In this brief blog article, we will explore and define what a data artifact is.

A data artifact is something that, “appears to exist because of the way that it is measured” (Cooper, Heron and Heward, 2007).

A Data artifact refers to a behavior that looks like it occurs more or less often than it actually occurred as a result of the way in which the behavior is measured. It does not refer to data being inaccurate or that it wasn’t measured correctly. It means the measure itself was flawed.

The most accurate form of data for measuring how often a behavior occurs is rate data. Rate data tells you how many times a behavior occurs per time. For example, Angelina gets out of her seat 7 times per hour. 

Sometimes, you cannot measure rate because a behavior occurs at a very high rate or because of limiting factors. In the example with Angelina getting her seat out of her classroom, the teacher’s ability to measure Angelina’s behavior would be constrained because her teacher must pay attention to all the kids in the classroom. 

When this occurs you would want to use an interval based procedure. There are three types of interval measurements: whole interval recording, partial interval recording and momentary time sampling. All three measurements yield an approximation percentage of the time the behavior occurred in a given time period. The problem is there is a risk with all three measures of over or underestimating how often a behavior occurs especially when intervals are longer. When this happens, there is a data (also known as a measurement) artifact. 

Example of data artifact using whole interval recording

When using whole interval recording, a behavior is recorded only when a behavior occurs during an entire interval. It is typically, although not always used to increase a behavior. In this example, the requirement might be for Angelina to sit in her chair for the entire time interval. Let’s say that Angelina has a 60 minute work session. Her teacher was asked to use whole interval recording for 20 minute intervals to see how often Angelina is sitting in her chair. Data is recorded as a plus if she is in her chair for the entire interval and a minus if she is out of her chair for any part of the interval. In this example, let’s pretend Angerlina stands up in the first twenty minute interval for three seconds and stretches and sits back down. This interval would be scored as incorrect. During the second interval Angelina gets up to sharpen her pencil after it breaks. The second interval is scored as incorrect. During the third interval, Angelina stands up for ten seconds and sits back down when her teacher asks her to. According to the whole interval recording method, Angelina sat down for zero percent of the intervals in the work period. This data is accurate because it is correct but it does not really reflect what happened. In reality, Angelina only got up three times, at least once of which we know was justified and she was actually sitting for most of the time. This is an example of a data artifact. It appears Angelina was out of her seat almost the entire class because of how the behavior was measured but that’s not really the case. 

Example of data artifact using partial interval recording

Partial interval recording is when a behavior is recorded as occurring when it occurs at any point during an interval. Let’s consider another example of how partial interval recording could create a data artifact. In this example, Lisa will talk out of turn during class. Her BCBA asked her teacher to record whether or not Lisa talks out of turn at any point during a 30 minute interval during an after school period. Lisa whispers to her friend once during each thirty minute interval and each time says only two to three words. This data would show that Lisa spoke out of turn 100 percent of the time. This is accurate according to the data measure but doesn’t really reflect what happened. In reality, Lisa hardly talked to her friends at all. Therefore the measure is considered an artifact. 

Example of data artifact using time sampling

Time sampling is when a behavior is recorded when it occurs at the end of an interval. For example, Talen will engage in frequent hand flapping. It occurs almost constantly. The BCBA sets a timer for 10 minutes and at the end of ten minutes if Talen is hand flapping, the behavior is recorded. Let’s just say that during a one hour ABA session, Talen is flapping his hand at 55 minute out of the session. It occurs almost constantly. However, it just so happens that for five out of the six intervals, when the timer goes off, Talen is not hand flapping. The behavior is recorded as occurring 17 percent of the time which is accurate based on time sampling. In the same example, partial interval recording would have put it at 100 percent of the time, which was likely more accurate. This is another example of a measurement artifact. 

In all of these examples, the way in which a behavior was scored influenced how often it appeared to occur. Rate is always the best measure for recording how often a behavior occurs. When it is not possible to use a rate because the behavior occurs too often or because of limited constraints, another measurement procedure should be carefully selected and a therapist must take into mind the limitations of the measurement and the possibility of a data artifact. 

This article is useful for registered behavior technicians (RBT) or students who are studying to become board certified behavior analysts (BCBAs). Understanding (Applied Behavior Analysis) ABA terms is critical for both being an effective ABA therapist and passing your BCBA exam.

In an effort to help you study for your BCBA exam more effectively, this post is written in a “study note” form rather than as a long form blog post. They are my personal study notes I am sharing with you as a gift. I am spending my time studying so they are not edited. I am grateful for your understanding in overlooking the grammar! Happy Studying! 


Cooper, J. O., Heron, T. E., & Heward, W. L. (2007). Applied Behavior Analysis (3rd Edition). Hoboken, NJ: Pearson Education.

Add Your Comment