A New Methodological Framework for Project Design to Analyse and Prevent Students from Dropping Out of Higher Education
Round 1
Reviewer 1 Report
Abstract should be rewritten stating the problem, solution provided and results.
Literature review/ related works can be added as a separate section.
Overall presentation is good
All the figures and tables can be explained in detail.
Author Response
The authors would like to thank the reviewer for their comments and suggestions. The changes made to the paper to address them are described below:
“Abstract should be rewritten stating the problem, solution provided and results.”
The abstract has been rewritten as suggested.
“Literature review/ related works can be added as a separate section.”
We have included section 2 to deal with this suggestion and modified section 1 subsequently.
“All the figures and tables can be explained in detail.”
We have reviewed the paper and checked that all information in tables and figures is commented on in the text. Due to the format, sometimes a table or figure does not appear right next to the text, but all figures and tables are referenced from the text.
Moreover, the text of the paper has been revised to improve the English, some typos have been corrected.
Reviewer 2 Report
This is an interesting theoretical paper describing the process required to use big data in solving the dropout problem in universities. However, while the paper is based on previous literature, it doesn’t provide new information.
As you rightly point out on Line 273 - the causes of the problem should be analyzed, starting from the premise that the solution should be based on an adequate analysis of the data stored in educational computer systems and taking as a reference the benefits provided by Big data and data analytics techniques.
In addition on Line 374 you allude to the fact that sufficient information is needed to identify potential dropouts and solutions.
What I feel this paper lacks is a discussion of the challenges one would face in implementing this type of system. You do an adequate job in describing the steps in the process but before the conclusions, you need a section that describes how someone might actually implement it and the challenges they would face. As the paper reads now there isn’t anything concrete for the reader. It needs something they can do to prepare for this to happen.
This discussion would be primarily based on the need for accurate data. The literature has many examples of how big data is used in business. The challenge in education is the fact that schools do not capture data on individuals beyond academic achievement and basic demographics. To use big data to predict and help alleviate the dropout problem using this process, additional information would need to be gathered about students. Then the ML algorithms could produce predictions flagging students as potential dropouts.
The dropout problem is very complex. the problem we have is that we might identify possible factors that might contribute to the dropout problem but we don't track crucial information about individuals that might help us predict that someone is likely to drop out. for example, you note that pregnancy is a potential factor that might cause someone to drop out. However, not all students who become pregnant dropout. There are other interacting factors. What is their health situation, financial support, family support, personal interests, and career goals,? In addition, there are likely other factors that will affect an individuals decision to dropout that we don't know about. These are not simple dichotomous checkbox variables. This means we need accurate measures of these attributes. Unfortunately many of these factors will change over time, sometimes slowly other times instantly. The biggest challenge in implementing your proposed process is that we simply don't track the data we need to implement these types of systems. We have lots of data, but we don't know if we have the correct data or whether the data we have about individuals is accurate. You need to discuss this limitation and perhaps suggest what can be done to make this work in the real world.
Author Response
The authors would like to thank the reviewer for his/her comments and suggestions. The changes made to the paper to address them are described below:
Comment: "What I feel this paper lacks is a discussion of the challenges one would face in implementing this type of system. You do an adequate job in describing the steps in the process but before the conclusions, you need a section that describes how someone might actually implement it and the challenges they would face. As the paper reads now there isn’t anything concrete for the reader. It needs something they can do to prepare for this to happen. ...
You need to discuss this limitation and perhaps suggest what can be done to make this work in the real world."
Answer:
We have included section 4 to address this suggestion, where we discuss the challenges and limitations of our approach in the real world.
Moreover, the text of the paper has been revised to improve the English, some typos have been corrected.
Round 2
Reviewer 2 Report
I believe the authors have addressed my suggestions and concerns to some extent. While I have no objections to publishing the paper as a theoretical opinion. One of the biggest challenges to what the authors suggest, which was not meantioned, was the ability of institutions to actually measure student characteristics and get access to the important information they felt might be needed. One additional issue is the fact that in social sciences, models rarely are accurate for the majority of students. There are simply too many important factors and because students have free will (agency) they often change their perspectives, interests, and learning goals. The authors should explore this issue a bit more in the discussion.
the ideas the paper presents are not new or novel, several institutions do a form of this. However, the reason we don't do it well is the lack of accurate measures of the important variables.
