[R] Scripting Problem--Linear Regression Using Logit of the Response Variable

John, Larry Larry.John at anser.org
Thu Jan 7 21:22:22 CET 2016


Dear fellow citizens of the R-verse,

I'm a non-quant analyst trying to do some statistical analyses on a large data set unimaginatively named "data."

  sympathy trust fear greed sharer_prob
[1]      3     2    0     0    0.669593
[2]      2     1    2     3    0.669593
[3]      2     2    2     3    0.494675
[4]      2     2    1     2    0.494675
[5[      2     2    2     0    0.556837
[6]      2     2    1     1    0.556837

"sharer_prob" is the continuous dependent variable with values 0 > sharer_prob > 1, so I know need to run a logit transformation before I can do any linear regressions to study the main and interaction effects of the four independent variables.

I've loaded "car" and tried for several hours to create a one-line script that will give me the overall linear model, but keep having problems. It seems to me that something like the following should work


> EvacLM <- lm((logit(sharer_prob) ~ sympathy + trust + fear + greed) adjust=TRUE) na.action=NULL)

But R politely tells me I've gotten something wrong:


Error: unexpected symbol in "EvacLM <- lm((logit(sharer_prob) ~ sympathy + trust + fear + greed) adjust"

I'm hoping one of you can spot and help me correct my mistake-I sure can't figure it out.

Thanks for any help you can offer!

Larry John
Principal Analyst
ANSER (www.anser.org)

	[[alternative HTML version deleted]]



More information about the R-help mailing list