The State of BCBA/RBT Compensation and Turnover in Applied Behavior Analysis
2019-2020 Trends In Pay Scales And Turnover Across Direct Care Staff, BCBAs, And Clinical Directors
We are going to be talking about BHCOE’s most recent compensation report with Dr. David Cox, BHCOE’s Chief Data Officer. We are also going to be doing a conference rundown. Let’s dive in. Welcome, David.
How are you doing?
We’re good. Anna is here too.
I’m so glad to be back. Welcome, David.
I don’t know about you guys. We got out of this amazing leadership meeting and I’m feeling good. I’m feeling bonded and connected.
Sara, that’s what I was thinking. What a great day. It’s so fun working with people that you like. That’s like, “Wow.” Everybody is so connected. It’s fun. It’s a great day.
There’s so much fun and cool stuff going on across the departments. I’m excited, 2022 is going to be a pretty big year.
David, we’ve got to ask you. Tell us about your love of data. How did you become the Chief Data Officer at BHCOE?
Data is critical to anything that you’re talking about.
I started getting interested in data when I was a kid. I have a fascination with life on Earth. That led me naturally into the hard sciences and how this all come to be here, behavior, how does that all work out, and then the hard sciences, with data, it’s critical to say that you’d know anything that you’re talking about. I got into the science data generally, but for big data, that love came about when I pivoted away from ABA for individuals with autism spectrum disorders and developmental disabilities toward pro-health behaviors, diet, physical activity and substance abuse.
In that camp, you don’t get 30, 40 hours a week. You maybe get like 15 minutes every 2 weeks to do something. You have to get creative if you’re approaching it from an operating perspective and how you’re going to collect data on those environment behavior relations. Technology is one way you can do that, but then that comes with the challenges of you having this tech collecting all this data. How do you handle that, process it and get it into a single format across the different technologies to analyze?
I found big data techniques to do that. I started getting interested in there. Artificial intelligence and machine learning have a lot of promise and opportunities. I got fascinated with that. We were on a random panel together, Sara, on ethics or something. I started talking about what was going on. This opportunity existed and was perfectly aligned with my interest.
You were at Blue Cross Blue Shield when you joined. You got to see the other side.
I did. It was very interesting. It’s six months of peeking behind the curtain.
What would you say is the hardest thing about working with data for all of us out there who don’t live and breathe in that world every day?
There are two things come to mind with that question. The first is this idea of data distributions. Many audience are quite familiar with the Bell Curve, normal distribution, and using the average score as one number to describe that whole distribution. It turns out there are 30, 40 distributions that exist in the wild. When we talk about one data point or one number that’s going to describe a large data set, you have to understand the distribution behind those data. Also, when you start running analytics, you might want to say, “This variable has this effect on some outcomes.” Unless you’re playing around with the distributions appropriately, those analytics can be off. You may be thinking that you’ve observed something that you actually haven’t. That’s the first one.
The second one is human variability and problem definition. You even set up a case with a kid that you’re trying to measure some toothbrushing or whatever. It can be challenging to get two BTs and the BCBA all collecting data the same way. When you think about some of the stuff we’re doing or you scale to big data where you have tens of thousands of people all thinking about the same thing in slightly different ways and collecting data slightly differently, on the backend, taking those different data definitions and trying to make something of it is challenging.
I’m sitting here laughing because I’m like, “We are very familiar with that problem here at BHCOE.” For those of us who are toe-dipping into the world of data collection, talk to us a little bit about the big picture. What role does data play in our field or any field? In the broader world, we’re here talking about standards, standardization and benchmarking. Why is it important to collect large amounts of data, particularly for our field? What’s so important about standardizing the way we collect those data?
I’m biased. I love data but I do firmly believe in Clive Humby’s quote, “Data is the new oil.” Generally, the ease, efficiency and costs of which we can collect data made it central to much different business and operational processes. We’ve moved away historically from this idea of behaving from intuition. We have data to support and guide the decisions that we make from a business standpoint. From a clinical perspective, data is also central to understanding and reflecting on how well you’re doing at something, consistent metrics and comparing to some benchmark.
It’s that field for ABA. As a clinician, I can’t know how well I’m doing unless I can compare my data to something else, to another set of providers. Unless we’re all talking about data and collecting it the same way, that comparison becomes almost impossible. That’s where standardization comes in if we’re going to reflect on who we are and how well we’re doing. You have to have standardized data collection of some sort to make those comparisons.
