Hello everyone,
I hope some of you can help me out with a confusion.
For my PhD project. I am examining regional variability of the effect of immigrations status on positive mental heallth in Canada. Here the dependent variable, positive mental health, is a scale varaible constructed based on 14 items. The two other main variables are province of residence, GEO_PR which is collapsed into five regions (Atlantic Canada=AC, Quebec=QC, Ontario=ON, Praisirs, and Britisch Colombia=BC), and immigration status, immigrant, which contains three categories-- Canadian born, Recent Immigrant, and Long residing immigrant.
While I am ok with the regression analysis where I will use interaction term between province and immigration status variables to see the regional whether the effect of immigration status on prositive mental health differs across different difference regions. I am planningto run OLS regression. However, I am a bit confused the bivariate level results. If I collased the scale of positive mental health variaiable in to "High"= 1std above the mean, "Low"= 1 std belwo the mean, and "Medium"= inbetween and run trivariate frequency tables with province and immigration status, I get some cell size problem at doesn't allow me to get the result vetting out from the RDC. So, I am planning the following analysis:
As my dependent variable is scale, I am thinking to run mean value for the positive mental health varible over province and immigration status variables and then run a marginsplot. In this case, insted of reporting table, I am thinking to report the marginsplots. It clearly shows the mean value for each of the immigration status variable differs across different provinces; some province shows Canadian born group with higher mean, which other province shows recent immigratns have higher mean of positive mental health.
I would really appreciate it if you kindly help me with your valueworthy suggections whether my plan is ok or I have to follow a different parth. Looking forward to gracefully read your suggestions.
Thank you in advacne,
Iqbal Chowdhury
I hope some of you can help me out with a confusion.
For my PhD project. I am examining regional variability of the effect of immigrations status on positive mental heallth in Canada. Here the dependent variable, positive mental health, is a scale varaible constructed based on 14 items. The two other main variables are province of residence, GEO_PR which is collapsed into five regions (Atlantic Canada=AC, Quebec=QC, Ontario=ON, Praisirs, and Britisch Colombia=BC), and immigration status, immigrant, which contains three categories-- Canadian born, Recent Immigrant, and Long residing immigrant.
While I am ok with the regression analysis where I will use interaction term between province and immigration status variables to see the regional whether the effect of immigration status on prositive mental health differs across different difference regions. I am planningto run OLS regression. However, I am a bit confused the bivariate level results. If I collased the scale of positive mental health variaiable in to "High"= 1std above the mean, "Low"= 1 std belwo the mean, and "Medium"= inbetween and run trivariate frequency tables with province and immigration status, I get some cell size problem at doesn't allow me to get the result vetting out from the RDC. So, I am planning the following analysis:
As my dependent variable is scale, I am thinking to run mean value for the positive mental health varible over province and immigration status variables and then run a marginsplot. In this case, insted of reporting table, I am thinking to report the marginsplots. It clearly shows the mean value for each of the immigration status variable differs across different provinces; some province shows Canadian born group with higher mean, which other province shows recent immigratns have higher mean of positive mental health.
I would really appreciate it if you kindly help me with your valueworthy suggections whether my plan is ok or I have to follow a different parth. Looking forward to gracefully read your suggestions.
Thank you in advacne,
Iqbal Chowdhury
