You guys come up with some amazing stuff! Of course, however, I'm not exactly able to more than just acknowledge your input at some point... I feel I'm lacking some vital basics. Be that as it may, I thought I'd let you know what's happened to your input on my side:
glm(recode(PLANMOV, "'yes'='no'; 'no'='yes'")~ log(8-ATTMT_AVG), family=binomial(link="probit"), data=dta), with PLANMOV as a dichotomous variable on whether at the time a move is planned, and ATTMT_AVG as the average of some 9 Likert-Scale coded items, where 1 is high agreement and 7 is high disagreement. Since the underlying theoretical rationale is that greater attachment to the home will per definitionem mean a smaller tendency to move, combining the recode and inversion-by-subtraction approaches means to represent functionally the structure of the theoretical argument. So while of course the coefficient is the same, I think it is an achievement to be able to express a thought statistically in the same way I do mentally, and not just by implication (i.e. when I find that low attachment means greater willingness to move as is the direct reading of the untransformed function). Cheers, Alexis --- [EMAIL PROTECTED] wrote: > On 04-Aug-07 22:02:33, William Revelle wrote: > > Alexis and John, > > > > To reverse a Likert like item, subtract the item > from the maximum > > acceptable value + the minimum acceptable value, > > That is, if > > x <- 1:8 > > xreverse <- 9-x > > > > Bill > > A few of us have suggested this, but Alexis's > welcome for the > recode() suggestion indicates that by the time he > gets round to > this his Likert scale values have already become > levels of a factor. > > Levels "1", "2", ... of a factor may look like > integers, but they're > not; and R will not let you do arithmetic on them: > > > x<-factor(c(1,1,1,2,2,2)) > > x > [1] 1 1 1 2 2 2 > Levels: 1 2 > > y<-(3-x) > Warning message: > "-" not meaningful for factors in: Ops.factor(3, x) > > y > [1] NA NA NA NA NA NA > > However, you can turn them back into integers, > reverse, and then > turn the results back into a factor: > > > y <- factor(3 - as.integer(x)) > > y > [1] 2 2 2 1 1 1 > Levels: 1 2 > > So, even for factors, the insight undelying our > suggestion of "-" > is still valid! :) > > Best wishes, > Ted. > > -------------------------------------------------------------------- > E-Mail: (Ted Harding) <[EMAIL PROTECTED]> > Fax-to-email: +44 (0)870 094 0861 > Date: 05-Aug-07 > Time: 00:09:58 > ------------------------------ XFMail > ------------------------------ > ____________________________________________________________________________________ ______________________________________________ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.