set of models, followed by psychology, education, business, then child and youth care and organizational behavior management (OBM). Nevertheless, 23 different subspecialties were represented in the secondary field of the ethical decision-making models.
Stacked-Bar Graph Showing the Number of Ethical Decision-Making Models Based on the Primary and Secondary Literatures from which It Came
Table Table3 3 presents a list of the synthesized models and their respective fields of study. The most common field of study across the 55 models was medicine (n = 34; 62%). Seventeen of the models from medicine were specific to the subfield of nursing (50%) and three were specific to the subfield of psychiatry (9%). Of the remaining models from the field of medicine, one each was specific to critical care (3%), dentistry (3%), emergency medicine (3%), geriatrics (3%), internal medicine (3%), and oncology (3%). The remaining models from the field of medicine were coded as “general medicine” because they did not indicate a specific subfield.
Field of Study of Included Models
| Primary field | Secondary field | Models |
|---|---|---|
| Business | Leadership | Zeni et al., 2016 |
| Management | Jones, 1991 | |
| Child and Youth Care | Not Specified | Garfat & Ricks, 1995 |
| Education | Administration | Green & Walker, 2009 |
| Teaching | Ehrich et al., 2011; Johnson et al., 2017 | |
| Engineering | Not Specified | Fan, 2003 |
| Medicine | Critical care | Kanoti, 1986 |
| Dentistry | Johnsen et al., 2020 | |
| Emergency medicine | Marco et al., 2011 | |
| Epidemiology | Soskolne, 1991 | |
| Family medicine | Tunzi & Ventres, 2018 | |
| Geriatrics | Kirsch, 2009 | |
| Internal medicine | Kaldjian et al., 2005 | |
| Nursing | Bolmsjö, Sandman, & Andersson, 2006b; Cassells et al., 2003; Cassells & Gaul, 1998; Christensen, 1988; DeWolf, 1989; Ferrell et al., 1991; Greipp, 1997; Haddad, 1996; Hough, 2008; Hughes & Dvorak, 1997; Macpherson et al., 2020; Murphy & Murphy, 1976; Park, 2012; Phillips, 2006; Schaffer et al., 2000; Sullivan & Brown, 1991; Toren & Wagner, 2010 | |
| Oncology | Nekhlyudov et al., 2009 | |
| Psychiatry | Grundstein-Amado, 1991; Hayes, 1986; Hundert, 2003 | |
| Not Specific | Candee & Puka, 1984 (Deontology); Candee & Puka, 1984 (Utilitarian); Harasym et al., 2013; Schneider & Snell, 2000; Siegler, 1982; Tsai & Harasym, 2010 | |
| Organizational behavior management | Business | Bommer et al., 1987 |
| Psychology | Coaching | Duff & Passmore, 2010 |
| Counseling | Cottone, 2001; Forester-Miller & Davis, 1996; du Preez & Goedeke, 2013; Sileo & Kopala, 1993 | |
| I/O psychology | Heyler et al., 2016 | |
| Pediatric psychology | Shahidullah et al., 2019 | |
| Psychobiography | Ponterotto & Reynolds, 2017 | |
| School psychology | Boccio, 2021; Laletas, 2018 | |
| Not Specified | Tymchuk, 1986; Hill et al., 1998; Liang et al., 2017 |
Thirteen models were specific to the field of psychology (24%). Four of the psychology specific models were from the subfield of counseling (31%) and two were specific to the subfield of school psychology (15%). Other specified psychology subfields included coaching (n = 1; 8%), industrial/organizational psychology (n = 1; 8%), pediatric psychology (n = 1; 8%), and psychobiography (n = 1; 8%). The remaining models were coded as “general psychology” because they did not indicate a specific subfield.
Three models were specific to the field of education (5%). Two of these were specific to the subfield of teaching (67%) and one was specific to the subfield of administration and leadership (33%). Two models were specific to the field of business (4%); one of these was specific to the subfield of management (50%) and the other to the subfield of leadership (50%). One model was specific to the field of child and youth care (2%), one was specific to engineering (2%), and one was specific to OBM (2%).
Figure Figure2 2 shows the number of models that contained a problem-solving approach. A total of 23 models included a problem-solving approach (42%) and 32 did not (58%). Most of the models with a problem-solving component came from medicine (n = 15; 65%), followed by psychology (n = 7; 30%), and engineering (n = 1; 43%). No models from the fields of business, education, or OBM included a problem-solving component.
