Thank you very much Hector!
I knew there was a solution, although I did not remember it, Thanks again!
best
Staffan
-----Ursprungligt meddelande-----
Från: Hector Maletta [mailto:
[hidden email]]
Skickat: den 9 mars 2007 19:31
Till: 'Staffan Lindberg';
[hidden email]
Ämne: RE: Output of zero counts in frequencies and graphs
The trick you vaguely remember does not change the population size.
It works by adding a new case to the file, but also giving that case a very
small WEIGHT. If the analysis is not using any weight yet, a new weighting
variable should be introduced with a value of 1 for all the other cases and
a very small value (such as 0.0001) for that additional case. That case,
multiplied by its weight, would be rounded to zero weighted cases, but will
appear in the tables and charts. The size of the (weighted) population will
remain unchanged.
Hector
-----Mensaje original-----
De: SPSSX(r) Discussion [mailto:
[hidden email]] En nombre de
Staffan Lindberg Enviado el: 09 March 2007 16:00
Para:
[hidden email]
Asunto: Output of zero counts in frequencies and graphs
Dear list!
I know this has been dealt with in the list before but I cannot
rember the
solutions.
A collegue of mine has a rather small material and would like to
have even
the zero counts in the frequency tables (i.e if the number 5
"disagree" has
zero counts she wants to have that shown in the frequency table).
Likewise
in a graph she wants to show 5 "disagree" on the x-axis with zero
counts on
the y-axis.
I remember reading something about adding a record with an
infinitesimal
value (i.e. 0.00001) but she does not want to change the population
size. Is
there another way to accomplish this?
best
Staffan Lindberg
National Institute of Public Health
Sweden
Locked
