The Youth Start – Entrepreneurial Challenges used the Random Control Trials (RCT) methodology as our experimentation design since we have full control over the educational “treatment” and are able to randomize which students experience it. All participating schools used the educational modules at some point, except in some cases where the schools were part of the “pure control group”. The questionnaires used to evaluate the programme were first developed through the ASTEE project, which developed measurement tools to assess the impact and the influence educational programmes had, not only on participants’ entrepreneurial self-efficacy, attitudes, mindset and intentions, but also their future intention to work in innovation-oriented professions. During this project, we tested the effects of an extensive and an intensive version of Youth Start – Entrepreneurial Challenges at primary and secondary schools in four countries. Since it is particularly important to test different moderators in order to inform policymakers in the most efficient way, we tested the programmes in different types of schools and different educational levels.
There are many ways the collected data in Youth Start – Entrepreneurial Challenges can be processed and analysed. Since the educational intervention has been randomly allocated it might be enough to only use the data collected at the end of the school year and analyse on which variables the students in the experiment group differ from the students in the control group. The matched data from the teacher survey could be used as control variables or as moderators in order to investigate the influence teachers had on their students. The analysis should include the school effects as well. The simple way to do this is was to nest the data at school level or even at the class level and account for these clusters in the analysis.
In experimental settings where the educational intervention has been randomized, only the variables that can be expected to have an influence on the treatment effect should be included and the analysis should investigate whether or not these variables interact with the educational intervention. Including additional control, variables would only lower the statistical power and it would be unnecessary to include these since their occurrence in the groups should be at random. Controlling for baseline data can be important since Likert-scales, unlike true continues variables, have a natural ceiling. If a respondent has a max-score in one of the dimensions of the baseline test they will not be able to increase their score further.
In order to test whether students have acquired the relevant core competencies, we offer several forms of assessment for all our challenges. An essential feature of our model is the self-assessment to be done at the end of the challenge. The students reflect on their own behavior, using questions such as: Have I reached my goals? What role did I play in the group? How can I apply what I learned to my daily life? This self-assessment helps students assume responsibility and makes them aware of how important it is that they actively contribute to projects.