Some Statistical Methods of Item Analysis.
Date of Submission
December 1984
Date of Award
Winter 12-12-1985
Institute Name (Publisher)
Indian Statistical Institute
Document Type
Master's Dissertation
Degree Name
Master of Technology
Subject Name
Computer Science
Department
Electronics and Communication Sciences Unit (ECSU-Kolkata)
Supervisor
NA
Abstract (Summary of the Work)
One of the main featuree of the development of payohology as a quantitative rational science in the poseibility of making roliable and valid observatione on the payahologioal oharaoteristios of interost. The main tool for making suoh observations is the objeotive test - an it 1s popularly kmown- waioh is being inorea- singly used in education and o ther fields of applicatione of peychology. This 1s owing to the ease of administre- tion, and, of evaluation and asses ment of the responses, that ia assoointed with euch a test. However, this tool ie as delionte an it is pevarfulj for, if oarelesely handled, it ia likely to lead the user astray. Thiat is why, the branch of item analyode of an objective test plays such a signifioant role in quantitative payahology.An important aspect of item analysis ie to find an index to measure the effieiency of an item, so as to lay dowm an objeotive oriterion for the inolusion of a trial 1 tom in a battery and to enable the items to be rank-ordere B0 that the best iteme aan be selected. (These problems are generally dealt wi th in terms of the proportion of correat reaponsee). The ariterion for the effioiency of an item 1e ite ability to dietinguish between good and bad subjeote and alno the consistenoy with which it does this in repéated trials. Item analysis esn be iproved, if, beeidee the acotraoy or otherwise of the responses, some other relevant charaoterietic of performance on the item, such as the responae tine, 1s observed. Purthermore, there are aituations in whieh items are needed to bring out the differenoes betwe en several populations regarding the tratt the item is intended to meae ure, for inatance, if feminini ty 1s the tradt, ono may be intereeted in items which bring out differencen be tween male and female populationa or if the 1 tems are to be dsed with a speoial group, one may be interested in those items that bring out differenoes between a control group and the special group. In suoh inatanoes, item analyeis would be based on a diserininatory anslyeis of the two groups.A problem of thie nature aroee when Rao (1963) was preparing direotiona testa for bilingual ohildren in order to relate their adjustment diffioul tdee to attainment in their second language. A eet of trial iteme wan adminie- tered to a oontrol group and the experimental bilingual group.The ability of a subjeat was moas ur ed not only by the aocuracy of the respanse, but also by his response time. However, the time taken for a wrong renponse was not neaningful, beaause in some oasen there were no suoh reapone and in others becanse there wore several different rong responses. Therefore, only the time taken for a correct response was noted, for each item and subjeet. The problem was to eelect itee which best disoriminated between the twe groupa.A solution to this problem is provided in Chapter 2 in which the two variables (after a suitable trane foraation of the proportion of responses) are combined linearly in euoh a way that diserimination between the two groupa i nexinised.
Control Number
NA
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.
DOI
NA
Recommended Citation
Krishnan, T., "Some Statistical Methods of Item Analysis." (1985). Master’s Dissertations. 362.
https://digitalcommons.isical.ac.in/masters-dissertations/362
Comments
ProQuest Collection ID: http://gateway.proquest.com/openurl?url_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:dissertation&res_dat=xri:pqm&rft_dat=xri:pqdiss:28843445