One of my friend like to code 1 = Yes and 2 = No for a dummy variable instead of 1 or 0 respectively. Anyone have idea how SPSS reads that coding (1 & 2): As on/Off or as 1 and 2?
It depends on the analysis, as a grouping variable in an ANOVA model the coding does not matter, in a regression model it does definitively matter and changes the outcome, here SPSS really reads it as 1 and 2! And it has to be that way in a regression model, because it is also possible to chose other coding systems (e.g. effect coding) to investigate other aspects/comparisons within the model.
Rainer is correct as far as he goes. While he is correct the regression output looks different between the two coding schemes a proper interpretation of the meaning of the regression coefficients between the two schemes will sort out the differences and eventually give the same results. Be careful though. Best, David
I agree with Rainer. Whether (and how) it matters depends on which procedure(s) your friend is using. For many things, 1=Y, 0=N coding is more convenient. But if one is using CROSSTABS to compute an odds ratio (as just one example), it is probably more convenient to use 1=Y, 2=N coding.
HTH.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS