Re: Recode

[Ben Pfaff]

On Wed, Jan 09, 2013 at 10:37:55AM -0800, pohaku wrote:
> 
> 
> 
> Ben Pfaff wrote:
> > 
> > On Wed, Jan 09, 2013 at 09:22:12AM -0800, pohaku wrote:
> >> 
> >> In recode you use (num1 thru num2 = ?). But, suppose you are recoding
> >> continuous data, then it is likely that num2 of one code is num1 of next.
> >> Example recoding student GPA
> >> 
> >> (2.222 thru 2.555 = C)
> >> (2.555 thru 3.000 = B)
> >> 
> >> the question is -- is the logic of the recode 
> >> 
> >> (2.222<= X <2.555 =C)
> >> (2.555<=X  <3.000 = B)
> > 
> > It's 2.222 <= x <= 2.555 becomes C, 2.555 < x <= 3.000 becomes B.
> > 
> > to be clear... the firs recode is inclusive on upper and lower end of
> > range and successive recodes are exclusive and inclusive

All of them include both ends of the range, but they are checked in the
order they are written and only the first match is executed.

> > (2.222<= X <=2.555 =C)
> > (2.555 < X  =< 3.000 = B)
> > (3.000 < X  =< 4.000 = A)
> > (4.000 < X  =< 4.200 = A+)
> > (4.200 < X <= highest =A++)
> > 
> > is this the logic above? for real/continuous data the boundaries are
> > important. all the documentation is more or less categorical or discrete
> > in nature thus this boundary issue doesn't come up. just that i am using
> > PSPP for first time in intro stats course and chapter 2 is qualitative
> > descriptive stats and recoding of continuous to categorical is rather
> > essential to create a meaningful frequency distribution table with a
> > reasonable number of rows.

Well, your example only included the first two, but yes it would be easy
to rewrite the informal notation above into a real RECODE command with
those semantics.