Dear members,
I am asking for your kind help on a data analysis for my thesis.
Main goal of the analysis: I want to measure the difference of learning outcome for two different learning method groups. Also, I want to see if visual literacy has a moderating effect on this relationship.
A short overview of the model and dataset:
- I have a categorical Independent variable, which is learning method. It has 2 levels: text ot video. It is used as a dummy variable in the analysis (0=text, 1=Video)
- The dependent variable is learning outcome. It is measured with a 10 question recall test, scores range from 0 to one. The score shows how many percentage of the questions did the participant answer correctly.
- The visual literacy acts a potential moderator. It is measured by an assessment test with 19 questions. Each question is is scored from 1 to 5. The final viual literacy of a participant is measured by the mean of the 19 questions. ranging from 1 to 5.
- Additionally, I have 4 control variables: age, gender, educational level and English level. Gender, education and english proficiency are all dummy coded. Age is already a numerical variable.
I constructed a correlation matrix for the variables. None of them was significant, and there is only a negligible correlation between them anyways. I am not sure if it makes sense to run the hierarchical multiple regression where the first model includes the control variables, and in the second model the predictor (learning method) is entered.
However, I did run the multiple regression and I got the following results:
Model R Square Adjusted. R Sq. R Sq. Change F Change Sig. F Change
1 .126 .059 .126 1.892 .080
2 .127 .050 .001 .092 .762
What should be the takeway/interpretation of these results, and is it possible to run a moderation analysis even though there seems to be no relationship between learning method and learning outcome?
Thank you very much for your help in advance
My recommendation is to talk to your advisor who I assume approved this thesis project. But in the meantime notice that your DV IS NOT CONTINUOUS and thus is not suitable for ordinary least squares regression. You might look at Rosner Fundamentals of Biostatistics which covers what to do in these cases. David Booth
After you talk to your advisor look at the attached material. Best wishes, David Booth
Yes, I would do that if your reserch questions asks for adjusted results. These give possible supression effects. However, also predictors with different mening after the partializing (not that much in this case). Intetaction effects differentiate main effects, with different regression values over different levels inn the moderator. Rolf
Hello! I am interested in your work! I am engaged in multifactor modeling of the educational process at a university.
You can send me part of your data tables in Excell format and I will create multifactor computational models that solve direct and inverse problems. I propose to publish a joint scientific article.
Victor Abrukov
- Badges
- Science topic
- Similar topics
- Linguistics
- Philology
- Language
- English