- guardian.co.uk, Tuesday October 31 2006 00.01 GMT
Characteristics of the sample
· Opinionpanel Research surveyed 22,334 first and second year members of their "Student Panel". Panel members comprise students who choose to sign up to participate in surveys, for which they receive a contribution to an Amazon Gift Certificate for each survey undertaken. No quota or other controls are set for membership of the panel, though quotas may be set subsequently by Opinionpanel Research for specific surveys. No quotas were set for this survey. As will be seen from the below, where appropriate, weightings have been used in the analysis in order to enable comparisons to be made.
· The achieved sample contained 14,666 full-time first and second year undergraduate students in English universities. Some data was collected from part-time and postgraduate students and from students at Scottish, Welsh and Northern Irish institutions but this was not used.
Subjects
· Combined studies are not shown (because all students in the Hepi survey were coded as belonging to a simple subject)
· Veterinary sciences has been combined with agriculture and related studies (owing to small numbers in each category)
Detailed analysis of total workload (teaching and private study) by institution and subject
· A minimum level of response has been enforced at both subject and institution levels for the results to be treated as reliable and therefore used in the analyses. This has been necessary because students studying the same subject at the same institution do not generally return the same number of hours, either because of errors in their reporting, or because of variations in provision, with different options or programmes within the same subject.
· In detail, for an institution to be identified in the subject-level analysis, we required at least 5 responses each from first and second year students (though in fact the mean number of responses across all subjects and all institutions was 26). It should be noted therefore, that even with 20 respondents a mean value of, say 14 hours, should be thought of as lying between something like 11.5 and 16.5 using the conventional 95% confidence intervals. However, it should also be noted that other than in the case of some outliers, in almost all cases the majority of responses are clustered around the mean.
· For a subject to be reported at all, we required at least five institutions to meet those requirements in that subject. So although we carried out analysis of 17 subject groupings, only 15 are reported on (at institution level - there are other reports at higher levels); and the number of universities for which data are reported ranges from 5 to 51. Two subjects - architecture, building and planning as well as veterinary sciences, agriculture and related subjects - did not meet the thresholds required for reporting.
