test validity

test validity - Transcript

TEST VALIDITY TEST

Degree to which a test measures what it claims to Degree measure Its not self correlation of test rather its correlation with some outside independent criteria which are regarded by experts as the best measure of trait ability being measured by the test

Validity like reliability is a matter of degree not an Validity all or none property a test cannot be either perfectly valid or not valid at all its usually more or less valid Validity is concerned with generalizability when generalizability when a test is a valid one its conclusion can be generalized in relation to general population

Types of Validity Types Content or Curricular Validity Content Criterion Related Validity Criterion Construct Validity Construct

Content Curricular Validity Validity

Also known as Intrinsic validity
When a test is constructed so that its content of term measures what the whole test claims to measure Such a validity is needed in Psychometric tests to to measure how well examinee has mastered the special skills or a certain course of study

Content validity is examined in two ways Content Expert Judgment Statistical Analysis Statistical

Face Validity is often confused with content validity but in strict sense its quite different FV but FV refers not to what test actually claims to measure not but to what it appears to measure superficially but When test items look valid to grp of examinees the test is said to have face validity

Express purpose of face validity is to establish face rapport secure cooperation because when test items do not appear to be valid to examinees they may not cooperate in responding and may give irrelevant answers

Criterion Related Validity Validity

It s a very common and popular type of validity and It s
is obtained by comparing or correlating test scores with scores obtained on criterion available at present or to be available in future There are two sub types of criterion related validity There Predictive and Concurrent Validity Predictive Concurrent

Predictive Validity Predictive

Here a test is correlated against criterion to be made criterion available sometime in the future Test scores are obtained and then a time gap of months Test or years is allowed to elapse after which the criterion scores are obtained then these two scores are correlated and obtained correlation becomes index of index predictive validity For example in industry mgt wishes to select SEs who For can exhibit best performance on job so mgt is administering tests of communication skills for selection of SEs

Obtained scores on CS may be correlated after 6 months a year against their average performance as measured in terms of units of products services sold If obtained correlation is high CS test is said to have If high predictive validity Predictive validity is needed for tests which include Predictive long range forecasting of success in industry

Concurrent Validity

Similar to predictive validity except that there is no time gap in obtaining test scores and criterion scores The test is correlated with a criterion which is available The at the present time scores on newly constructed present JS test may be correlated with scores obtained on JS already standardized test of JS and resulting JS correlation is the indicator of concurrent validity If correlation is too high then it may be said that new test is needless duplication of previous one

Concurrent validity is most suited to tests meant for Concurrent diagnosis of present rather than for prediction of diagnosis future outcomes For the same test predictive validity is usually For lower than concurrent validity coz degree of association between the test and the criterion decreases overtime overtime Similarly if the concurrent validity of a test happens concurrent to be zero then its predictive validity is most likely zero then predictive to be zero or close to it zero

Whether a test constructor is calculating predictive or Whether concurrent validity he she is faced with one vital problem of selecting an appropriate adequate criterion and this problem is more acute with predictive validity so obtained validity coefficient predictive be corrected for attenuation Correction for attenuation is done with the help of Correction following formula which takes into account unreliability measurement errors or chance errors in in test as well as criterion rc rxy r ttx rtty

Where rc correlation b w test x and criterion y corrected rc corrected for attenuation rxy obtained validity coefficients rttx rxy reliability coefficient of test X rtty obtained reliability rtty coefficient of the criterion Y So if reliability coefficient of test and criterion are 0 87 and So 0 80 and the validity coefficient b w test and criterion is 0 50 the corrected validity coefficient will be 0 60

Construct Validity Validity

1st time it was introduced in 1954 in Technical 1954
Recommendation of APA since then its been frequently APA used by measurement theorists Also known as factorial trait validity trait A construct is a sort of concept which is normally proposed construct with definition is related to empirical data Examples of constructs anxiety intelligence extroversion anxiety neuroticism stress work motivation job satisfaction and the likes

The process of validation involves following steps The Specifying the possible different measure of construct Specifying Determining the extent of correlation b w all or some of Determining those measures of construct Determining whether or not all or some measures act Determining as if they were measuring the construct

Convergent and Discriminant validation are important Convergent are for establishing satisfactory construct validity When a test correlates well with its expected referents When the process is known as convergent validity and convergent When a test correlates poorly or not at all with measures When with which it should not because it differs from those referents or measures the procedure is called discriminant validity discriminant

Factors Influencing Validity Factors

Length of the Test Length
Range of Ability or Sample Heterogeneity Range Ambiguous Directions Ambiguous Socio Cultural Differences Socio Cultural Addition of Inappropriate items Addition

Relation of Validity to Reliability Reliability

A test having poor reliability is not expected to yield high test
validity validity is dependent upon reliability but this prediction is true for the homogeneous test only true If a test is heterogeneous such as biographical data If blank validity may be high even without high reliability particularly the internal consistency reliability this is coz in heterogeneous test each part measures an independent function

Reliability is the sufficient but not necessary condition for Reliability validity validity of a test may be higher than the reliability but not higher than the index of reliability the Index of reliability sets the maximum limit of correlation Index that the test can yield with its true measures the true measures being the criterion A test constructor should not always aim at having high test reliability and high validity in the same test the requirements for high reliability and high validity are opposite to each other

High reliability requires items of equal difficulty and High requires high inter correlations b w the items whereas the high high validity requires items of different difficulty values and requires low inter correlations among items But if inter item correlations range from 0 10 to 0 60 one 0 10 can expect to have both reliability validity to a satisfactory degree Trucker 1946



NORMS of PSYCHOMETRIC NORMS INSTRUMENTS

It s average performance on a particular test made by a It s standardization sample a sample which is true representative of population takes test for express purpose of providing data for comparison subsequent interpretation of test scores For adequate representation the sample must include a For cross sectional representation of different parts of the population In order to compare raw scores with performance of In standardization sample they are converted into Derived Scores Derived

Types of Norms Types
Derived scores are divided into four common types and Derived
depending upon each of these four scores there are four types of norms as Age Norms Age Grade Norms Grade Percentile Norms Percentile Standard Score Norms Standard

Age Norms Age Its an average performance of a representative sample Its
of a certain age level on the measure of a certain trait or ability Such a norms are most suited to those traits or abilities Such that increase systematically with age

Grade Norms Grade Its average performance of a representative sample Its of a certain grade or class Such a kind of norms are mostly utilized in colleges Such schools but hardly in industry

Percentile Norms Percentile Its most popular common type of norms that are Its used in psychological testing A percentile norms indicate for each raw score percentile percentage of standardization sample that falls below that raw score It provides a basis for interpreting an indiv s score It on a test in terms of his own standing in a particular standardization sample

For PN to be meaningful it should be based upon a PN sample which has been made homogenous with respect to age grade sex occupation and other factors otherwise separate tables for percentile norms for age grade sex and occupation should be prepared Such a norms is easy too construct easy to understand Such that even an untrained person can freely use it but its not without limitations

Standard Score Norms Standard A norm which is based on a standard score is known as norm
a standard score norms and it s a superior form of derived score It has a fixed mean and SD and an individual score is fixed and interpreted in terms of SD unit from the centre of interpreted unit distribution There are several types of standard scores as Z Score T Score Deviation IQ etc Score

THANK YOU THANK