Bar Graph Showing the Number of Decision-Making Models with and without a Problem-Solving Component, and Models that were Sequential or Nonsequential
Figure Figure2 2 also shows the number of models that were sequential. A total of 52 models were linear or sequential in nature (95%), whereas 3 were not (5%). Most of the models that were sequential came from medicine (n = 32; 62%), followed by psychology (n = 14; 27%), education (n = 3; 58%), business (n = 2; 4%), engineering (n = 1; 2%), and child and youth care (n = 1; 2%).
The goal of this literature review was to identify and analyze published ethical decision-making models in behavior analysis and allied disciplines to determine consistency in recommended approaches. We examined 55 ethical decision-making models to collect data on what recommended steps were included and what approaches were most frequently emphasized. Three general themes within ethical decision-making models arose from our analysis. These include: (1) What steps were included within models; (2) Whether the steps were sequential (i.e., a behavior chain); and (3) Whether the entire process could be labeled as problem solving (i.e., Szabo, 2020). We discuss each of these findings in turn.
The first main finding surrounds the variability in recommended steps of ethical decision making across models. We found that each of the nine steps coded appeared in an average (arithmetic mean) of 58% of the articles (range: 20%–93%). This suggests that some consistency exists in what behaviors various scholars recommend practitioners should engage in when faced with an ethical decision. However, the wide variability in how frequently each behavior appeared also highlights that ABA practitioners would benefit from researchers clarifying at least three important characteristics of ethical decision-making models. These are: (1) What behaviors are necessary and sufficient to make an optimal ethical decision in ABA contexts (i.e., component analysis)? (2) What are the conditions under which specific steps are and are not needed (i.e., conditional discrimination analysis)? (3) Is there an optimal functional result of ethical decision making that is more important than the specific topographies a practitioner uses to contact that outcome (i.e., functional analysis; see Cox, 2021)? Practitioners and researchers may begin to explore some of these questions when engaging in ethical decision making.
More than half of the articles examined emphasized the need for consulting ethical codes. It is interesting that more ethical models recommended practitioners reference codes of ethics from outside their discipline (n = 44; 80% of models; e.g., personal, religious, organizational) than their own discipline’s code of ethics (n = 40; 73%). To our knowledge, the conflict between personal and professional codes of ethics is an underexplored topic in the ABA literature. Nevertheless, the slightly greater emphasis on other codes of ethics in addition to one’s own discipline suggests this might be an important area where practitioners could use guidance. Also, the field of ABA would likely benefit from future research and scholarship surrounding the conditions and functional outcomes of ethical decisions where personal and professional values conflict.
It is important to mention that our review was done prior to the publication of the BACB’s (2020) ethical decision-making model. The BACB’s model was published in the analysis and writing stage of this review. Our findings suggest a robust literature spanning 40+ years, 60+ articles, and 50+ models all clustered around similar ethical decision-making steps published by the BACB. Perhaps most intriguing is that we identified the nine steps from our review prior to the publication of the BACB’s model, and no previous models had incorporated all nine ethical decision-making steps until the BACB published their decision model (BACB, 2020). Practicing behavior analysts would benefit from future component analyses, conditional discrimination analyses, functional analyses, and empirical support surrounding the BACB’s ethical decision-making model.
Our analysis also suggests that behavior analysts and allied professionals approach ethical decision making similarly. Given the complexity of ethical decision making and the shared types of dilemmas human service professionals contact, some convergence is expected. However, there are many reasons that two professionals from different disciplines may come into disagreement (Boivin et al., 2021; Bowman et al., 2021; Cox, 2019; Gasiewski et al., 2021). Having familiar systems with empirical support for how to navigate ethical dilemmas might improve the likelihood that a positive resolution occurs. Further, such interprofessional similarities in ethical decision-making processes allows future interdisciplinary dialogue to focus more on specific areas of agreement because what and how information will be used to make a decision is already agreed upon.
We found that 95% of the ethical decision-making models could be described as a behavior chain (e.g., Catania, 2013). Framing ethical decision making as a behavior chain might be useful as it highlights the interrelated and sequential nature of ethical decision making. That is, completing one step in an ethical decision-making behavior chain leads to a context wherein the next response in the chain is more likely to contact reinforcement. For example, until you have gathered all relevant information about how the decision will affect all relevant parties, your ranking and weighing of information seems less likely to lead to the best outcome. That said, the temporally delayed nature of behaviors and consequences involved in ethical decision making is different than how behavior chains have been studied in laboratory settings (e.g., Baum, 2017; Cox, 2021; Slocum & Tiger, 2011). Future research will likely be needed to better understand the effects of temporal relations on behavior chains and thus determine what approach best provides a behavioral description of ethical decision making.
