Popolazione potenziale e reale nelle indagini sulle opinioni degli studenti sulla didattica universitaria: dati mancanti casuali o non casuali?

The definition of the correct population, as well as the evaluation of the results with incomplete coverage, are crucial in student's opinion surveys on university teaching. In fact, they can strongly affect comparative analysis on different degrees and faculties. Thus the rate of students' participation to lectures - as well as the reasons of students' behaviour on these matters - must be considered when lecture attendance is not compulsory, since they can cause bias in the estimates of teaching quality. To tackle the problem, from 2004-05 the survey conducted at Turin University asks respondents to estimate the percentage of students that, while attending the first lessons of a given course, have since quit them. This paper tries a first analysis of results of this experiment, with regard to both correctness of students response and possibility of ex-post reconnaissance of the type of missing data through a conditional comparison of the collected opinions.