When selecting candidates for a company a psychologist learn about someone's personal sexual preferences. The Psychologist? 

Does not report to the company this information - this information is confidential

Selection test in a criterion validity study

The investigator ^test cutoff score to decrease false positives  

Criterion-related validity coefficients for selection test for F/M. Females are statistically significant but nor males?

This may be due to differential validity   

If a test on one area example cynism correlates highly with a test in another area example anger? 

Low construct validity

Standard error measurament

Use to construct a confidence interval around an obtain test score

Variability in test scores

Test not reliable if coefficient is ↓ .80

A combination of true scores, and variability due to random or measurement errors

Reliability coefficient = correlation coefficient ► provide an estimate of the true score variability

Example- compare two scores of exam of the same subjects sample → test-re-test, alternate forms (use to measure a stable trait), internal consistency (better for measurment of charasteristic that fluctuate over time) and inter-rater (for subjectively score measures) -use Kappa -assess agreement between 2+ raters

To maximize  and increase reliability

• ↑ length of test by adding items that assess the same charasteristics
• range of score is unrestricted = examinees are heterogeneous

Confidence Interval

Reliability = test score variability reflect true score variability

The range within an examinee's true score is likely to fall given the obtain score

Derive using standard error - standard deviation that serve as index to measure error

Example - (95%) → + & - 2 standard error to obtain condidence interval

Validity - that the test measure what was intention to measured. 

3 types:

• content
• construct
• criterion-related

Multitrait-multimethod matrix

• Monotrait-monomethod - reliability coefficients
• Monotrait-heteromethod - correlate 2 dif measures beleif to measure same trait →evidence of covergent validity
• heterotrait-monomethod- correlate 2 similar measure that assess dif traits
• heterotrait-heteromethod- correlation between 2 dif measure that assess 2 dif traits  

A table of correlation coefficient that give information about a measure's covergent and divergent (discriminant) validity  

• covergent - correlate high with measures design to measure same traits
• divergent - low correration with measures that assess unrelated traits
• both = evidence of contruct validity

Criterion related validity

Reliability coefficient is never more that the square of root

example - coefficient of 81 → validity cannot be greater than 90

90 is the square root of 81

Norm reference score

Subject performance according to people in the norm group.

z-core

distribution with a mean of 0 and standard deviation of 1

+1 = examinee is one standard deviation above mean

t-score

distribution has a mean of 50 and a standards deviation of 10

so 60 ► examinee is one standard deviation above mean

Percentile rank

Example 84%= that a person did better than 84% of the norm group

Criterion/content/domain referenced scores

Provide information about an examinee's performance in an absolute rather than relative form

example- 80% = examinee answered 80 out of 100 correctly