I want to control for gender, which is a dichotomous variable (M/F), and was hoping to do this with ANCOVA. In this video, you switch from ANOVA to Regression and show 'union' as a categorical variable. I'm guessing this is because in Stata, for ANOVA you can't identify a categorical covariate as being different from the DVs and IVs?? Or is there a different solution??
With the *anova* command, variables are assumed to be categorical, so the i. notation is not needed. However, the c. notation is needed to indicate that tenure is continuous. There is a note in *help anova* that mentions how variables are treated. On the other hand, with *regress* the explanatory variables are assumed to be continuous, so if you want to include a categorical variable, you should use the i. notation, as Chuck did with i.union. Please email us at tech-support@stata.com if you have any other questions.
In this video, Chuck first created a scatterplot to examine whether there was a difference in the relationship between tenure and wage across levels of union. But it seems that the relationship is the same for union members and non-members, which is why he doesn't interact tenure and union. In Stata's factor-variable notation # refers to the interaction of the variables and ## refers to both the main effects of each variable and their interaction. (By interacting two variables you are indicating that the effect of one variable may depend on another variable.) When fitting your own models, you would base this decision on theory and examination of your data.
Thanks for teaching me in particular. Could you kindly help me how I can use covariate augmented
Dicker-Fuller in Stata. Thanks.
I want to control for gender, which is a dichotomous variable (M/F), and was hoping to do this with ANCOVA. In this video, you switch from ANOVA to Regression and show 'union' as a categorical variable. I'm guessing this is because in Stata, for ANOVA you can't identify a categorical covariate as being different from the DVs and IVs?? Or is there a different solution??
With the *anova* command, variables are assumed to be categorical, so the i. notation is not needed. However, the c. notation is needed to indicate that tenure is continuous. There is a note in *help anova* that mentions how variables are treated. On the other hand, with *regress* the explanatory variables are assumed to be continuous, so if you want to include a categorical variable, you should use the i. notation, as Chuck did with i.union. Please email us at tech-support@stata.com if you have any other questions.
Great. Thanks!
Why in some cases people says ancova should be: anova wage union##c.tenure ??????????????????
In this video, Chuck first created a scatterplot to examine whether there was a difference in the relationship between tenure and wage across levels of union. But it seems that the relationship is the same for union members and non-members, which is why he doesn't interact tenure and union. In Stata's factor-variable notation # refers to the interaction of the variables and ## refers to both the main effects of each variable and their interaction. (By interacting two variables you are indicating that the effect of one variable may depend on another variable.) When fitting your own models, you would base this decision on theory and examination of your data.
StataCorp LLC thank you so much! Stata is the best :)
great!
good one