Chapter 3 Methods for Recording Behavior EDP 7058.

Chapter 3Methods for Recording Behavior

EDP 7058

Introduction

• When recording dependent variables we are looking at the effectiveness of the independent variable (intervention)

• When the dependent variable is not readily observable (such as cognitive processing) then the dependent variable needs to be defined as an observable outcome or time elapse.

Anecdotal Records• Involve measurement of antecedents to the behavior, the actual

behavior and the consequences of the behavior.• Used as a tool to operationally define the dependent variable• Two forms of records:

1. Running records – these are continuous and record what is going on over a long period of time.

– PROS: Provide a wealth of detailed information– CONS: Time consuming2. Record an instance of behavior that occurs over a set period

of time or only when the behavior is observed.-- PROS: Less time consuming and intrusive– CONS: May lose valuable information* Independent observers are desirable but required at this stage

Measuring Quantitative Dependent Variables

• Two general types of measures:• 1. Event based – Measure frequency, accuracy

and intensity of a behavior.• 2. Time based – Used to measure how long it

takes for a behavior to occur or begin.• Permanent products – may be gathered

through use of videotape or paper and pencil so that it can be analyzed frequently.

Event Based Methods1. Percent correct2. Frequency3. Rate4. Trial to Criterion5. Cumulative Recording6. Interval Recording

A. Whole intervalB. Partial intervalC. Momentary Time Sampling

7. Magnitude8. Topography 9. Levels of Assistance10. Task Analysis

Percent correct

• Used when the number of opportunities to respond varies from session to session.

• There should be a minimum of 10 opportunities to observe behavior per session (20 is optimal)

• Must first determine what is correct and incorrect response

• Divide the number correct by the total of correct and incorrect responses and multiply by 100%

Interobserver Reliability

• Must have at least 2 observers to calculate interobserver reliability.

• Each instance of the behavior must be looked at for agreement.

• # of agreements is divided by the Total of agreements and disagreements and then multiplied by 100%

• 70-89% agreement is adequate• Less than 70% raises doubts about the

procedures

Frequency

• Involves recording the number of observations that occur during an observation session.

• Length of the session must remain constant• Used when target behavior is short in duration and is

easily observable and counted.• Opportunities to emit the target behavior should not be

restricted or controlled.• Interobserver Reliability: can be calculated by comparing

each observance as in percent correct or by dividing the smaller # of occurrences by the larger when that is not possible

Rate

• Used instead of Frequency when the length of the observation period varies

• Rate is determined by dividing the number of responses by the length of the observation.

• Rate is generally expressed as behaviors/minute or seconds

• Used for high frequency behaviors that would be difficult to measure over long periods of time.

Rate

• It is necessary to vary when the observations occur to make sure that responses are not a function of when the individual is observed.

• Rate is an estimate of frequency

• Interobserver reliability is determined the same as in frequency

Trials to criterion

• Measuring the number of responses needed to achieve a preset level of acceptable performance (criterion)

• Provides information on how quickly one attains a skill as well as which type of instruction requires less trials.

• Interobserver reliability- determined the same as frequency using # of trials in place of responses

Cumulative Recording

• Recording the total number of responses that occur across each observation session toward some final level of achievement.

• Ex: Selling candy

• Interobserver reliability- determined the same as frequency (smaller # of occurrences divided by the larger)

Whole Interval Recording

• Interval recording estimates the frequency of response when a high frequency of occurrences makes it difficult to record across an extended length of time

• Whole Interval- the observation period is divided into equal intervals and the target behavior must occur during the entire interval

• Used when you are concerned with episodes that last at least the length of an entire interval

• Can provide an estimate of duration• Interobserver reliability – based on each

interval

Partial Interval Recording

• Observances are scored if target behavior is exhibited at any point during the interval.

• May be a better estimate of frequency than whole interval but

• Does not estimate duration because individual may exhibit behavior in only a small percentage of the interval.

Momentary Time Sampling

• Records whether behavior is occurring at the end of the interval only.

• May yield the grossest method of frequency if intervals are too long or infrequent. Best when behaviors have a long duration

• Intervals should be short enough as to not miss behaviors.

Magnitude

• Magnitude is the strength force or intensity of a behavior.

• May be measured mechanically or manually through human judgment.

Topography• Topography is the correctness of how a

behavior is performed or how it looks.• Challenge is determining what represents

examples and non examples of the target behavior

Level of Assistance Recording

• Measures the degree of assistance required to complete a task

• Researcher is recording his or her own behavior rather than the subjects.

• Frequently paired with task analysis

Task Analysis Recording

• Similar to level of assistance recording• Behaviors to be learned are outlined in

clear successive steps• Often paired with trials to criterion in order

to learn how many opportunities it takes the individual to achieve each step.

• May also use percent of correct responses based on the number of opportunities to perform each step.

Time Based Methods for Recording Behavior

• Duration recording – involves measuring the length of time from when a behavior begins to its termination

• Used when the length of time a behavior occurs is of primary importance.

• Allows for frequency estimates

• Must make sure that duration is not artificially altered because the target behavior is not allowed to occur. You may decide to record frequency of alternative behavior as well.

Duration recording

• Allows researcher to determine:

1.Total duration- total time spent on behavior

2.Duration per occurrence- the amount of time spent engaged in target behavior for each event.

3.Average duration per occurrence- total time divided by number occurrences

4.Percent of time observed engaged in target behavior- total duration divided by total time observed.

Duration recording

• Interobserver reliability should be calculated for each occurrence as well as total duration in order to report differences accurately.

Latency recording

• Involves measuring the length of time from the delivery of an antecedent stimulus to the actual beginning of the target behavior.

• Interobserver reliability is measured in the same way as duration recording.