Sara, before we dive into more of the details of the report we’re going to discuss, there’s one thing I would like to ask. As someone newer to BHCOE, how does BHCOE decide what data we collect? Sara, for organizations out there who might be interested in becoming accredited, I’m thinking they would want to know how we keep that data secure and make sure all the things that we’re putting together what it’s being used for. I’m asking both of you in that world as someone who’s talked to a lot of provider organizations what that looks like.
I’ll start with a little bit big picture. Ultimately, when we collect data, it typically has some function to it. All the data that we typically collect are related to being able to make an educated and legally defensible decision about an accreditation. Essentially, we have to be able to showcase that we have all the information we need to be able to showcase that an organization is, in fact, quality. We need to be able to stand behind that if we get audited or if there’s any type of legal issue that might arise with that organization. Typically, those data elements that we collect are for that purpose. The first bucket is the information that we have to have to be able to make an accreditation decision.
There’s a second bucket related to benchmarking. David talked a little bit about that. The requests that we’ve seen come in through providers that we worked with over the years is that most industries have some type of repository of a larger data set that they can benchmark against. Our industry, because it’s so new, lacks that data set. We took it upon ourselves to understand those KPIs and metrics that organizations need to be intelligent about their business, maintain competitiveness within the marketplace, as well as understand they’re hitting the mark or not. Also, to be able to improve what they’re doing continuously.
There’s the second bucket that we use that’s related to benchmarking and business intelligence that allows organizations to make better decisions about trends in the market. That’s a little bit about what we’re going to talk about. The third bucket, which I’m sure we’ll have an episode on in the future is the future of ABA and care. Those are things related to things like value-based care or outcomes, research and measurement.
Some of the components that we do measure allow us to be innovative and be forward-thinking, what we’re collecting and how we’re informing accreditation decisions because ultimately, we want that first bucket to be useful and evolve as the industry evolves. Those are the three ways in which we collect data or why we collect them. David, do you have anything else to add on those?
You’ve nailed it and you’ve been doing it longer than I have. You’ve covered it well. The only other thing I’d add is when we think of a standard and how we’re going to collect those data, there is a standard process where we consult subject matter experts who throw things out to the field and get feedback. We review the peer-reviewed literature and all of that gets compressed into options. It’s the question of, “Realistically, what can we practically ask organizations to provide without being too burdensome?” That goes down to that final decision of, “This is the thing that we’ll look for data on.”
What about the security on that for organizations that may be like, “How are you keeping this information secure that I’m sending you?”
Compared to designing a standard, that’s one of the easier parts. The nice thing is that there are a lot of laws, regulations and best practice standards for the technological infrastructure for how data can and should be transferred, how it needs to be stored and protected behind paywalls and multifactor authentication, things like that. That technology and stuff exist. We’ve followed best-practice standards in that regard.
You can’t run a sophisticated organization without thinking about cybersecurity. Security risks are at an all-time high, especially considering how virtual we’ve all gone. It’s something that any sophisticated organization has to have at the back of their mind at all times, especially when you’re collecting data. It’s part of our general governance as well. We’re consistently monitoring security threats that arise. We’re regularly reporting internally and externally, any risks that we have on the cybersecurity side and any possible negative scenarios or actions we’re taking in the last 90 days, plans for the next 90 days. It’s a core function that we have internally to monitor this and make sure that we’re meeting the best standards.
Not all turnover is bad turnover.
I’ll also throw out there for any of the audience that is into machine learning and big data analytics, the cyber security realm is fun. As you can imagine, as security measures get put in place, bad actors figure out ways to get around that, so you have this game theory system going on. There’s a lot of fun modeling going on there.
That sounds great. I know organizations certainly are something that are important. That’s awesome. Sara, we want to dive into the report.
I wanted to share a little bit about the need that we saw in the field that drove the report. If you’ve been working with us over the last years or consumed any of our information, we typically would report on compensation and turnover as a page in our annual report. You’d have to dig a couple of pages in to try and find it.
We noticed that people were referencing that broader document for that information. This is the first year we released it as a standalone. The goal is to continuously refresh it so that it’s available to our members pretty consistently. I’ll start with David. First of all, what can you find in the report? What is something that you think is essential that people should know? If they’re in the report, what are they going to find there?
Generally, within the report, there are two halves to it. One is the general compensation for direct care staff and supervisory staff. We’ve broken that out by five different regions that are ranked based on their cost of living index. Fifty states’ territories break those into 5 groups of 10. Within that, we then report the average salary, median salary, minimum, maximum and total sample size. When I say salary, that’s for the supervisors directly. For direct staff, we report the hourly rate. That’s the big thing for compensation. Similarly for turnover, we look at some different trends and turnover for direct care staff by the year 2016 through 2020.
