I'm an undergraduate university student conducting research on the effects of internet use on depression and anxiety among undergraduate Ghanaian university students: Mediation role of Internet addiction and fear of missing out. My sample size is about 848 undergraduate Ghanaian students belonging to diverse fields of study and all in different levels of their programme of study. Unfortunately, after collecting about 853 responses from this population, I realized I had not used standard scales to collect responses on the psychological symptoms in my research. i.e., Depression and Anxiety severity of these students. Instead of using standard scales like the PHQ-9 for depression and the GAD-7 for Anxiety, I used my own fabricated scales, and now I'm stuck. Though my scale has some questions in common with the PHQ-9 scale, it's not really the same thing. Please help me out with some advice from your experiences in research. Thanks in advance.
This has two serious problems. Without standardised scales, you won't be able to compare your results with the current literature on the subject.
The other big problem is that you are going to have to do psychometrics on your scales to show that the scores follow the characteristics of unidimensional scales, and that the items scale together.
I'd recommend Mokken scaling, which is nonparametric and assembles scales from an item pool based on a strong definition of unidimensionality. You can do this in R or Stata. The R package mokken has excellent help resources and worked examples. But you have a lot of work ahead of you.
Don't blame yourself here. An undergraduate who gets to do large scale research using non-standard methods shows a failure of supervision and oversight. This problem should have been spotted way before the survey ran. Sadly, you are now learning the things that the people who are supposed to be teaching you didn't tell you.
Good luck
I completely agree that this is a problem that should have been caught by the person advising you before you even started the work.
Apart from following Dr. Conroy's excellent suggestion, you may also want to think about the purpose of this paper. If you wish to submit it for publication, then certainly you would need to be able to compare your results to existing literature and do the needed statistical analysis.
But maybe it is just a learning experience as part of your studies. You can discuss the noted limitations of your scale and the accompanying results and look ahead to another study which may involve your scale and the better standardized scales. This suggestion will depend on what the requirements were for the study you undertook and whether cutting your losses will still fulfill the requirements.
As another thought (and I am not a statistician), you could administer your scale and the PHQ-9 to two other matched samples of students and report on the correlation of scores between the PHQ-9 and your scale. Even if you do not get a lot of data points to have a significant correlation, you may be able to discuss some qualitative differences or similarities based on item responses.
Best wishes
I would like to explore with you the opportunity to present your research as a mixed methods research where you first constructed a questionnaire that is relevant to Ghanian students today, and a wider wellness concept that is an appropriate measure for FOMO. Later on in recommendations you can say that if someone wants to measure only depression and anxiety, they could use standardised tests.
For the more qualitative part, is there a way you can describe how you constructed your questionnaire? Did you include a number of questions about other variables than only depression and anxiety, e.g. general feeling of loneliness or having friends etc etc? You tried to operationalize FOMO and constructed questions for that, it is an experiment. Did this cluster come from only your own thinking, from the literature; or was there some kind of pilot with Ghanian students that produced that particular set of questions? Interviews, group discussions? Try to remember what did and how you did it. Make that process transparent.
If so, could you make a case that these set of variables are relevant to student life in Ghana today, and that you measured wellness or contentedness or perceived quality of life as an indication for FOMO, relevant to Ghanian students today?
The next challenge is to relate your outcomes from the sample of students to FOMO and try to validate why your list of questions was a good indicator. Or not! That can go three ways:
1. Questionnaire for FOMO was a good enough indicator
2. Questionnaire for FOMO needs improving
3. Standard questionnaire can be used if you want to measure anxiety and depression are linked to FOMO.
You donot have to copy other people’s methods. That does not lead to innovation. What you have to show to examiners is which choices did you make and what were the reasons for that; how can you you substantiate your choices and decisions. You did original work as a researcher, so try to explain that in methodological terms and show how it fits into methodology. That requires more thinking than immediately diving into statistical methods. Maybe statistics will be the only way forwards, but you can try to show how you constructed a set of questions and if that set is trustworthy.
Success!
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs