How many hypotheses should one have ? How to decided count of hypotheses where there are four independent , one mediator and one dependent variable?
Should I have 7, 10 ,13 or another number? Can you please list the way, how you will make the hypothesis?
It depends upon your objectives.I believe you want to test the mediating effect of D on all the IVs-DV linkages, if so then 10 hypotheses.
Noor Ul Hadi Muhammad Farrukh , Thank you for your reply. I absolutely agree that it depends on objective, but I was wondering What if I select only 7?
1-A--> E
2-B-->E
3-C-->E
4-A-->D-->E
5-B-->D-->E
6-C-->D-->E
7--D-->E
Because, based on 4th, 5th and 6th hypothesis, I have already assumed that the independent variable has a relationship with the mediator. So I can avoid below possible hypothesis?
8-A-->D
9- B-->D
10- C-->D
You can, but if you want to test the model simultaneously with mediating effects, you have to proceed with 10.
For 7 hypotheses, you have to change the direction of your hypothesized relationships.
regards.
Which approach of mediation are you planning to adopt?
Muhammad Farooq
I will suggest to go for PLS-SEM, or you may go for AMOS Graphics.
You can find my paper on mediation Analysis which might help you.
https://www.researchgate.net/publication/299537761_Making_Sense_of_Mediating_Analysis_A_Marketing_Perspective
Muhammad Farrukh Sorry, I dont the approaches. Is it something related to full mediation or partial mediation ?
Noor Ul Hadi Thank you so much Sir for sharing the article. I just downloaded, read the abstract. I found it very-useful. Will read and cite the article in work. Thank you for sharing.
Muhammad Farrukh I just read the approaches from Noor Ul Hadi shared article. I hope you are talking about these three approaches.
Baron and Kenny’s (1986) mediation analysis
The Sobel test (1982)
The bootstrap method (Preacher and Hayes, 2004; 2008)
Muhammad Farrukh I am learning this mediation analysis. Please share with me approaches or anything else which can help me in analysis or I need to learn. Thank you.
You can have A -> D -> E but you can also have D modifying the A -> E reaction. That gets to ten. Then you can have A, B, and C interact with each other, for another three interactions for a total of 13. If D or E influence these interactions then one could get 19 interactions. D could modify the (A, B, C) -> D interactions for a total of 22. If there are feedback loops like the level of E changes the A -> E transition then there are six more for 28, plus one more for the D -> E feedback.
My total is 29. With luck, substantive theory can be used to remove a few of these to make the experimental design more manageable.
A -> D -> E has two hypotheses, but A -> D , B -> D, and C -> D are three and D -> E is one unless the quality of D changes depending on whether D comes from A, B, or C and then things can get more complicated. If A has three components (quantity, quality, and duration) Then the number of potential hypotheses just increased by a great deal.
Timothy A Ebert Thank you so much for your valuable response. Thank you for sharing the reference/technique ( Substantive theory). I was confused because of all these options. Now using Substantive theory, I will justify my choice.
Timothy A Ebert The next thing, I am thinking, I hope you can guide me better. Let us say, with substantive theory , I designed 7 or 10 hypothesis. Later on after analysis I realized that there are 13 relationships or more. How can i justify this thing? What researchers normally do?
Substantive theory depends on the field of study. I could suggest a system using plants and insects, but that would be silly if the problem is in psychology or another field.
For insects: an experiment that allows you to test all 29 (or more) relationships will be too large. Laboratory option: You focus by controlling some or all the relationships, and the result may or may not make a difference in the field. Field option: You accept that you cannot control everything and there may be alternative explanations to your results. With a bit of luck and hard work you can arrive at a result that others can duplicate, thereby showing that your initial observation was correct or that others can make the same mistake.
In part the solution is to understand the literature and project goals. This may provide context justifying reducing the problem.
A hypothesis follows when one has a comparative or relational objective stated in a study. The number of relational or comparative objectives you have will determine also similar number of hypotheses in a study.
Hypothesis dependent upon research questions and research objectives.According to substantive theory elemets can be transferable to context of action under the study.
Another option is to redo the experiment. The first run was a pilot study to get preliminary data. Depending on the topic it might also work as justification for further funding.
Use logic to eliminate some of the options. If A, B, and C are elements, say nitrogen, potassium, and phosphorus then it is hard to envision how the number of atoms of nitrogen would directly influence the number of atoms of phosphorus. Thus the A -> B hypothesis doesn't make sense. If A influences the availability of B to E then this is A influences the B -> E hypothesis.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics