Personality 2
Criterion keys, good and bad.
1. Nature of criterion key; how it differs from homogen
2. Good: doesn't matter what people do; it's what they say
Hard to fake
3. Bad: don't know what it means
4. Example, criterion keys for exam scores
Method variance -- MTMM
Suppose we go and ask about 100 college students (targets) to rate themselves on a few traits of interest, and that we also ask them to grab their best friend. The best friend will also rate the target person on the same traits. What shall we ask them? We will ask them to judge statements about themselves and their friends about questions such as the following:
Dominance: In groups of people, I usually "take charge." I like to tell people what to do. There are few things more satisfying than winning an argument. I talk more than most people in groups.
Risk: I like to stir up some excitement. I like to take chances. It's better to risk all your money for a chance at winning big than to never try at all.
Esteem I'm a better than average student. I'm good at many sports. I'm good at doing most everything I try. If I make a mistake, it doesn't matter too much because I'll do better next time.
After we get scores from targets and their best friends, we can compute a correlation matrix which shows the correlations between all the ratings. There are two sections which correspond to the correlations of traits within rater -- one for self and one for best friend. For example, the targets' correlation of self ratings of dominance and risk is .48. There is also a section which corresponds to the convergence of ratings -- self and best friend's ratings of target dominance correlate .30.
|
Dom |
Risk |
Esteem |
Dom |
Risk |
Esteem |
|
|
S Dom |
1.00 |
|||||
|
S Risk |
.48 |
1.00 |
||||
|
S Esteem |
.80 |
.30 |
1.00 |
|||
|
BF Dom |
.50 |
.20 |
.05 |
1.00 |
||
|
BF Risk |
.21 |
.10 |
.10 |
.50 |
1.00 |
|
|
BF Esteem |
.12 |
.05 |
.15 |
.75 |
.21 |
1.00 |
The above matrix is typical of results in personality and performance ratings.
Why don't they agree more?
Traits defined differently by different people
Response styles
Access to information
Situational determinants of behavior
Person situation interaction -- speaking in class
Intentional Biasing of Self Reports
1. Ability tests are hard to fake
2. Personality tests are easy to fake -- if you know what they want.
3. Responses to distortion.
a. infrequent responses, missing responses
carelessness, random response, answers off by 1
b. lie scales
fake good responses -- always investigate qualifications of political candidates, don't get mad at people, don't take it out on someone else, etc. Ways of developing tests to avoid these?
c. forced choice: matched for desirable response
Ethics
Is personality testing invasive? Is it reasonable for an employer to give personality tests to applicants for jobs?
Safety, theft, more productive work force, smoking
When is deception reasonable?