Page 133 - DEDU504_EDUCATIONAL_MEASUREMENT_AND_EVALUATION_ENGLISH
P. 133

Unit 10 : Item Analysis : Item Difficulty, Discrimination index, Effectiveness of Distractors


            According to Harper and Chauhan, “ Test analysis is a general term which covers a wide range  Notes
            of statistical and analytical techniques which may be applied for the purpose of improving tests,
            examinations, quizzes and other mental measurement devices.”
            —  Harper, Proceedings of the 62nd Indian Science Congress, Part 2nd Presidential Address (1974)
            Criteria for Item-analysis : The purpose of item-analysis is to select suitable and good test items
            from the many items the test-maker has made. For this, each item is evaluated. Besides, it is
            determined on the basis of a certain criterion which item is good and should be selected, and
            which item is not good and should not be selected. This decision is taken on two bases :
            1.  Difficulty index, and
            2.  Discrimination index or validity index.
            If the difficulty index and discrimination index of an item are of the desired level, then that item
            is selected. There is no definite formula or principle to determine what should be the difficulty
            and discrimination indices of an item. This depends on the objective of the test and the insight
            and vision of the test-maker. If a test is survey-type, then the questions with difficulty index
            between 20-80 and with discrimination index. 3 or more, are considered suitable. If a test has
            been made with a view of selection, then its difficulty index should be between 10-50 and
            discrimination index should be about .4 or .5. To calculate difficulty index of a question, it is
            found how many people in the sample have solved the item correctly, and then this is simplified,
            and if it is lower, then it is made more difficult, Besides, each item’s discrimination index is
            correlated with the entire test. It is assumed that the whole test measures that trait for which it
            has been constructed. Therefore, if the correlation of an item is more than the whole test, then it
            is understood that item measures that trait, that is, the item is valid.





                    Item-analysis is also called ‘evaluation of the test’ or ‘try-out stage of test construction’.
                    This is an extremely important step of the test, and no test can be completed without this
                    process.

            10.3 Process of Item-Analysis

            To undertake item-analysis, at first the test is administered to a sample of those individuals or
            students for whom it has been constructed. After that, all test papers are marked and the following
            processes are sequentially followed :
            1.  Test-papers or answer-sheets are arranged in the ascending order of scores.
            2.  To prepare item-analysis chart. In this chart, at first, the serial number of items is written in
                the columns. On the left, the names or serial number of students are written in rows. Thus,
                the chart is divided into many rows and columns, and when lined, they make several cells.
            3.  After this, the answer-sheet of each student is taken and it is seen whether he has solved it
                correctly or not, and a right tick mark is given for the right response and a cross mark is
                given for the wrong response. Thus, all test items are marked on the chart. At the end of the
                chart, two rows are made where the difficulty index and discrimination index are written.
            4.  In each column of the chart from top to bottom, the right tick marks are counted and are
                changed into percentage in order to write them in the form of difficulty index. Thus, difficulty
                index is calculated in respect of all test items.
            5.  To calculate discrimination index, we divide the analysis chat into almost three equal parts
                from top to bottom.




                                               LOVELY PROFESSIONAL UNIVERSITY                                    127
   128   129   130   131   132   133   134   135   136   137   138