Stata Tutorial: Nonlinear Transformations

When, Why and How to use simple log and square transformations in an OLS regression. We run through an example using scatter plots, histograms, the coefficient of skew, and interpreting OLS results in Stata.
Link to "Gentle Introduction to Stata"
www.amazon.com/gp/product/159...
Link to the excellent Introduction to Econometrics Textbook by AH Studenmund:
www.amazon.com/gp/product/933...
Link to Jeffrey Wooldridge Introductory Econometrics Textbook:
www.amazon.com/gp/product/813...
My Twitter is:
/ michaelrjonas
My Google Scholar Page:
scholar.google.com/citations?...
ResearchGate:
www.researchgate.net/profile/...

[@GomadaTube]

you are my best teacher

[@zhongzhihong3435]

Very helpful video. thank you!

[@mikejonaseconometrics1886]

Thanks!

[@adityavedam1174]

Suppose we have multiple indep variables in the equation would do suggest checking non linearity pairwise (i.e. among y and each indep var or iv1; then y and iv2 and so on or else
check non linearity by checking the skew value (positive or negative) variable wise and then transform them appropriately to reduce their skew and run the regression using the transformed variables....
Please clarify? Thank you in Advance Mike

[@petertraulsen600]

I want to make a regressions where I test log(immigration) between all U.S. states depending on the stringency of the covid-19 restrictions. (I use an index created by Oxford).
My theory is that people moved toward the "extreme" states, so the states with highest and lowest stringency got the most immigration during that period. Because of that I want to include a quadratic term in my regression so that the effect of stringency depends on the value of stringency (high/low stringency should mean higher effect of stringency on immigration, while moderate stringency should mean less effect).
The regressions model is: reg logimmigration stringency stringency2, but I'm not sure that is how it should be done.
Thank you in advance

[@MAli-jw2eq]

Thank you. Can we use the squared terms in xtreg command for panel data?

[@jyotsnarosario2204]

If education is taken in log, should we take (log education# log education) or first square the education term and then take log.

[@iHazardGaming007]

Could someone explain what a fitted value even means? And why would you not just plot lwage and education?

[@georgeadu4055]

Hello Mike. Please I need help . Am estimating a simple non linear model with stock of capital as the dependent variable, ( gdppc, bank captial to adequacy ration, unemployment), exp(public sector corruption) and regional dummies. I keep getting errors when I issue the command. Could you please offer some adivice?
Thank you

[@mikejonaseconometrics1886]

George, email me at [e-mail] with a screenshot of the command and error message and I can take a look.

[@georgeadu4055]

Thanks

[@xarawormald4189]

which is the dependent variable and which is the independent variable?

[@ahmadouly4878]

The first variable is always the dependent variable, and the following variables are the independent variables

[@madsfristednavne9632]

Urgent! How do I find the function of the graph of the Linear regression in Stata?

[@mikejonaseconometrics1886]

You mean you want to graph the fitted values following a regression?
You can use:
reg y x
predict y-hat
scatter y-hat x
if you mean that you have the graph, and want to find the function that made it, that is a different problem.

[@madsfristednavne9632]

@@mikejonaseconometrics1886 Thank you for the quick response! Yes, I want to find the function that made it.