Threats to Valid Measurement Flashcards
What are the threats to valid measurement?
The validity of behavioral data is threatened when measurement
is indirect, when the wrong dimension of the target behavior is
measured, or when measurement is conducted in such a way that the data it produces are an artifact of the actual events.
What is indirect measurement?
Indirect measurement occurs when the researcher or practitioner measures a proxy, or stand-in, for the actual behavior of
interest
What is an example of indirect measurement?
Using a student’s score on a standardized math achievement
test as an indicator of mastery.
What is direct measurement?
when the behavior measured is exactly the same as the behavior that is the focus of the
investigation or behavior change program.
What is an example of driect measurement?
a student’s score on a properly constructed
test consisting of math problems from recently covered curriculum content.
Why does indirect measurement rarely occur in ABA?
Meeting the applied dimension of ABA
includes the targeting and meaningful (i.e., valid) measurement
of socially significant behavior.
When would an indirect measure be appropriate?
the researcher or practitioner has no direct and reliable access to the behavior of interest, to make inferences about private events or affective states
What must the researcher do when using indirect measurement?
The researcher must provide a convincing case for the validity of the data.
What is compromised when measurement produces values for a dimension of behavior ill suited for measuring the behavior?
validity
What are some common causes of measurement artifacts?
Discontinuous measurement, poorly scheduled measurement periods, and using insensitive or limiting measurement scales.
What is continuous measurement?
Measurement that detects all instances of the target behavior during the observation period.
What has produced an increasing trend of continuous measurement in behavioral research?
Development of automated data collection technologies
What is discontinuous measurement?
Discontinuous measurement is any form of measurement in which some instances of the response class( es) of interest may not be detected
What problem does discontinuous measurement present?
A data set produced by discontinuous
measurement may be
an artifact of the measurement system rather than accurate portrayal of behavioral events.
When might we use discontinuous measurement?
Continuous measurement is prohibited in the absence of automatic measurement technology
What other factors besides absence of automatic measurement technology may prohibit continuous measurement?
The nature of the target behavior being measured; the number
of participants or clients whose behavior must be measured; the number, duration, and scheduling of observation periods; and the availability of trained personnel to observe and collect the data.
What are the three primary methods of discontinuous measurement? (Chapter 4)
Partial-interval, whole-interval, momentary time sampling
What study shows the artifactual variance of discontinuous vs. continuous measurement?
Thomson, Holmber, and Baer (1974) contiguous, alternating and sequential.
What was meant by contiguous, alternating and sequential in Thomson, Holmber, Baer (1974)?
contiguous: subject 1, 2, 3, 4 (16 minutes each)
alternating: subject 1 , subject 2, subject 1, subject 2 (4 minutes each)
sequential: subject 1, 2, 3, 4 (4 minutes each)
How did Thomson, Holmber, and Baer (1974) compare discontinuous vs. continuous measurement?
Recorded 3 discontinuous measurement types to a single continuous measurement called “actual rates”
How do you minimize the threat to validity from discontinuous measurement?
Frequent measurement, as a general rule… daily or frequent basis, valid sampling procedure
Which intervals are more preferable to limit the threat of validity to discontinuous measdurement?
Numerous brief intervals over longer, less frequent
When is a sampling procedure sufficient for decision making and analysis?
the samples represent a valid approximation of the true parameters of the behavior of interest
What two boxes should the measurement period check to prevent poorly scheduled measurement periods?
- Equal opportunity for occurrence/nonoccurrence of behavior
- Consistent environmental conditions
What observation times should be selected to assess the effects of an intervention?
the most conservative times, or when their rate of occurrence is most likely to differ from the predicted outcome
When should you measure behavior targeted for reduction?
At times when those behaviors are most likely to occur at their highest response rates.
When should you measure behavior targeted for increase?
At times when high-rate responding is least likely.
What problems with measurement scales pose threat to validity?
Insensitive (over or under sensitive), limited (does not detect full range of relevant performances)