Measuring and Manipulating Variables Flashcards
Operational definition
- Define how variable is to be measured (within our experiment)
- Allows others to repeat experiment precisely
- Reduces generality of interpretations
Delayed gratification
Amount of time willing to wait for reward
Deciding how to operationalize
- Use operational variables that have worked in other studies
- Select operational definition that best fits what you really want to test
Common types of IVs
- Manipulated variables
2. Subject variables
Manipulated variables
- Situational variables (bystander effect)
- Task variables (diff. levels of IV)
- Instructional variables (by differing instructions)
- Effect of ‘treatment’ (therapy vs. no therapy)
Subject variables
Age, gender, etc.
-Varies by the subject
Common types of DVs
- Self-reporting measures
- Behavioral measures
- Physiological Measures
- Indirect dependent variable
Self-reporting measures
- Rating scales
- Opinion polls/ questionnaires
Behavioral measures
- Accuracy or speed response
- Frequency of response
- Verbal responses
Physiological Measures
- Way to operationalize person’s inner state
- Heart rate, bp
- Functional brain imaging (brain activity)
Indirect dependent variable
Use observable behavior to measure something unobservable
A given experiment can use (DV)
- A single DV
- Two + separate DVs
- Composite DV (several measures combined into single DV – intelligence)
Criteria for selecting variables
- Must be reliable (repeatable with same results- does NOT mean accurate)
- Must be valid (measure what was intended to measure)
Causes of unreliability- assess using
- Variations in how measure is administered – test-retest correlation
- Variations in scoring or data recording – inter-rater correlation
- Subject performance variations – split-half correlation
How to establish a measure’s validity
- Face Validity
- Discriminant Validity
- Concurrent Validity
- Content Validity
- Predictive Validity
Face Validity
- Appears to be valid
- Weakest type of validity
Discriminant Validity
- What else might be measured other than what you intended?
- Show that your measure DOES NOT correlate with the other factor
Concurrent Validity
-Showing your measure DOES correlate with another measure of the same thing
OR
-Show two groups differ in your measure when they should differ
Content Validity
Measure should accurately SAMPLE content it’s supposed to measure
Predictive Validity
Show you measure accurately predicts what it ought to predict
