Continuation from this post...
So continuing our discussion on different ways to take behavioral data, here are some of the most commonly used discontinuous methods.
Discontinuous Measurement
These methods provide an estimate of a behavior rather than true representation. You are not counting every time a defined behavior occurs but rather grabbing a "snapshot" to give you a good idea of it. The advantages of doing this are that it requires data collection once per interval instead of every instance. This works well for a behavior that is hard to count or happens way too often to count accurately. It is also sometimes easier if many things are being managed or in an environment where you're trying to capture behaviors on multiple students. On the other hand, these methods do not collect all instances of behavior, so they may over- or under-estimate the response.
In general, it comes down to IF you can define and count every instance of a behavior and do so without staff running away screaming from being over-whelmed, then use a continuous or count method. It's more accurate. But if you are trying to juggle a lot and just need an idea of what's going on and a way to compare over time if your strategies are working, these methods are useful.
Partial Interval Recording
Interval recordings (whole, partial, and momentary time sampling) are used for estimating a behavior in which observers periodically look at the client at predetermined intervals and record whether or not a behavior is occurring. In the case of Partial Interval, did the behavior occur at least ONCE during the short observation interval? Then the interval is scored positive if the behavior occurred at any point in the interval. It doesn't matter if it happened once in the interval or 100 times, it's still just one "count."
For example, maybe we want to take data on nose picking. That's a good one. We define nose picking as any instance of the tip of the finger passing the nostril opening for more than 1 second. We ask the staff to take three measures a day in three different settings, intervals every 2 minutes for 20 minutes (total of 30 intervals). So the staff at 10:00 start the tracking. They watch the student for nose picking. At 10:02, they mark yes or no if they saw ANY nose picking during those two minutes. Then, at 10:04, they mark yes or no if they saw any nose picking during those two minutes, and so forth. At the end, we can count the number of intervals there was nose picking and the number of times there was an absence of nose picking. We can say something like 60% (18 out of the 30 intervals) involved nose picking that day. We can also make deeper analyses by looking at the settings, the activities, the people involved during the time, etc.
Now, Partial Interval Recording generally under-estimates the frequency of the behavior. If you are only marking one YES no matter how many times the behavior is occurring, you may not get a complete picture of how often the behavior really IS happening.
Partial Interval is appropriate to use when you want to decrease the duration or frequency of a behavior.
Whole Interval Recording
If you can grasp the concept of Partial Interval, this will be easy. In the case of Whole Interval, you want to know, Did the behavior occur for the whole interval that you are looking for it? The interval is scored if behavior occurred for the duration of the ENTIRE interval. This is often used to measure continuous behaviors that occur at a high rate such that the observer has difficulty distinguishing one response from another.
For example, maybe we want to take data on staying on task completing a math worksheet. We define "on task" as eyes on paper with no more than 10 seconds staring off (to allow time to think) and answers marked at least every 10 seconds. Or maybe we want to track a student's peer conversation, and we define "conversation" as eyes on speaker when listening and on-topic commenting when speaking with no more than 3 seconds processing time between turns. In either of those cases, for each of the intervals, we will mark yes or no if the behavior as defined occurred during the entire interval. Much like Partial Interval, but now instead of it needing to happen just once, we want it to occur the whole time.
Whole Interval generally over-estimates the frequency of behavior because if the behavior occurs for most of the interval but not at the last bit of the interval, then we do not record this data. So the result can be a little skewed. But, it can still be a useful tool!
We use this when we want to increase the duration or frequency of a behavior.
Momentary Time Sampling
This is the last discontinuous measurement I'll go into, and these are really the three most well-known and utilized in our field. There are other, obscure ways of collecting data on a discontinuous scale, but this will keep you plenty busy!
In MTS, you are still creating time interval to record your data and defining your behavior clearly, but now you will look up at the client immediately at pre-designated points and record whether the behavior occurred at that precise moment. In other words, at the end of each interval, you look up, and you see if that student is engaged in the defined behavior AT THAT VERY MOMENT and then mark yes or no.
For example, maybe we want to track the presence or absence of a student's stereotypic behavior (stimming). We define it as hand wringing: both hands brought up to under the chin and squeezed together for more than 5 seconds with or without vocalizations. We set the interval for every 30 seconds, and at the end of each 30 second mark, we look up and see if the student is engaged in hand wringing at that precise moment. We mark yes or no.
I do like MTS in the case of a group of students I'm trying to gather data on. It's pretty easy to define a behavior for each one, then set an interval on my watch, and each time it goes off, quickly check and see if each student is engaged in the behavior or not.
MTS can over- or under-estimate a behavior because you are just looking at that one slice of time. It's more accurate, the shorter the interval you create.
Permanent Product and Anecdotal Data
Permanent Product and Anecdotal Data are technically forms of discontinuous measurement, but they're pretty loosey-goosey. Permanent Product is simply a way to look at the physical product of a behavior. This can be used when the behavior you are assessing results in a lasting product or outcome. For example: number of written assignments completed, coloring, math facts. Most of our education system is built on this type of data collection. We look at the product a student produces and assess if they got the concept or not.
Anecdotal Data also considered permanent product. This is just note-taking on a behavior. It is a method of descriptively recording the behavior emitted by the learner, the response of others, and information about the environment. It is not a true measurement nor objective but may be useful information.
That's it for measurement for now! Feel free to let me know if there's a subject you'd like me to cover as I add to my posts!
Find data sheets here.