Transcript
Anders SjöbergAssociate ProfessorDepartment of Psychology
Disposition
• The background and development of two test review systems
• Advantages and disadvantages of each system is presented from the perspectives that validity is (or is not) a characteristic of a test.
• Example of a selection process validity study
• Questions
Swedish Psychological Association
• Swedish Psychological Association is the union and professional organization for the country’s psychologists
• One task is to review different kinds of psychological assessments carried out in the work and organizational area, such as personality and cognitive ability test used for selection and development
Swedish National Board of Health and Welfare (NBHW)
• The National Board of Health and Welfare is a government agency in Sweden under the Ministry of Health and Social Affairs
• One of the tasks is to review different types of psychological assessments carried out such to detect violence in marriage, abuse of alcohol and other health related problems
SPA Review Model Procedure
• SPA Review Model for the Description and Evaluation of Psychological Tests is a procedure that employs two anonymous reviewers for each test review, with a third person to oversee the review (Consulting Editor)
Swedish Psychological Association
• EFPA Review Model for the Description and Evaluation of Psychological Tests. Version 3.42, (2008)
EFPA Review Model Sources
• British Psychological Society (BPS) Test Review Evaluation Form
• The Spanish Questionnaire for the Evaluation of Psychometric Tests (Spanish Psychological Association);
• the Rating System for Test Quality produced by the Committee on Testing of the Dutch Association of
• American Psychological Association [APA], American Educational Research Association [AERA], and National Council on Measurement in Education [NCME]. US AERA/ APA/NCME . Standards for Educational and Psychological test
EFPA Validity
• The framework to operationalize validity is based on Standards for Educational and Psychological Tests [APA], AERA], [NCME], 1954). This conceptualization of validity holds that there are three approaches to the validation of tests.
• Content validation (demonstration that test items are a representative sample of the behaviors)
• Criterion-related validation (demonstration that scores on a test are related to an outcome)
• Construct validation (collection of evidence that a psychological concept or construct explains test performance)
Practice
• EFPA Review Model for the Description and Evaluation of Psychological Tests. Version 3.42, (2008)
2.10.1Construct Validity - Overall Adequacy(This overall rating is obtained by using judgment based on the ratings given for items 2.10.1.2 – 2.10.1.6. Do not simply average numbers to obtain an overall rating.)
2.10.1.2Sample sizes:[ -2] No information given.[ -1] One inadequate study (e.g. sample size less than 100).[ 0 ] One adequate study (e.g. sample size of 100-200).[ 1 ] More than one adequate or large sized study.[ 2 ] Good range of adequate to large studies.
2.10.1.4Median and range of the correlations between the test and other similar tests: [ -2] No information given.[ -1] Inadequate (r < 0.55).[ 0 ] Adequate (0.55 < r < 0.65).[ 1 ] Good (0.65 < r < 0.75).[ 2 ] Excellent (r > 0.75)
2.10.1.5Quality of instruments as criteria or markers:[ -2] No information given.[ -1] Inadequate information given.[ 0 ] Adequate quality[ 1 ] Good quality.[ 2 ] Excellent quality with wide range of relevant markers for convergent and divergent validation.
• American Psychological Association [APA], American Educational Research Association [AERA], and National Council on Measurement in Education [NCME]. US AERA/ APA/NCME . Standards for Educational and Psychological test.
• EFPA Review Model for the Description and Evaluation of Psychological Tests.
• Buros Center for testing
Swedish National Board of Health and Welfare (NBHW)
NBHW Procedure
• NBHW test Review Model for the Description and Evaluation of Assessment have a procedure that employ two anonymous reviewers for each assessment review, with one person to oversee the review, (Consulting Editor)
NBHW Validity
• Validity is defined as the degree to which evidence and theory support the interpretation of assessment scores proposed by the service provider of the assessment.
• Instead of talking about different kinds of validity, the service provider of the assessment must state explicitly what interpretations are to be derived from a set of scores and how to use these scores for decision making.
• In this way, the strength of the validity evidence refers to the probability that the inference is correct.
• Thus, it is critical for service providers of the assessment designing and conducting validation studies to concentrate their efforts on ensuring evidence for the inferences they wish to make in much the same way that they would otherwise “defend” their conclusions in an hypothesis testing situation.
Practice
• NBHW test Review Model for the Description and Evaluation of Assessment
Validity
The process of validation involves accumulating evidence to provide a sound scientific basis for the proposed score interpretations. It is the interpretations of assessment scores required by proposed uses that are evaluated, not the assessment itself. When test scores are used or interpreted in more than one way, each intended interpretation must be validated.
Evidence that the interpretation of the assessment score are correct.
Describe the validity studies
X Evidence that the interpretation of the results are correct is not possible to value due to lack of or insufficient informationX Evidence that the interpretation of the results are correct, should be revised and clarifiedX Evidence that the interpretation of the results are correct, should be supplementedX Evidence that the interpretation of the results are correct is good
Justification of valuation:
Proposals
Validity of a test
•Easy to evaluate
•Concentrates on statistics
•Difficult to evaluate
•Concentrates on content and evidence
Validity of the use of
a test score
As a reviewer
Validity of a test
•Difficult to evalute
•Concentrates on statistics
•Easy to evaluate
•Concentrates on content and evidence
Validity of the use of
a test score
As a client
Selection practice
• SPA model - psychometric properties of the test
• NBHW model – the selection process and decision
Example Selection
• Organization A use intelligence test in the selection process (N=200)
• Organization B use intelligence test in the selection process (N=200)
Validity argument
Results based on the validity argument
Test score
Low High
Performance
Low
High
85
85
15
15
r = .70
Question and Analysis
• The relationship between the test score and the selection decision (Not selected or Selected)
• Is the selection decision based on intelligence score
Organization A
Test score
Low High
Decision
Not selected
Selected
60
60
40
40
r = .20
Organization B
Test score
Low High
Decision
Not selected
Selected
95
95
5
5
r = .90
Conclusions
• Psychometric quality is important but not sufficient to ensure good test use
• Both psychometric quality and practical use of the test score should be included as criteria in the review models
• Start to discuss the validity definition in your test-review models
Thanks
● Questions?
Supplement
EFPA Version 3.3: November 2004
• When judging overall validity, it is important to bear in mind the importance placed on construct validity as the best indicator of whether a test measures what it claims to measure. In some cases, the main evidence of this could be in the form of criterion-related studies. Such a test might have an ‘adequate’ or better rating for criterion-related validity and a less than adequate one for construct validity. In general, if the evidence of criterion-related validity or the evidence for construct validity is at least adequate, then, by implication, the overall rating must also be at least adequate. It should not be regarded as an average or as the lowest common denominator.
top related