Chapter 6 - Visual Analysis of Behavioral Experiments Flashcards
4 Tactics in using Behavioral Strategy…
Develop a behavioral definition
Use methods of Direct Observation
Check for reliability and validity.
How to design a single-subject experiment.
Do a visual analysis of the data.
To decide whether differences between base line and treatment look DIVIDED and STABLE.
To decide whether observed differences in behavior between baseline and treatment look convincing.
Do they convince you that the level of behavior has changed as a result of the treatment?
Do they convince you that the treatment actually helped?
- Use with single-subject design
The Principle of Visual Analysis
The ranges of the last three points of two conditions are mutually exclusive.
The numbers are MUTUALLY EXCLUSIVE if one condition consists only of low numbers while the other condition consists only of higher numbers.
_____ conditions help convince scientists that additional observations will continue to show differences.
2,3,1 and 7,9,8 - None of the baseline numbers are as large as even the smallest of the treatment numbers.
Divided Conditions
Means that one condition has a range that includes values from the other condition.
A condition consisting of high values but one low value might not exclude the values of the other condition.
Example: 4,1,0 (baseline) and 7,3,9 (treatment)
Failure to mutually exclude values from the other condition raises a doubt that added observation in treatment would still show a difference.
Observations that are ______ are NOT CONVINCING.
NOT Divided Conditions
The last 3 numbers of one condition are not moving CLOSER to the numbers in the other condition.
Numbers are not moving closer to the larger values (or are moving away)
Stable Condition
Four Steps of Visual Analysis
- Ask is the ranges of behavior in the two conditions are DIVIDED.
- Ask if the rates of behavior in the two conditions are STABLE.
- Ask if the differences are CONVINCING in terms of the first two questions.
- Ask if the treatment CAUSED the differences.
Every pair of adjacent conditions must be DIVIDED; every condition must be STABLE.
IF the rates of behavior fail to be BOTH divided and stable, then the differences are not convincing.
Convincing Differences
What experiment designs rule out TIME COINCIDENCES and INDIVIDUAL DIFFERENCES.
Only an experiment that uses a REVERSAL or a MULTIPLE-BASELINE design rules out both TIME COINCIDENCES and INDIVIDUAL DIFFERENCES.
You can conclude that the treatment caused the difference only if the difference is convincing and if you used either of these designs.