The profound and comprehensive accompanying research at CyberMentor provides insights into successful mentoring processes which are used for the continuous improvement of the program. The results are made available to other mentoring programs in scientific publications. The four most important components of the accompanying research are:
(1) longitudinal monitoring of program participants over at least three measurement points, and, in some cases, over several years,
(2) impact measurement by using several comparison groups,
(3) extensive data collection using standardized scales, and
(4) the combination of several data collection methods.

Some mentoring programs measure their success based on participant satisfaction. However, it is known that the success of the mentoring programs is often greatly overestimated due to a meresatisfaction survey. For this reason, CyberMentor is particularly interested in examining how certain participant characteristics change over the course of the mentoring.
To test the effectiveness of the program, several variables such as interest in STEM, confidence in one's STEM abilities, knowledge of STEM fields of study and careers, intention to study a STEM subject, STEM activities, and several aspects of the students' home and school environments are surveyed using standardized scales. To study their development, participants are followed longitudinally over several measurement points. Surveys are conducted using online questionnaires before the mentoring begins, during the mentoring year, and after the mentoring ends.
In addition, feedback questionnaires are collected to assess the mentoring round. Follow-up surveys are partially conducted with former mentees to record actual decisions regarding study and career choices.

The participants’ development is compared with three comparison groups. The first comparison group is a group of girls who have registered for the program but will not be admitted to the program until the next round due to limited places. These girls show a comparable interest in STEM as the participating female students. The other two comparison groups consist of one group of girls and one group of boys who have an average interest in STEM. The participants and the comparison groups receive the identical questionnaires at fixed times. All data are stored and analyzed in anonymized form. From a research perspective, the comparison of the participating mentees with female students in the comparison group with similar interests represents a special feature in the research design compared to existing research: It allows for an actual proof of impact of the CyberMentor program, as a comparison is made to female students with comparable prerequisites who, however, cannot benefit from mentoring.

Follow-up surveys with former mentees regarding study and career choices show the following results:

62 percent of former female participants later chose STEM majors and careers. This means they were twice as likely to choose STEM as female students in their respective age cohorts (31%).