When working in any setting as a practitioner claiming to be practicing ABA or implementing the principles of ABA, it is a requirement that data be collected. As Jose Martinez-Diaz warns, "If you are not taking data, you are not doing ABA." At minimum, data collection is vital in order to measure progress and ensure fidelity to the treatment plan.
When considering which type of data collection system to utilize with your learner, an important concept to familiarize yourself with is that of the dimensional properties of behavior. Behavior has 3 fundamental distance properties that can be measured. As a practitioner, you should be aware of and familiar with which dimensional property you are measuring. The property of repeatability refers to our ability to measure behavior because BEHAVIOR CAN BE COUNTED. When you are taking frequency data, measuring the rate of engagement in a behavior, or recording celeration data (perhaps we can discuss standard celeration charts at another time), you are measuring the repeatability of behavior. Behavior is something that happens and then can happen again, and again, and again. It is repeatable. If you are attempting to measure some phenomena that is not repeatable, then you are not attempting to measure behavior.
A second fundamental dimensional property of behavior is temperature extent. Temporal extent refers to the fact that behavior occurs during some period of time. This "period of time" is measured by collecting duration data. Duration data allows the observer to record the length of time that passes while the behavior is engaging in a targeted behavior. If you are attempting to measure some phenomena that does not take up time, then you are not attempting to measure behavior.
Finally, a third dimensional property of behavior is temporal locus. Temporal locus describes how behavior occurs at some time relative to other events in time. More specifically, temporal locus helps describe "when" behavior occurs. When measuring temporal locus, the observer may collect data on response latency or data on inter response time. Both of these measures will allow the practitioner to determine information regarding when a behaver is engaging in a targeted behavior. If you are attempting to measure a phenomena that does not occur at some point in time, you are not attempting to measure behavior.