It is interesting that the order in which steps were proposed differed across models. We are unaware of any research that compares the effectiveness of different sequential ethical decision-making models to understand whether the order of behaviors recommended as a chain are more or less useful. Nevertheless, future research that identifies the extent to which rigid sequences of behaviors need to occur to optimize decision making would be helpful for the field of ABA. Such information would likely improve behavior analytic training programs and prove useful for clinical directors, ethics committee chairs, case supervisors (e.g., BCBAs), and direct staff (e.g., RBTs).
Recent attention has been given to the common-sense problem-solving approach (Szabo, 2020), which we used to score models within the current analysis. This problem-solving approach may offer great utility and is observed across various fields (e.g., cognitive psychology; Szabo, 2020). Within behavior analysis, this problem-solving approach has increasingly been applied to teach complex skills (e.g., Suarez et al., 2021). Our review involves an interesting extension of this analysis to ethical decision making and indicates the steps of the models may also point to additional precurrent behaviors or mediating strategies that could prove to be important elements of the behavioral chain.
We found that 42% of the ethical decision-making models could be described as including problem solving (e.g., Kieta et al., 2019). Framing ethical decision making as involving problem solving is advantageous because of the existing empirical literature on how to teach problem-solving skills and recognition of the importance of verbal stimuli and verbal behavior (e.g., Kieta et al., 2019). However, this also might have the drawbacks of adding complexity and less empirical support specific from the behavior analytic literature on describing, predicting, and controlling problem solving. This suggests that there are either components of ethical decision making outside of problem solving or that there are components of problem solving that might be missing from current decision-making models. Future research using concept analysis (e.g., Layng, 2019) combined with laboratory experiments may help clarify which of the above scenarios is more likely (or if there’s an unknown third!).
We also found that 58% of the ethical decision-making models could not be described as including problem solving. We are unaware of any research that has directly compared the effectiveness of ethical decision-making models with and without problem-solving components. Nevertheless, a practically useful set of empirical questions might identify the conditions under which ethical decision-making models with and without problem-solving components are more helpful for practitioners. Behavior analytic training programs subsequently could teach fluency toward ethical decision making via problem solving under some conditions and ethical decision making without problem solving under other conditions.
The current study included several limitations. One limitation centers on the procedures used for rater agreement. Article ratings were completed in a group format and by consensus among the authors. It is possible that reactivity to other members of the group affected overall ratings (e.g., Asch, 1956). It is also possible that the search terms we used failed to capture relevant ethical decision-making models or that additional search terms would have led to different results. Further, we also restricted our inclusion criteria to specific human service fields allied to ABA. Thus, it is possible that a more comprehensive search of ethical decision-making models across more varied professions would lead to different outcomes. Finally, we did not include ethical decision-making models published in books mainly due to access issues and a typical lack of peer-review for books. Regardless, these limitations may provide greater support for our primary findings that the existing variability in ethical decision-making steps and overall lack of empirical support suggest this area is ripe for future research.
The development of an ethical decision-making skill set is vital for behavior analysts and for other human service providers. Dilemmas present as complex circumstances, with specific and unique contextual variations that require nuanced assessment. The process of training behavior analysts to meet these demands is daunting. There is a need to identify strategies for navigating dilemmas and for making ethical decisions. Allied professions and behavior analysis have identified steps in this process. Many of these models use problem-solving techniques. The BACB’s Decision Making Model overlaps substantially with existing literature across professions, and uses a problem-solving, sequential approach. These results are especially interesting as we had completed identifying the decision-making steps scored in the current article before the BACB model was released. It seems that the field has built a model that is entirely aligned with and built upon this interprofessional database. It will be important to empirically evaluate this new model. It will also be important to explore other decision-making approaches, to compare models, and to (potentially) match models to the contextual variables embedded in the presenting dilemma. The field of behavior analysis has, at times, been insular, and this has been a source of internal and external criticism. However, this review of the literature supports the substantial overlap across fields and provides concrete hope for mutually beneficial interdisciplinary collaboration. So, although decision-making models can be field-specific, ethical dilemmas appear to be universal and so are the intended outcomes. As behavior analysis tackles this complex skill set, it is important to learn from colleagues in allied disciplines, examine the component skills likely to be crucial to the development of this behavioral repertoire, and develop procedures for measuring, teaching, and training clinicians to methodically approach ethical dilemmas.
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.
