# Statistical Analysis help for thesis - Correlation, Probit, Tobit and Moderation

#### Uni_student

##### New Member
Hello Talkstats users!

I am writing here cause I need some guidance on my statistical analysis which has turned out far too complex for my basically begineer stats skills and my self research.

Before I ask my specific, I will give some background to my work. The scope of my analysis is to produce three models answering these questions:

Scope 1. Which of my 11 IVs predict if a firm licenses (dummy variable DV coded 0/1)

Scope 2. Which of my 13 IVs predict how much a firm licenses (DV is continuous scale of 0-100% of sales)

Scope 3. Is the relationship identified in scope 2 moderated by where they license to (two dummy variables for USA and China with each coded 0 and 1). Basically I am interested in seeing if there is a difference in IVs between the firms that license to these nations.

I have approached this problem, based on past research's recommendations, by utilising a probit analysis for scope 1 and a tobit analysis for scope 2 and 3. I have used robust models of each scope and have conducted outlier removal analysis based on Studentised rs and cook's SD.

Whilst running these analyses in Stata, I have however been faced with the below problems I cannot seem to figure out.

1. Am I meant to run an initial correlation matrix for each model I create or an initial one with all the IVs and DVs together? I would lean towards making an individual one for each.

2. Is it alright if one IV has a correlation of .8 with scope 2's DV and then has a tobit dy/dx of 105 with significance at the 0.000 level (much higher than the second highest one)? This seems illogical as from my understanding and should prompt me to drop this variable from the analysis. If I do drop it, many other variables gain significance or have their dy/dx completely change.

3. What is the best tactic to use my moderation variables in such a complex analysis compromised of so many IVs and with two dummy moderators?

I am at everyone's disposal to provide more information about my work.

Thanks for your time and help.