LECTURE 5.1

I. Measuring a variable

A. Search through

1. Textbooks

2. Psychological Abstracts (but check the Thesaurus first)

B. Consult

1. Dictionary

2. Theory

C. Consider

1. Self-report measures

2. Observation

3. Physiological measures

A. Sensitivity

1. Avoid dichotomous measures by assessing response's

a. Frequency

b. Accuracy

c. Latency

d. Speed

e. Intensity

2. The shorter the chain of inferences, the more sensitive

B. Reliability

1. Usually a prerequisite for validity

2. Reliability also aids sensitivity

3. Looking for a reliability coefficient of at least .75

C. Validity

1. Reliability usually necessary, but never sufficient

2. Look for evidence of:

a. Convergent validity

b. Discriminant validity

c. Content validity

3. Interview subjects to get their ideas about what the measure was assessing

D. Measurement means assigning a number to an observation, but not all numbers are equally informative

1. Nominal scale measurements: Different numbers mean different properties. Not very rigorous

2. Ordinal scale: Bigger numbers more, but can't say precisely how much more. Ex: Ranks, grades

3. Interval scale: Describe how much more of a quality. Ex: degrees Fahrenheit, Rating scale data??

4. Ratio scale: Meaningful zero point (0 means absolute absence of quality) and has interval properties. Ex: Weight, Magnitude estimation.

5. Conclusions: Don't assume that you have ratio scale data, you probably don't

Back to Chapter 5 Menu