1606071216 Destring-ing data with entry errors or coding problems

The "ds" command already mentioned is a powerful tool to clean up and analyse rebellious data. Say you know that your data is numerical but for some reason has been stored as string, you can:

• spend time identifying by hand the string variables and destring them all at once or one by one
• execute this:

ds _all, has(type string)

local tolist

foreach var of local tolist {

di "var'" `

count if real(var')==.`

if r(N)!=0 listvar' if real(var')==.

}

This allows you to ex-ante check whether your supposed numerical variables take non-numerical values, and props them in a list. You can then work on them, and ultimately destring the variables needed.