We also look at turnover by type. Was it a disciplinary termination? Did they leave voluntarily? What impact did COVID have? We do that same thing for supervisory staff. It’s the general content. There are some interesting trends generally that we observed in the data over time in terms of increases and overall compensation increases in turnover. One big thing that’s challenging with this data set is COVID. What role did that play in the status set? With the trends that we’re observing here, are they real or not real? What does that look like? That’s the high-level overview.
One of the biggest things I hear is that many providers see high compensation quality as a trade-off. They’re looking at it as like if you pay too much, you must be providing bad service or a bad work environment. If you pay less, you’re providing a better work environment. You’re a data guy, but looking at everything that you see here, do you think there’s a trade-off? What do you think is the relationship between pay and quality?
It’s a complicated relationship. That’s my short answer. It reminds me of the pro-profit versus nonprofit discussion. You could be a phenomenal organization that pays well and have high quality. Similarly, you could pay well and have poor quality and vice versa. Maybe you don’t pay well and have high quality or you don’t pay well and have poor quality. There are so many things that go into running a successful clinic and business that are going to impact how well I can pay or not pay my employees. It’s a complicated question and much bigger than any assumption in that capacity.
David, do you think this report can answer the question? Does pay rate contribute to turnover? To what percentage? Do you feel like it’s a large contributor? From what you’ve seen, if somebody asked that simple question from this report, what would you answer?
The data in this report can’t answer that question. In 2020, we ran a benchmark report specifically down turnover where we looked at some more advanced analytics. With all the data that we had to be, we try to predict turnover. If we can, what variables seemed important? The general trend from that and the literature on turnover is we know that staff satisfaction relates to turnover. Pay rate is one component of staff satisfaction that influences overall scores there but it’s certainly not the whole story.
When we ran those predictive analytics turnover benchmark report, staff compensation or employee satisfaction with compensation was toward the top of the important variables but it wasn’t the most important. We found in employee satisfaction that career development was the most important predictor. It’s a complicated and more nuanced relationship that exists. There are a lot of ways people can be compensated or gain benefits from their job, not only their hourly rate or salary.
There’s this beautiful map of the United States and it has all the five different cost of living, color-coded states in there. You see this range of the lowest cost of living, which is about $18 an hour pay rate for direct care staff. On the higher end, you have about $21 an hour. When you go on the next page, you have all these columns where it talks about not just the average, but things like the median hourly or the average minimum or average maximum. Can you talk to us a little bit about what do we do with information like the median or the average min or the average max? Why did you include those in this report?
This is my way to try to sneak in distributions. Those data are there to provide context. An average hourly rate is interesting in and of itself, but what’s not shown here is, is that distribution fat? Are we seeing a lot of variabilities? It’s very narrow and we see very low variability. With the one data point in the average, you can’t see that. Adding in the mean and comparing that to the average, if they’re relatively similar, it might be a normal distribution or probably close. With the min and max, try to give the audience some range of the variability that we’re seeing in our data set. If I were to do this again, I’d maybe add in a visual like a histogram, standard deviation or something else to give a bigger picture of that variability.
I was going to laugh because there’s an element in this report that David did suggest that we include a standard deviation in this report. I was like, “You people aren’t going to know what that means.” People are asking what standard deviation is. I will go on record by saying that if you’re mad that there’s no standard deviation, you can send your hate mail to the office and I will receive it gladly.
When we’re looking at the supervisors and you have that range of $70,000 up to $87,000, I’m going to the highest cost of living index and see that the max average is $150,000 on average, what do we do with that? People see that and they get nervous like, “My staff is going to ask me to pay them that.” What would you say to someone whose supervisor brought them this and said, “This is how much I should get paid or I’m not being compensated appropriately?”
In the nicest way possible, remember that’s also the max. These are people that have probably been in the field for 10, 15 years. Some people get compensated in various ways that $150,000 at 1 agency doesn’t mean the same thing as $150,000 at another agency. When I report that I’m paying somebody $80,000, maybe I did or didn’t include other things. If you bill more than 25 hours, you can make extra money or things like that. Going back to those data definitions pointing out, this is all context. That’s an average max. Who knows how long those people have been in the field? Who knows the true reimbursement or compensation structure of that organization?
