Page 79 - DEDU504_EDUCATIONAL_MEASUREMENT_AND_EVALUATION_ENGLISH
P. 79

Unit 6 : Validity – Types , Methods and Usability



                                   X                                       Y                         Notes
                             Vocabulary Test           25%            Grade Agerage

                Now, supposing college students are also administered the intelligence test in addition to the
                vocabulary test, and the correlation between intelligence test and grade average is .5 and 0 by
                the vocabulary test. The intelligence test explains 25% variation of average variance. Thus,
                vocabulary and intelligence tests explain 50% variance together.
                            X                         Y                       X
                                                                               2
                       Vocabulary Test  25%      Grade Agerage    25%   Intelligence Test

                This joint variance which is described by predictor tests, is used in calculating correlation
                coefficient. This variance is equal to the square of multiple correlation. In this examination, the
                value of multiple correlation will be equal to .50 or .71, and it will be displayed by R . In its full
                                                                                 123
                form, following is the formula for multiple correlation :
                R
                 1.234.....n
                This symbol displays correlation of all tests 2, 3, 4,.... with test-l.
                Thus, if we link several tests in the test battery, whose mutual correlation is zero and they are
                correlated with the criterion, then multiple correlation will be equal to the total sum of squares
                of the correlation of criteria of predictors, that is :
                R 2  ...n = r 2  + r 2  +.....+ r 2
                  1.23   1.2   1.3   1.n
                General Definition of Multiple Correlation :
                “Multiple correlation is the correlation between scores on one variable and the composite
                scores on several other tests while these scores are optimally weighted.”
                                  Table : Semesters Completed in Percents
                      Decile     2       3        4       5        6       7        8

                        X       97       94      93       91      89       88      86
                       VII      93       86      84       80      78       77      77
                        VI      91       81      78       76      72       71      71
                        II      84       70      65       63      60       58      57
                        I       80       66      57       50      49       48      45

                It is clear here that the students who are given X item (maximum), there is possibility that 70%
                students will complete the semester; while in the lowest decile group, there are only 45% students
                who will be able to pass the semester.
                Prediction can also be displayed by another method, called bar diagram. By it, the predictive
                ability of a test is calculated. Under this, a test battery which has been used for selecting workers
                for pilot training, the percentage of students on stanine has been displayed. It is clear from it
                that on the basis of test, the individuals who obtain 9  stanine are least likely to be discharged
                                                          th
                from their work (only about 4% pilots were discharged due to inefficiency), while those obtaining
                1  stanine, 70% of them were discharged due to inefficiency. Thus, explanation can be given on
                 st
                the basis of other stanines.
            7.  Cut-off Score Method : Cut-off score is a test score which is used as the demarcation point for
                accepting or rejecting students. For example, supposing the following scores indicate adjustment
                in some adjustment list, and we consider 9 as the cut-off score, then it would mean that the
                students who have scored 9 or less on the adjustment inventory will be accepted, and those




                                               LOVELY PROFESSIONAL UNIVERSITY                                     73
   74   75   76   77   78   79   80   81   82   83   84