Generate dummy based on part of the value of a string variable

[Ilse de Groof]

Dear reader,

I am dealing with a database containing executives and their title. Two string variables are attached: title and titleann, some examples of the string values of the variables:

title

group vp-structures & systems, maintenance, repair and overhaul

chairman, president & CEO

vp, general counsel & secretary

vp & chief finance officer

group vp-aviation supply chain

group vp-structures & systems, maintenance, repair and overhaul

president & CEO

president & chief operating officer

executive vp & chief finance officer

senior vp; president-Integrated Solutions Group

chairman, president & CEO

vp-worldwide operations

titleann

group vp-structures & systems, maintenance, repair and overhaul

Chairman, Chief Executive Officer and Chairman of Executive Committee

vp, general counsel & secretary

President, Chief Operating Officer of Expeditionary Services and Director

Vice President of Commercial Development

group vp-structures & systems, maintenance, repair and overhaul

president & CEO

president & chief operating officer

Chairman, Chief Executive Officer and President

senior vp; president-Integrated Solutions Group

chairman, president & CEO

vp-worldwide operations

I want to generate a dummy variable being equal to 1 or "CFO" when the text includes "chief financial officer"/"principal financial officer"/"chief finance officer". Under variable title, "executive vp & chief finance officer" is one observation. The term I am searching for may be part of a string variable, rather than the only value in that specific cell.

Does anyone know how to generate such a dummy from a string variable containing more than only the item I am searching for?


Kind regards,

Ilse

[Jesse Wursten]

See help strpos. E.g.

gen isCFO = 1 if strpos(title, "chief financial officer") > 0 | strpos(...) ...

[Clyde Schechter]

Jesse Wursten's advice will do what Ilse de Groof asked for, quite literally. But it may not be what she really wants. The code in #2 generates a variable coded 1 for true and missing value for false. That kind of variable is difficult to work with in Stata since, when used in any boolean expression, Stata treats missing value as true. Stata is really set up to use boolean variables coded 0 for false and 1 for true. To do that, the code would be:

 gen isCFO = strpos(title, "chief financial officer") > 0 | strpos(...) ...

Sometimes we actually want a trichotomy: 0/1/missing value where missing value denotes "can't be evaluated as true or false," for example, in an observation where the variable title is itself missing. So that would look like:

 gen isCFO = strpos(title, "chief financial officer") > 0 | strpos(...) ...    if !missing(title)