I would direct them back to the average and median. Where are most people falling around in terms of what they’re compensated for? Not to belabor the point but this is where the standard deviation also is helpful. The min and max are informative but what you don’t know is how much of an outlier were those effectively? Most of the distribution centered around $80,000, $90,000, $85,000, $95,000. Is it spread out? $150,000 and then $145,000 was next on the list or something like that. We can lose that without the visual.
We did get some questions about this that’s why I’m imposing this to David. We decided to run a couple of other analyses on this data set because we were trying to understand. Let’s say there was one provider that paid that big amount. What would happen if we pulled that data set out? Do you want to talk to us a little bit about winsorizing and the two keys methods? The results, if I remember, were not that significant when we did pull the max averages. Tell us what you did and what our finding was generally.
With the winsorizing and the two-piece method, the idea there is how can we control for outliers in our data set? Those that can be defined in several ways, those high or low numbers, they’re going to impact, especially the average where that falls. Winsorizing your data set is effectively where you take all extreme values depending on how you do it beyond the 5th and the 95th percentile. You reassign those values to the numbers that are at those percentiles. The advantage there is to get to keep all the data that’s in your data set. The downside is that you start placing extra weight on those edges of your distribution at those specific percentiles.
An alternative method that I tend to prefer is called the two keys method. You take the 25th percentile and the 75th percentile. The difference between those, multiply that by 1.5. Essentially anything that falls above the 75th percentile number, below the 25th percentile on that number, you drop those from your data set. I like that method because it’s nonparametric. You don’t have to necessarily worry about the distribution of your data, though it can influence that. In any event, two methods for handling those extreme values may influence your average in particular.
It’s not always about how much you pay someone. What a lot of people value is the opportunity, leadership, and meaning that they can contribute.
We played around with a couple of different methods. At the end of the day, the numbers didn’t change much, which gave me confidence in particular with the numbers that we were putting into the report, and also those questions again, it’s good feedback to figure out how we might do this differently for the one that will explain to you.
David, how much do you think reimbursement rates within a specific region or state contribute to this? I can hear a provider in my head saying, “I can’t pay those rates because my primary business is in a state, which has one of the lowest reimbursement rates.” How would you talk about how that plays into this report if it does? For providers, especially if you’re operating in one state and that state, in particular, has low reimbursement rates.
I’m going to point you to the cost of living difference. In group five, there’s a cost of living index score. When you look at the difference between Rhode Island, which is 119, to Hawaii that’s 192, that range is very wide. You do have to be a little cautious in your interpretation. If you’re in Rhode Island and grouped in group five, you might be at the very much lower. That’s why that standard deviation that David was talking about is important, that max and min as well.
Piggybacking off that, that’s probably where I direct most readers. If you’re in a specific state or one state in particular or even across states, which one of those buckets do you happen to fall in? Thinking about the average and those median salaries relative to the cost of the living index within that.
It’s something that probably affects some people more than others. It depends.
I’m also thinking about the function of pay within an organization, and if there are other benefits that you can offer. There are economic constraints on what I can pay somebody but, a plug for BHCOE, you can also give opportunities for professional development and things like that. There are alternative ways to make the work environment more reinforcing, and that’s also hard to do.
Speaking of retention, we could dive into the whole issue with turnover. I’m hearing from so many providers that it’s a difficult environment. There have been a lot of contributing factors but I’d love to get both of your opinions on what you’re also hearing. We are seeing it be a real inhibitor to being able to take more kids, access to care, those things.
I couldn’t agree more. What’s interesting when you look at the numbers here is when we look at turnover for direct care staff, it’s steadily increasing but not in a way that’s too surprising. When you look at the 2016 data set and there was a sample of around 5,000 clinicians included, their turnover was about 50%. When you’re looking at 2020, 4 years later, you have around 6,000 folks in the sample and it’s around 59%. It is going up. What was interesting to me is when we think about turnover by type, I remember during COVID, I was thinking that these poor technicians are probably getting the grunt of all of it. I was thinking there would be a lot of COVID-related terminations for the technicians and furloughs.
What was surprising to me looking at our data set is the clinical directors took on the brunt of all the distress during COVID. When you look at terminations, about 49% of the terminations happened at the clinical director level were because of COVID-related reasons in 2020. When you look at supervisors or direct care staff, it was 14% or something like that. What was interesting to me is the clinical directors who I thought were holding up the organization were the ones that got cut first when you had this huge crisis. Those are oftentimes more senior staff.
