Test 1 Flashcards
Behavior needs to be
observable, reliably recorded, quantifiable
Why do you need measures of behavior?
provide best evidence that an intervention designed to change behavior is or is not working
To be useful in addressing human problems your measurement systems need to…
- measure the real problem
- be reliable
- be sensitive to the amount of change
- be practical
Goals in Applied Behavior Analysis
- Make things better
- Create solutions that are replicable
- Develop a better basic understanding of human behavior
Mechanical/electrical recording
Advantages: easy to use, highly reliable, constant recording possible
Disadvantages: expensive
Self-reports
Advantages: easy, inexpensive, no intrusion on privacy
Disadvantages: not always believable unless there is another person
Direct observation by observers
Advantages: maximize likelihood of recording behavior you want to measure
Disadvantages: person time and intrusion on privacy
Direct observation by observers is most common. To use it you need (3 things)
- definition of behavior
- method of recording the amount of behavior
- evaluation of the reliability of recording
Frequency measure
count number of times behavior occurs during an observation
Duration measure
record how much time behavior occurs during an observation
Interval-by-interval measure
divide observation period into small time blocks and score whether or not behavior occurs in each of the time blocks
Time sampling
observe frequently for brief moments during an observation and score whether behavior occurred or not during each moment you observed
When would you use frequency count or duration measure of a behavior?
Frequency- hitting (number)
Duration- sleep (amount)
Why would you use interval-by-interval?
gives you an overall estimate of amount of behavior; shows agreement b/t 2 observers more precisely
Why would you use time sample recording?
- gives estimate of amount and shows agreement b/t 2 observers
- BUT need a very large number of time samples to get accurate amount of behavior
Difference b/t interval-by-interval and time sampling?
Interval-by-interval observe for a longer time period.
How do you calculate the reliability between observers?
Larger total divided by smaller total multiplied by 100 to get percentage of agreement.
13/15= .87x100= 87%
What is the typical interval size in interval-by-interval recording?
10,20, or 30 seconds
What is a problem with comparing totals of 2 observers?
You don’t know whether or not the observers are recording the same events at the same time.
What is the first question you want to answer about your intervention?
Does it work? Does it change behavior in a desirable direction?
To answer the first question of intervention you need (2 things)
- way of determining that the behavior you are measuring has changed in desirable direction
- adequate experimental design to rule out extraneous variables
Experimental-control group designs
intervention or no intervention at the same time with different people who are randomly assigned to groups
Experimental-comparison group designs
intervention or no intervention at the same time with different people who are selected to be comparable to the people in the experimental group
Reversal designs (ABAB)
intervention or no intervention with the same person at different times
Multiple-baseline designs
intervention or no intervention at different times with the different people or in different situations or for different behaviors
Changing-criterion designs
systematic manipulation of behavioral requirements with a person to see if behavior matches the requirements
Alternating-treatment designs
alternating several treatments multiple times with a person for a single behavior
Reversal design steps
- baseline
- intervention on baseline
- maintain intervention until change
- return to baseline (remove intervention) until goes back to baseline
- reinstate intervention until behavior is acceptable or can’t improve further
Changing-criterion design steps
- take baseline until behavior shows no trend or moving opposite than expected
- intervention with criterion for reinforcement slightly higher than baseline
- interventions increase criterion for reinforcement
Alternating-treatment design steps
- take baseline on 2 tasks until both show no trend or going opposite than expected
- start intervention and alternate
Second question of intervention
Does the intervention work enough? (consumer satisfaction/normative measures)
Third question of intervention
What parts of the intervention contribute to the effectiveness of the intervention?
How do we evaluate consumer satisfaction?
rating scales (1 to 5)
What are normative measures?
How do the outcomes of the intervention compare to “normal” amounts of the behavior?
Evaluations of components of intervention
cost, difficulty of implementation, undesirability of some components