Collecting accurate data is a foundational part of any top-quality behavior intervention program for patients with autism spectrum disorder (ASD). Data are crucial for a variety of reasons:
Bad Report | |
---|---|
Data | |
Caregiver Report | “Calvin’s aggression has gotten better. I feel like he knows that we’re not going to give in. He definitely doesn’t seem as upset as he used to be.” |
Explanation | This is an example of a bad report. We see that the caregiver’s report and the data don’t necessarily coincide. The intervention is having a slight effect, but not to the extent that the caregiver seems to be reporting. |
Good Report | |
---|---|
Data | |
Caregiver Report | “Melanie’s aggression has gotten better. I feel like she knows that we’re not going to give in. She definitely doesn’t seem as upset as she used to be.” |
Explanation | This is an example of a good report. We see that the caregiver report is more accurate. But we only know the caregiver’s report is accurate because the data show us so. |
ABA treatment should be data-driven. BCBAs don’t make decisions based on feeling, or even on caregiver reports. which you have seen can be unreliable. Instead, BCBAs make decisions based on what is actually occurring. Data collection is how you and the BCBA know what is really happening.
What does this mean for you? Well, the data you collect are very important!
Without your data, your BCBA will have nothing on which to base their decisions. Accurate data collection is key; if your data are not accurate, the BCBA won’t truly know what is happening with behavior.
You're saying that is like a reoccurring thing. Then we can just teach him the rule. You know, if they don't have what you want, you say, oh, I wish they had milk, but OK. You know, teach him the replacement thing to say, and then we'll take data on [INAUDIBLE] is he able to use the replacement behavior that we've taught.
What did you have for snack yesterday at school? What was there, but you didn't like it? What was there? Do you remember?
Peaches.
Peaches. Peaches are good.
And what else?
Peaches. Are peaches good for me?
I love peaches. I think they're so good, and they're very good for you. They have lots of vitamins.
I don't like them!
That's OK. You can just say, I don't like peaches. So you give a rule, have him practice, and then we'll take data on if he's able to implement the rule.
We have discussed that data are extremely important to ABA, but this leads to additional questions:
For problem behaviors, some types of data you may take include
You may also take data on caregiver behavior, technician behavior, and inter observer agreement, or IOA. This is a method used to check the consistency of our data, which will be discussed more in a later part of the training.
You should constantly be tracking data unless the BCBA has specifically told you otherwise (e.g., do not collect new data about food and snacking).
In relation to the behavior, collect data as soon after the behavior target as you can. In lessons, data should be taken after each trial, because every second that goes by compromises the accuracy of your data. If you aren’t 100% certain of what the response was, the accuracy is already gone.
When collecting problem behavior data, safety is obviously your number one priority.
EXAMPLE
Don’t let a patient bang their head against a window just so you can mark things perfectly.In most cases, do what you need to in order to manage the behavior and then mark your data. In some instances, you will be able to track the data as the behavior is occurring.
Data should be both accurate and consistent so that we can make the best treatment decisions possible for the patient. As mentioned already, accurate data are very important for BCBA decision-making. Consistency in data collection is very important as well.
All behavior technicians on a team should be consistent with skill acquisition data and behavior management data.
Type of Data | Criteria |
---|---|
Skill acquisition data |
What is considered correct? What is not correct? What is considered extraneous or irrelevant behavior? |
Behavior management data |
What are considered occurrences of the problem behavior? What are non-occurrences of the problem behavior? What is considered one instance or episode of the behavior? |
One method of checking consistency in the data is IOA, or inter observer agreement. All behavior technicians take data on the same behavior at the same time. Data are compared and calculated for accuracy; 80% accuracy is considered good.
One way that instructors make sure data is reliable is called inter observer agreement. Inter observer agreement or IOA measures data reliability by calculating the degree by which two people agree on a behavior that they have both observed, measured, and recorded. The data that they collect is then compared to determine the level of similarity. If both individuals have recorded data in an identical or very similar way, then the data is then considered to be reliable.
However, if there is low level of agreement or many differences in the data sets between the two observers, then agreement is low and their data collection results are considered to be unreliable. Inter Observer Agreement, or IOA, is something that should be measured and incorporated regularly into ABA programs in order to ensure that the data being collected by instructors is reliable and that everybody is measuring the same behavior in the same ways. The type of IOA procedure used for any particular client will greatly depend on the type of data collection procedure being used.
Your program supervisor or BCBA will advise you in regards to what type of IOA procedures will be used for your particular client. Let's look at an example of how IOA works. Here we see an individual within a discrete trial training session. In order to determine if data collection is reliable, two different observers would watch the same responses from the individual and record those responses.
At the end of the discrete trial activity, the two data sets would be compared. If there is agreement between the two data sets per trial and that agreement is high, then we say that the data is reliable. However, if there are discrepancies between trials of each of the data sets, then the behavioral data is said to have low agreement. In other words, there are problems with reliability.
This may mean that additional training is required for all of the staff in order to make sure that data collection procedures are being utilized appropriately. It may also mean that adjustments need to be made to the data collection procedures. In any case, IOA is something that should be occurring on a regular basis within any ABA program to ensure that the data being collected is accurate and reliable.