Certainly, I’m no expert when it comes to the stock market. But I do think there is something we can all learn from technical analysis because it superficially seems similar to what we do in behavior analysis. Technical analysis refers to attempting to make predictions about stock market prices from historical stock market data. Essentially, people are analyzing graphs and other data to try to make predictions about the future. That may seem like what we do in behavior analysis. But it really isn’t, or it shouldn’t be.
A deep dive into technical analysis is both beyond the scope of this blog and my current abilities, but there is minimal evidence that it is effective, and there seems to be problems with the data in studies that found it to be effective.
Don’t we do the same thing in behavior analysis? We look at data on historical performance and then use that data to make predictions on future performance. If a child has increasing levels of problem behavior, we might change something to try to decrease problem behavior. If the child has had no problem behavior in months, we might assume that nothing needs to be done about the behavior.
A likely reason why technical analysis doesn’t work to predict the stock market is that there are literally millions of variables that may impact the price of a stock. Things like politics, labor negotiations, weather patterns that affect the price of raw materials, or whether I need to sell my shares due to a family emergency. All of those things can have an impact on the price of a stock. Therefore, the relationship from previous patterns in a graph are probably not a good predictor of future performance. But an occasional win by chance can lead to superstitious behavior, and make people think they can predict the future when they probably can’t.
But, in behavior analysis this shouldn’t be the case. In behavior analysis there is a clear relationship between what we do and behaviors observed. We specifically decide when to implement the behavior plan. So, if problem behavior is high and I implement a behavior plan on Wednesday, and then get an immediate drop in the problem behaviors, that’s much better than technical analysis because I controlled when to do the intervention. It is probable that my intervention worked and was not due to chance. Although behavior analysts understand that the results could be due to chance, there are a variety of methods in single subject design that make that very unlikely.
Unfortunately, in practice it is easy to mess this up and fall into a pattern of doing something like technical analysis. Graphs of behavior without the analysis are not useful for making predictions. As an example, if we collect data on problem behaviors across the whole school day, there are so many variables that might impact the data that it will be hard to know what is causing what (e.g., a peer made a negative comment, there was a fire drill, the assignment was hard today, his pencil broke, the teacher usually calls on him frequently but didn’t today, etc. etc.).
For me, the lesson of technical analysis is that decisions based on data that don’t show “the why” are not effective. It is easy to believe you are making scientific decisions based on relevant data, but instead fall into superstitious patterns. In behavior analysis, we also make decisions based on data, but it is absolutely essential that we know WHY those data went up, down, or are staying stable. If we don’t know “the why,” the data are essentially useless for making predictions about the future, especially when there are any changes to a program.