st: RE: Understanding the mechanics of the sum() function

[Nick Cox <n.j.cox@durham.ac.uk>]

`a' and `b' I take to be quite secondary here. There is a difference as -sum(`exp')- is affected by what is in the current observation if `exp' is so affected, and will return 0 in circumstances in which `exp' + what's inherited will return missing. 

For example set 

set obs 10
gen z = . 
local a 0 
local b 1 
local exp z 
generate y1 = `a' + `b' * sum(`exp')
gen y2 = .
replace y2 = `a' + `b' * (`exp' + cond(missing(y2[_n-1]), 0, y2[_n-1]))
l 

I doubt you could make your code faster as your code requires more interpretation. 

Nick 
n.j.cox@durham.ac.uk 

Billy Schwartz

Is there a difference between

. generate y = `a' + `b' * sum(`exp')

and

. generate y = .
. replace y = `a' + `b' * (`exp' + cond(missing(y[_n-1]), 0, y[_n-1]))

where `a', `b', and `exp' are arbitrary Stata expressions? If they're
not the same, is there a way of modifying the second to make it the
same as the first? (If they're the same in effect, is one faster than
the other?)
*

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/