Also thinking about the direct care staff 2020 that’s 59%. If you drop that one from the analysis also and you’d look across four years prior, it’s 50%, 52%, 48%, 53%. It seems like it’s pretty steady. That’s also comparable turnover percentages to other similar related industries. I am excited for the 2021 report that we’ll put out and see if the 59% hold. Do we continue to see an increasing trend? Is it back down to the 50% somewhere in between?
The other piece that’s interesting to me as well is in most organizations, you’re always dealing with supply-demand. Usually, there’s always enough product and it’s about getting them to the customers. Here, it’s flipped where you have customers but not enough product. In this case, it’s people. It’s a very unique dynamic to look at within an industry. Ultimately, it will level out but this is something that if we can crack as a field, we’ll be so much more successful for it and our patients will benefit so much from it.
There are a lot of factors that are very interesting with the RBT level of turnover because it is an entry-level position. In many cases, it’s trying to separate those with a career path, also those that are in it for a job and they’re going to move on to another job. It’s an important piece that organizations are reflecting on because I do think back to the beginning of our conversation, it’s not always about how much you pay someone.
What a lot of people value is the opportunity in the company, where they can go, their leadership and the meaningfulness that they can contribute. A lot of it too is figuring out that person to hire, which we did an incredible leadership assessment. That was interesting. Make sure you’re hiring the right person.
For those reading, we started using the Predictive Index, which is an interesting tool for predicting whether a person is a good fit at the organization and helping people work together.
That’s a huge piece of this puzzle. Knowing that you’re hiring the right person for the job is critical because I’ve been in so many clinics and myself, my child having an RBT and working with her is not an easy job.
I feel like this field and every organization isn’t for everyone. When I look at that turnover too, I go, “How much of that is because that person wasn’t a good fit for your organization, and how much that is because you need to fix something within your organization?” I also typically look at a sales pipeline or a customer pipeline, but you need to look at your hiring pipeline and figure out what we can adjust for at the top of the funnel to make sure that the right people are entering the funnel to begin with.
Related to that, at some point in time, some people find different calling, paths or areas in the industry. That’s okay. You give me the idea that not all turnover is bad turnover. As a field for collecting these data, we’re contributed to a large database. We’re also giving that on other variables. People start to get a sense of what is a good rate, whatever that means, and what are the variables that contribute to the challenging turnover versus the turnover that’s okay to live with?
I love that statement, “Not all turnover is bad turnover.” This data set is so informative. What advice do you have for ABA business owners or leaders who are thinking about how to utilize these turnover data sets? How can they integrate it into their regular business operations?
The biggest way organizations can start integrating these data would be to collect data with similar definitions. Going back to this idea of defining a problem statement clearly in a way that allows me to compare myself to others, people could collect turnover and define it however they want, but at the end of the day, if that definition differs from everyone else’s, it will be hard to know what to do with their data. One way you can use this report is to look and see what are the types of turnover that people seem to be talking about? How are people classified or bucketing their turnover? Internally, start analyzing processes collecting data. What are some variables in my company that might contribute to those larger buckets that people are reporting on in the industry?
Anna, you used to be in an operating role. What would you do with these data?
You can’t run a sophisticated organization today without thinking about cybersecurity. Security risks are at an all-time high right now, especially considering how virtual everything is.
It’s such a great report. These types of things are so needed for our industry. They can be so helpful in creating a long-term sustainable business that can make an impact for kids with autism, which is what we’re all here for. I love that we’re doing it. I also love that we are constantly evaluating how we can do it better. David is already talking about the next report and how it’s going to be even better. I love that.
David, what’s the best data set you’ve ever worked with?
The best can be defined in a few ways. One is the one that sounds the coolest. We conducted some quantitative analyses of people’s verbal reports of their psychedelic experiences. That was a lot of fun.
What kind of psychedelics are we talking about?
It was up to them. They reported on whatever psychedelic experience they had that led to changes in substance abuse in particular. They described the event as it unfolded.
Where was that dataset from?
That was at Johns Hopkins. That was fun. It’s neat. We have one publication out on that and a second one coming if people are curious. The other one is fun with Jake Sosine, who’s a Data Analyst at BHCOE. We downloaded every published behavior analytic article. We have one manuscript under submission. We have two more coming down the pipeline, but we analyze them for trends. What are people talking about? What’s the difference between the experimental and applied branch? What does that look like? It sounds fun. There’s a lot of neat stuff going off of that. My third favorite data set from interest is that I have ten years of every pitch thrown in professional baseball. We’re analyzing for patterns of choice. What leads to different decision-making? What does that look like? How does the matching offer them? That one’s cool.
