I am doing my master degree thesis on Earnings Management (AEM and REM). Right now I am working on Stata to shape my data and noticed the complexities of Stata handles missing data points (NA) values.
Is there is any particular ways I should consider in treating NA values ?
Hi, I guess that missing values is because researchers didn't answer this question in the survey. So, if you have additional information about the researchers in your data, you can impute these missing value in Stata. Some methodologies that you can use in Stata are: Hot Deck, Multivariate Imputation by Chained, Equations (MICE), Amelia II, missForest, Hmisc, Mi, Rf2e. I think you can look in Stata blog how to use these or what commands are for each one. Additionally, be careful to replace missing value with mean or median, if you do that, you can lost variance. So, the best way is use a methodology for impute data.
Depend on your data if you can get mean or mode u can just replace it . This is only for quantative data
For qualtitive data u have to go back to your data and try to reach the participants..
In the secondary data, you need to refer back to the original excel sheet where you have assembled your variables. Sorting the data with missing values and deleting that entry would be helpful in getting the regression results.
From my experience and since that the companies were not listed in all of the time period of your research you have to erase these companies from the research. You can't use partial data per company in your time span.
- Badges
- Science topic
- Similar topics
- Thesis Research