Psychological Testing Principle Types of Psychological Tests Mental ability tests Intelligence – general Aptitude – specific Personality scales Measure motives, interests, values, and attitudes Key Concepts in Psychological Testing Standardization Test norms Standardization group Reliability Correlation coefficient Validity Content validity Criterion-related validity Construct validity Classical Test Theory Goal of testing is to identify subject’s “true score” on a given attribute. Classical Test Theory says… Test Score = True Score + Error of Measurement Goal of test developers is to minimize this! Sources of Measurement Error Where do these measurement errors come from? Item sampling error Internal consistency error Time sampling error Random error Testing How do psychologists know how well their test is measuring a person’s true score? By assessing test reliability Assesses Time sampling error Test-Retest reliability Alternate or Parallel Form reliability Internal Consistency reliability Assesses Internal consistency error Assesses Item sampling error Test-Retest Reliability Judging Internal Consistency Correlation Matrix Q9 Q10 Q17 Q18 Scale Q9 Q10 Q17 Q18 I I I I Scale Mean if Item Deleted Q9 Q10 Q17 Q18 1.0000 .1829 .1888 .1730 1.0000 .5718 .4031 1.0000 .5302 1.0000 Corrected Variance if Item Deleted ItemTotal Correlation Anxiety Scale Items am carefree 8.9250 8.4753 .2220 am rarely9.4750 overcome 9. with 6.2272 worries .5460 I am carefree am not a 9.2875 worrier 5.3720 .6160 10. I am rarely overcome with rarely feel 9.1750 fearful or anxious 6.8804 .5141 .0494 .3452 .4367 worries .2999 17. I am not a worrier 18. I rarely feel fearful or anxious Reliability Coefficients Alpha = Squared Multiple Correlation .6846 4 items Standardized item alpha = .6749 Alpha if Item Deleted .7513 .5655 .5080 .5930 Testing How do psychologists know their test is measuring what they think it is measuring? By assessing test validity Content validity Criterion-related validity Construct validity Fig. 9-4, p. 339 Fig. 9-5, p. 340 Assessing Construct Validity Convergent ConstructValidity Validity Divergent Construct Correlations Our measure of extraversion Correlations HEIGHT SHOE EXTRA202 1.000 -.066 should not be HEIGHT correlated be correlated withwith NEOEXTRA .843** CLASSEXTRA NEO Extravers ion Scale SHOE .477** .843**1.0001.000 .011 things unrelated related to extraversion to extraversion Class ClassExtraversion ExtraversionScale Scale 1.000 -.066 .477**.011 1.000 **. **.Correlation Correlationisisssignificant ignificantatatthe the0.01 0.01level level(2-tailed). (2-tailed). Assessing Construct Validity Convergent ConstructValidity Validity Divergent Construct Correlations Correlations Our measure of impatience HEIGHT should be notImpatience correlated be correlated with Scale ‘02 with SHOEto impatience. Impatience ‘03 things related unrelated toScale impatience. HEIGHT ‘02 SHOE EXTRA202 ‘03 1.000 1.000 .857** -.085 .275* .857** .275*1.000 .039 1.000 Impatience Scale -.085level (2-tailed). .039 1.000 *. Correlation is significant at the 0.05 **. Correlation is significant at the 0.01 level (2-tailed). Testing Utility of Psychological Tests Reliability Alternate or Parallel Form Internal Consistency Test-Retest Validity Content Criterion-Related Construct