/* Style Definitions */
table.MsoNormalTable
{mso-style-name:"Table Normal";
mso-tstyle-rowband-size:0;
mso-tstyle-colband-size:0;
mso-style-noshow:yes;
mso-style-priority:99;
mso-style-parent:"";
mso-padding-alt:0in 5.4pt 0in 5.4pt;
mso-para-margin:0in;
mso-para-margin-bottom:.0001pt;
mso-pagination:widow-orphan;
font-size:12.0pt;
font-family:Cambria;
mso-ascii-font-family:Cambria;
mso-ascii-theme-font:minor-latin;
mso-hansi-font-family:Cambria;
mso-hansi-theme-font:minor-latin;}
This paper discusses possible uses of student course evaluations on a pair of courses developed to comply with the CDIO concept. It is seen that both similarities and differences in the evaluations can be found. These can in part be used to assess if the CDIO concept has been implemented as it was intended and possible adjustments can be suggested.
The data consist of 33 observations with full information on 8 general course evaluation questions on each of the two courses. The data has been collected over three years (2008, 2009, and 2010). This makes it necessary to consider methods which are able to handle possible differences between years.
We illustrate different ways to analyse the associations between the two courses by utilizing such data. Inferences about the mean differences between the courses are performed using analysis of variance techniques. In this context they may be considered as generalisations of the paired t-test. The generalisation to an analysis of variance makes it possible to handle differences between years. Inferences about the correlation structure of the data are performed using so-called canonical correlation analyses. A possible difference between years of the evaluations makes it necessary to consider adjusting the data for the year effect.
We find that one course generally is evaluated as more satisfactory than the other on five of the questions. Also we find a very strong effect of year, indicating the need to remove the year effect before proceeding with the canonical correlation analysis. The canonical correlation analysis is only significant at a 10% level of significance for these data and resulting associations must therefore be interpreted with caution. The interpretation results in a combination of evaluation questions for one course which correlate well with another combination of evaluation questions for the other course.