destring when numerical values with < or > signs

[celine steib]

Hello,

I am working on epidemiological data and I have a couple of variable from biomarkers for which I have values such as > 1.50 etc.
I cannot use the destring command as I have done with others because it says I have characters in my data.
How can I solve this issue?

Thank you in advance.

[Nick Cox]

Clearly -- meaning clearly, once you read the help -- you can just specify

destring whatever, ignore(><) 

and >1.50 will just get recorded as 1.50 and <666 as 666, and so on and so forth. But that just solves one problem to create another. How are going to treat these data? What justifies throwing away the inequalities?

In some situations, creating flags such as

gen gt = strpos(whatever, ">") > 0 

gen lt = strpos(whatever, "<") > 0

may allow you to carry data forward to special-flavour regressions.