Week 3 lecture slides.pptx

PSYCHOMETRIC PROPERTIES OF TESTS: VALIDITY PSY61204 Psychological Tests and Measurements Dr Michele Anne Overview What is validity Types of validity Face validity Content validity Criterion validity Construct validity Aspects of validity What is validity? Validity A...

PSYCHOMETRIC PROPERTIES OF TESTS: VALIDITY PSY61204 Psychological Tests and Measurements Dr Michele Anne Overview What is validity Types of validity Face validity Content validity Criterion validity Construct validity Aspects of validity What is validity? Validity Accuracy of test in measuring what it is intending to measure Types of validity Face validity Content validity Criterion validity Construct validity Nomothetic point of view Primary validity Basic validity where the test should correlate to similar areas Secondary validity Test ability to measure the underlying dimension or theory Tertiary validity Justification for developing a new test Face validity Face validity Whether a test looks like it is measuring what it should measure There is not statistics or calculation, rather it is based on subjective perception in regards to the area of measurement Can be affected by participant’s rapport and cooperation Can effect how seriously they complete the test Activity 1 Go to https://socrative.com (Student login) Room name: Anne4991 Content validity Content validity Whether the items of the test adequately covers and represents the area of measurement Can be divided into Content representativeness Content relevance Requires in-depth knowledge of the test area Best to include a team of experts to create and judge the items There is no agreement about number of experts needed and amount of agreement between both Content validity (cont.) Evaluating content validity Subjective methods Involving experts to judge relevance and representativeness of test items Empirical methods Factor analysis to show obtained factors match the test area (theoretically) Criterion validity Criterion validity Whether the test correlates with another validated test of the same / similar / related test area Types Predictive validity Comparing the 2 test to predict the future Concurrent validity Comparing the 2 test to predict the present Criterion validity is determined by conducting Pearson or Spearman’s correlation (depending on normality) between the scores of the 2 tests Criterion validity (cont.) Contrasted groups Refers to groups that differ significantly on a test area Administering the test to these different groups and still obtaining correlations increases criterion validity Criterion contamination When the criterion validity is influenced by factors not related to the area of measurement More often when there is involvement of subjective rating and experimenter bias Construct validity Construct validity Constructs are theoretical or abstract variables which cannot be measured readily These constructs are typically operationalised into quantifiable areas or measurements Construct validity measures whether the test is an accurate reflection of the underlying theoretical constructs / frameworks Validates the theoretical construct via the test Construct validity (cont.) Types Convergent validity When a test correlates with variables that it should correlate with Discriminant validity When a test does not correlates with variables that it should not correlate with Construct validity is determined by conducting Pearson or Spearman’s correlation (depending on normality) between the scores of the tests Construct validity (cont.) Methods to assess construct validity Group differences Correlation and factor analysis Internal consistency Test-retest reliability Studies of process Aspects of validity Bandwidth fidelity Bandwidth refers of range of applicability of test Broad-band test (wide area) versus narrow-band test (narrow area) Fidelity refers to thoroughness of test As bandwidth increases (becomes more wide), fidelity decreases Breadth versus depth Sample Group homogeneity Similarity or variability within the sample Higher homogeneity (similarity) will result in less correlation A distribution of high and low scores are needed Cross cultural validity The validity is stable across samples from different cultures Validity and results can be generalized Interpretation Validity typically uses correlation, but unlike reliability, there is no agreed upon score for “good” validity Variance (square of the correlation coefficient) are sometimes used in place of correlation coefficient, so determine how much variation is related to the test and how are explained by other factors Regression equations and expectancy tables (cross- tabulations) are sometimes used predict the relationship between test Questions?