I knew baseball was going to make its way into this conversation. I should have thrown a bed out to Sara but I was 100% sure.
For people that may have not attended previous presentations of mine, I always work baseball into all my slides. Play Station has one illusion to it if people are curious. Those are fun for different reasons, but the best all-around data set is the one that we’ve started to build and improve on here at BHCOE. There’s a lot of cool data on the industry that’s going to allow us to ask and answer questions that nobody has asked before. To me, that’s exciting and it will have a big impact on the field.
I was starting to feel a little left out that you weren’t saying the BHCOE data set, so I’m glad you’re saying that. It is a beast. You have to admit.
It is. There were so many distributions and how people define data differently. It’s presented so many fun challenges to figuring out what’s there and what we can do with it. It’s fun. I like it.
Thank you so much, David. This was very informative. It’s always so fun chatting with you on and offline. Anna and I are so lucky to work together. We have a great team. We can’t wait to have you back when our next benchmark report comes out.
I appreciate you guys having me on. This is a fun show. Thanks for having me.
David, where can people find you on social?
If you have any questions for us about this or anything else we talk about or if you have any topic, ideas, feel free to email us at TSR@BHCOE.org.
We are going to do our quick conference rundown. All the conferences in ABA are coming up very soon here. We have the Ohio ABA Conference coming up February 11th through 12th, 2022 in Columbus, Ohio. That is going to be a virtual and in-person hybrid, I believe. Everyone is scrambling a little bit. There’s the ABAI Annual Autism Conference in Seattle, Washington, March 5th through 7th, 2022. That same weekend, we have the CalABA Conference, March 4th through 6th, 2022, and then APBA in New Orleans, March 17 through 18, 2022. Nebraska will be from March 24th through 25th, 2022. Are you folks going to any of them? What are you thinking?
I’ll be at two of those, for sure. I’ll be at CalABA and APBA. Those are going to be fun. CalABA is always a good time. I’ll take the opportunity to promote data and analytics at BHCOE. We have a symposium there. Come check it out. It’s going to be fun.
We have a presentation at APBA as well. I believe Dr. Kazemi, our Chief Science Officer is an invited presenter at the Ohio ABA Conference, which I’m very excited about. Honestly, I’m praying for CalABA and APBA because years ago, they had to cancel their conference and go virtual. Everyone’s hurting so I’m crossing my fingers that they can still have their conference. It goes off without a hitch.
Everybody is holding their breath but if you are strictly going on location, New Orleans sounds great. That sounds fun but I’m super excited about CalABA. That’s going to be a great event. There are a lot of great things going on in California with some of the advocacy being done and things like that. It’s going to be a great year of exciting things happening.
The Behavior Analysis Leadership Conference is coming up too. That’s the same weekend as CalABA and ABAI Autism. We have three competing conferences. If you’re going to any of them, let us know. We can’t wait to hear how they are.
- Predictive Index
- Jake Sosine – LinkedIn
- Twitter – David Cox
- Research Gate – David J. Cox
- LinkedIn – David J. Cox
- Ohio ABA Conference
- ABAI Annual Autism Conference
- CalABA Conference
- Dr. Ellie Kazemi – LinkedIn
- The Behavior Analysis Leadership Conference
About David Cox
Dr. David Cox, Ph.D., M.S.B., BCBA-D is the Chief Data Officer at BHCOE. Dr. Cox began his work in behavior analysis in 2006 by providing in-home early intervention for children with ASD. He subsequently obtained a Masters in Bioethics, became a BCBA, and expanded the people he served to adolescents and adults with aggression and SIB across school, clinic, and residential settings. After practicing clinically for eight years, David went back to school to obtain a PhD in behavior analysis from the University of Florida.
There, his research focused on quantitative analyses of human operant behavior and the design and implementation of behavioral economics-based interventions for pro-health behavior (e.g., diet, physical activity, and drugs of abuse). David then completed post-doctoral training at Johns Hopkins University School of Medicine in computational behavioral economics and behavioral pharmacology, as well as a second post-doctoral fellowship in data science through the Insight Fellows Program. Prior to joining BHCOE, David was a Principal Analyst of Behavioral Science in the Department of Data Science at GuideWell. Dr. Cox joined BHCOE in 2020 to execute a unified data strategy across BHCOE to provide organizations pursuing accreditation a seamless and technology-enabled customer experience. In addition, he is responsible for leveraging BHCOE’s considerable dataset to advance the company’s predictive analytics and machine learning initiatives surrounding clinical benchmarking and analytics offerings