gms | German Medical Science

Introduction: Mentoring is important for a successful and satisfying career in academic medicine [1]. At our faculty we offer different methods of finding the right mentor. Increasingly, we are also using online systems for access to a larger quantity of profiles to increase a student’s chance to find their individual perfect match. This creates a new challenge for mentees and mentors, because the quality of a profile has a big influence on the matching decision but does not necessarily correlate with the quality of the potential mentoring relationship. But what contents of a profile’s text are of interest for students looking for a suitable mentor and therefore affecting their decision?

Materials and Methods: The contents of 147 anonymously extracted profile texts were categorized into main- and subcategories by two independent individuals using thematic content analysis. Randomly selected medical students differing in age, sex, mentoring status and years of medical training additionally ranked those profiles on a Likert scale ranging from 1 to 10 based on the following question: “How do you assess the quality of the profile with regard to the selection of a suitable mentor (regardless of your personal preferences)?”. To assess the quality of profile texts in terms of their relevance to students on a larger scale, we identified seven super categories based on the results and an earlier conducted survey by Dimitriades et al. where 120 students from the clinical years were asked which topics they want to discuss with their mentor [2]. Those super categories were validated by chi-squared test. For the 25% highest and lowest ranking profiles we then registered if the attribute from an earlier established subcategory was present in the text in a binary system. By conducting chi-squared tests on the generated data we statistically analyzed differences. In all analyzes Cohen’s kappa was used to verify inter-rater reliability.

Results:

1.

We succeeded in generating a category system with 5 main categories and a total of 74 subcategories. These categories enable us to systematically record the content and formalities of individual profiles in a standardized way and then evaluate them with the highest possible objectivity.

2.

The presence of 3 or more “super categories” in profile texts clearly correlates with a positive evaluation by students (p<0,01).

3.

Of all subcategories generated 10 were associated with positively- and 4 were associated with negatively-rated profiles (p<0,01).

Conclusion: In our opinion, this classification helps us to understand how profile texts are composed and moreover which aspects are important for students in order to choose a suited mentor. Furthermore, we can build a structured profile mask for the creation of new profiles to improve overall profile quality of our mentoring program.