The variable variable

1
The variable variable

• Yes, colleagues, it is a variable. But the
  question remains, What kind of variable is it?

2
Variables and research design

• Independent variable
• Varies across individuals or settings
• Manipulated by experimenter
• Is the hypothesized cause
• Dependent variable
• Varies Across individuals or settings
• Measured by experimenter
• Shows the effect

3
Another typology of variables

• By precision of measurement
• Discrete variables
• Continuous variables
• By content of measurement
• Qualitative variables
• Quantitative variables

4
The independent variable

• Manipulation strategy
• Presence-absence manipulation
• Manipulation of amount (quantitative)
• Manipulation of type (qualitative)
• Manipulation tactics
• Manipulation of instructions
• Manipulation of events
• Manipulation of organismic variables

5
Translating the IV construct

• Operational translation Empirical statement
• How much? How long? How often? When?
• Operational definition Empirical referents
• What is it?
• Agreed-upon techniques
• Beware stereotypy
• Choosing from among options

6
Construct validity of independent variables

• Clear definition
• Multiple measures of expected effect
• Convergent validation
• Multiple manipulation of different IVs
• Divergent validation
• Manipulation checks Pretesting.
• Multiple IVs and interactions

7
Dependent variables

• Sensitive to the effects of the IV
• Observable and measurable
• Dependent variables measure quantity
• How much? (measurement) How many? (count) How
  fast? (rate) How often? (frequency)
• Or quality
• What type? What color? What value? What elegance?

8
Kinds of dependent variable data

• Questionnaire data (self-report)
• Rating data (by observers)
• Life data Commitment
• Not subject to group differences based on
  organismic variables
• be especially careful with self-report data

9
Reliability and validity

• The dependent variable is measured, and so raises
  issues of reliability and validity
• Reliability of dependent variable measures is
  generally low, unless the measures are repeated.
  Consequently, replication is essential.

10
Epsteins reliability results
11
Dependent variable validity

• Convergent validation Do different measures of
  the same construct correlate highly?
• Divergent validation Do different measures of
  different constructs correlate lowly?
• Problem Different measures may assess different
  parts of the same construct.

12
Increasing respondent validity

• Problems of faking, carelessness, and
  self-deception
• Disguise techniques
• Embed key items in a larger matrix
• Disguise the reason for measurement
• Use whoops procedures
• Have a confederate collect the data
• Collect data as part of an unrelated survey

13
Increasing response reliability

• Non-awareness techniques
• Assess required behavior of significance
  Commitment.
• Collect data surreptitiously
• Use participant observers
• Use unsuspicious subjects
• Use physiological measures
• Use physical traces