We are investigating the impact of online learning environment on sustained attention and student engagement among undergraduate students (Year 1-3).
We had the participants watch a lecture video and answer multiple choice questions related to the content of the video. We also had them fill an engagement questionnaire which had questions about their involvement with peers and various university related activities.
Hypothesis I - 1st year students are able to focus more over a long period of time than 2nd and 3rd year students.
Hypothesis II - 1st year students actively participate and are more engaged with peers than 2nd and 3rd year students.
My questions are:
1. What statistical test would be most appropriate?
2. How do I set a value to prove significance...for example they had 13 MCQ questions, how do I know what number of questions they need to have answered correctly in order to say they paid more attention?
I'm very lost, i'd appreciate your help and any remarks that could provide me some clarity.
Hi Preet I am not a statistician but I think we need a bit more detail.
'significance' statistically is usually we work with a p value of less than 0.05 but I think what you are asking is significance of paying attention, you will likely have to make a value up (hopefully that you can back up) as to what you think are different levels of attention. If this was two groups you could probably do a t-test for significant differences.
Like I said I think more detail is needed.
Your description sounds like you are conducting a survey, or do you have the students divided into different groups with different treatments? The short answer to your first question is that your design will dictate your statistics if you intend this as an experiment.
As Gurvinder said, you need to provide more information.
If distribution of the data is not normal or if one is not sure about the distribution, it is safer to use non-parametric tests. When comparing more than two sets of numerical data, a multiple group comparison test such as one-way analysis of variance (ANOVA) or Kruskal-Wallis test should be used first.
From the description, I understand that you are administering a survey to your sample group. Are you running this survey on multiple groups with/without the online learning component or a pre/post online learning sample? If so, Nirmal's suggestion to use an ANOVA test would help you determine difference.
On the other hand, if you are looking to have a single sample group self-identify their perceived change in attention or engagement without a comparison, we'd need more information on your methodology to recommend an appropriate statistical method. Is there any additional assessment going on to support the self-perception reported in the survey (course performance, number of campus interactions, use of supplemental instruction, co-curricular participation, etc.)?
It sounds like you have two dependent variables, engagement and "focus" which are both interval-level variables, and you want to compare three groups on these scores. If so, then ANOVA is the most appropriate method.
Understanding out of the description of your inquiry, you have one independent variable and two dependent variables (DVs), and assuming your data meets parametric assumptions, you can go for one-way multivariate analysis of variance (one-way MANOVA), which is employed to examine whether differences exist between independent groups on two or more continuous DVs. The following could provide guidance on how to do MANOVA on SPSS.
Lund Research Ltd. (2018). One-way MANOVA in SPSS statistics. SPSS statistics tutorials and statistical guides | Laerd Statistics. https://statistics.laerd.com/spss-tutorials/one-way-manova-using-spss-statistics.php
Good luck,
Good morning
Maybe descriptive and study case.
Also, deep interview
Best regards
Ph.D. Ingrid del Valle García Carreño
The experimental method was used, but the appropriate statistical method is due to the sample distribution, is it normal or not, or if it is a moderate distribution, use the t-test with the brother / taking into account the sample size is greater than 30
I am in agreement with Heraiz regarding the use of the t-test as part of the statistical methods to analyze the data. Not sure if the questionnaire was designed to collect quantitative or qualitative data, that will determine how you analyze the data.
Hi Preet
Actually, your topic seems like an experimental research, but it has not been clearly described, so it needs more detailed information about your research.
However,
If it is an experimental research, and if you are taking covariates then ANCOVA will be more appropriate statistical test, but the assumptions regarding ANCOVA must be tested before implementation. If you are not taking covariates, then One way ANOVA and Post Hoc tests will be appropriate to compare the group differences and test your hypotheses.
From this we are not able to respond properly. Many relevant things are missing. Please add more details about your research work. Provide information about your research questions, objectives of the study and nature of the study.
Are you familiar with the expression, "getting the cart before the horse"? I'm afraid that's what you're facing. If so, that's not an impossible situation, but it can be a daunting one. Deciding what statistical analysis you will use is normally part of the research design, so you are sure that the data you gather can be analyzed in a way that will answer your research questions. that is normally done before you begin any data collection. I agree with what MOST of the others have recommended, but you cannot apply ANOVA, for example, unless you have quantitative data, and you did not describe the data you received from the surveys. If you solicited qualitative answers to the survey questions, you cannot apply quantitative statistical analysis to it. You will have to quantify that data or apply qualitative data analysis methods to it and have a blended study. There is nothing wrong with having a blended study, but is is more complex than a qualitative or quantitative one. You may need someone on site who can actually see what you did to help you sort it out. I wish I were there! I'd love the challenge!
good morning preet,i need more detail about your project,if you need to use experimenal designe you have to describe the independent and depended variables.however i think that one way ANOVA may be better way to analyzes your data
The experimental method was used, but the appropriate statistical method is due to the sample distribution
- Badges
- Science topic