Drawing conclusions
In commenting these pictures, we should keep in front of us both Figures \ref{581719} and \ref{216672}. We can see that in order for the distribution for the PUNTAJE to change as we would like also the other variables have to change distribution. Some of these variables tell us something obvious, some others something interesting or sometimes they don't give us much information. When looking at these variables, keep in mind the relative scale. You should compare the two pictures relative to the same variable, not different variables into the same picture.Drawing conclusion.
- TOTAL: The behaviour is not very clear and smooth, but we could say that it clearly show that there is not much of a difference between these two distributions. In fact some bars go up and some go down, and they don't do it with a precise trend. Looking at the actual distribution instead of the difference plot, it is clear that there is not much happening. We could say then that this variable is not one we would need to act on in order to improve the performances of students.
- AMBIENTE ESCOLAR: Here we can see that the variable squeezed towards the centre. The bars shrunk towards the tail and enlarged in the middle. This might mean that performances seem to be better where the school environment is not too bad, but neither too good. Whether or not this is a useful information or worth more investigation, this is left to you.
- ESTU TRABAJA: There does not seem to be a very big change with this variable.
- ESTU ESTRATO: Here the change is clear and expected, but not as important as expected. It confirms that if we wanted to improve the performance of students, we would need to increase their socio-economic status. If that is possible, and how to realise that, is out of the scope of this model.
- REPROBACION MEDIA: Surprisingly, the effect on this variable is that it shifts the distribution towards higher rates of students repeating the year. It seems that repeating a year affects positively, even if not in an important way, the performance inthe exams.
- EXTRAEDAD: Conversely, this variable tells us something someway different, i.e. that performances are better in schools where students are not older than the age they should be when they attend a particular year.
- DESERCION MEDIA: Finally, here we can see that this variable is not particularly correlated with the performance of students. In fact, students that drop school do not participate to exams, and thus it would be strange for them to affect this result. One could think that being in a school with a higher rate of students dropping school would affect negatively the performance of other students. It does not seem the case according to the data available to us.
Conclusion
Strong focus on:
Potential
Importance of collecting right data depending on the question.