For several years, in our country, it is most frequently realized the need to detect, treat, analyze and interpret data and information; the need to have therefore measuring instruments suitable for the detection and the organization of all those elements through which certain phenomenon occurs. One of these tools, even if not the most used, is represented by the tests. Each year thousands of students can see opened or precluded the doors of the university by the results of a test. To make all these things there must be some specialists, as rightly pointed out Bottani (2011); to create good tests it is necessary skill and practice, which cannot be replaced by a rapid reading of instructions (Lucisano 2011). This complex and articulated operation becomes even more difficult when tests are made to select, and, as in our case, to foreclose the possible inclusion to a degree course. It is thus necessary is to verify the quality of the entry tests used in order to check their validity and reliability and to suggest any change to make them congruent to the purposes for which they were built. This means to pay attention not only to the content, the minimum knowledge, but, "above all", the part does not immediately visible and still partially neglected related to the "technical construction" of the test, that is, its accuracy, its clarity, and the perception derived. The "quality, validity and reliability" of the questions, however, are closely related to the answers provided by students. The item psychometric examination of the test is, in fact, an important step in the test’s construction process. The results obtained through the item analysis will make possible the choice of questions with parameter values able to increase reliability requirements of the test and determine the level of difficulty and the discriminating capacity of it (Barbaranelli-Natali, 2005). This process of analysis can be carried out through two different approaches whose characteristics and assumptions are strongly differentiated: Classical Test Theory (CTT) and Item Response Theory (IRT). In the first one, the answers obtained are subjected to statistical analysis to highlight the items too easy or too difficult; if there are distractors which have a percentage of choice too low or too high; to check if the items are able to discriminate the candidates more competent than those less competent; to perform the appropriate calibrations in the tests. The IRT, on the other hand, allows to evaluate the performance of the subject as a function of a latent abilities through the specification of a statistical model-mathematician which allows you to reach not only the evaluation of the performance of the individual, but also the characteristics of each question. Having said that, in the research project of the present paper, the research team has decided to analyze and correlate the results of entry tests to the degree course in Primary Teach Education over the last five years by using both approaches but focusing the attention on the potential of IRT models with particular reference to the Rasch model which allows to obtain objective and universal measures transcending the context of the measurement and the instrument used.
THE ITEM RESPONSE THEORY (IRT) FOR THE ANALYSIS OF THE UNIVERSITY ENTRY TESTS
TAMMARO, ROSANNA;MARZANO, ANTONIO;NOTTI, Achille Maria
2012-01-01
Abstract
For several years, in our country, it is most frequently realized the need to detect, treat, analyze and interpret data and information; the need to have therefore measuring instruments suitable for the detection and the organization of all those elements through which certain phenomenon occurs. One of these tools, even if not the most used, is represented by the tests. Each year thousands of students can see opened or precluded the doors of the university by the results of a test. To make all these things there must be some specialists, as rightly pointed out Bottani (2011); to create good tests it is necessary skill and practice, which cannot be replaced by a rapid reading of instructions (Lucisano 2011). This complex and articulated operation becomes even more difficult when tests are made to select, and, as in our case, to foreclose the possible inclusion to a degree course. It is thus necessary is to verify the quality of the entry tests used in order to check their validity and reliability and to suggest any change to make them congruent to the purposes for which they were built. This means to pay attention not only to the content, the minimum knowledge, but, "above all", the part does not immediately visible and still partially neglected related to the "technical construction" of the test, that is, its accuracy, its clarity, and the perception derived. The "quality, validity and reliability" of the questions, however, are closely related to the answers provided by students. The item psychometric examination of the test is, in fact, an important step in the test’s construction process. The results obtained through the item analysis will make possible the choice of questions with parameter values able to increase reliability requirements of the test and determine the level of difficulty and the discriminating capacity of it (Barbaranelli-Natali, 2005). This process of analysis can be carried out through two different approaches whose characteristics and assumptions are strongly differentiated: Classical Test Theory (CTT) and Item Response Theory (IRT). In the first one, the answers obtained are subjected to statistical analysis to highlight the items too easy or too difficult; if there are distractors which have a percentage of choice too low or too high; to check if the items are able to discriminate the candidates more competent than those less competent; to perform the appropriate calibrations in the tests. The IRT, on the other hand, allows to evaluate the performance of the subject as a function of a latent abilities through the specification of a statistical model-mathematician which allows you to reach not only the evaluation of the performance of the individual, but also the characteristics of each question. Having said that, in the research project of the present paper, the research team has decided to analyze and correlate the results of entry tests to the degree course in Primary Teach Education over the last five years by using both approaches but focusing the attention on the potential of IRT models with particular reference to the Rasch model which allows to obtain objective and universal measures transcending the context of the measurement and the instrument used.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
