Nope, p value selection doesn't work. Stay away from things like Stepwise regression or anything based on it. Further if you have 3 DVs then you definitely want to consider using a multivariate method. The properties of the DVs tell you which type of regression you need. Stay with regression because it includes anything you might want to do and is easier to work with if your DV info is not experimental, eg survey. Lots of information to look at here. I'm attaching some info to get you Best wishes, David Booth

Another good method reference

And one more reference. Download of R software is free or you can run online from a browser. Google to see how. This software is my recommendation because your analysis will be a bit complicated. Best David Booth

David Eugene Booth Hi, thank you so much for responding!

I have heard of multivariate regression before but have not been taught this in class and can not find anything about it in my textbooks, which may be because we are being taught about how to apply statistical methods only using SPSS software.

So I think I'd need a method compatible with SPSS and hierarchal regression was the method I thought would work best, as we have only been taught about that and logistic regression. My DV data was gained using survey Likert scales too.

Thank you for your help though!

My friend you don't use what you have been taught, you use what gives you the answer to your research question. I am retired but look at the dates on my cites. I never heard of any of them in grad school. That is research. D. Booth

@laura I looked at your submitted research plan. It is good in detail and lays out the plan nicely but is missing the two things that are most important: the research question and the stat methodclearly outlined to answer the research question. You have a design. 1 what kind of design is it and the Statistical methods you propose to solve the bp4

Here's my suggestions:1. Use a standard design whenever possible. The reason is twofold: first you know what you will get from the design hopefully what your research question asked for and 2 you know the Statistical model to use. If you use a nonstandard design you may have much statistical thinking to develop the model.

Here's an excellent reference on design and Analysis. To choose the correct mod google what regression model do I chose. If your DVs are correlated you have more than one DV in your model so google the multivariate form of the regression model. As I said I suggest you use R. The. Best R book I think is attached above. No cost for R. If you have further follow-up questions do not fail to ask. This is what is required to do real research. It can be difficult but it is very rewarding. I wish you good luck and s good and long life researching your desired field. Always pursue the answer to valuable and important questions when possible. May the Force be with you. David Booth

Dear Laura McKirdy ,

One quick question first – you mentioned you collected data for your DVs using a Likert scale question, please clarify if you are treating the measurement level of your DV(s) data as continuous? Since you are using SPSS and running the regressions for each dependent variable individually by entering the sets of predictors into the model in blocks or stages, YES hierarchical regression analysis is very feasible.

Dear Laura McKirdy ,

You may as well please watch the video with the link herein below, I hope it will help too.

All the best!

https://www.youtube.com/watch?v=RyDteu6E7HY