Displaying 1 result from an estimated 1 matches for "f380".
Did you mean:
380
2010 Dec 30
1
Different results in glm() probit model using vector vs. two-column matrix response
...39;s for both
SF
y
ldose
log10x
# analysis with 10 records and 80 records
f1 <- glm(SF ~ group + ldose, family=binomial(link="probit"))
f3 <- glm(SF ~ ldose, family=binomial(link="probit"))
f180 <- glm(y ~ trt + log10x, family=binomial(link="probit"))
f380 <- glm(y ~ log10x, family=binomial(link="probit"))
summary(f1)
summary(f180)
f1$weights
f180$weights
# check weights divided by 8 to see if match -- match several decimal places,
# but not exactly
f1$weights/8
****
Shelly Lensing
Biostatistics / University of Arkansas for Me...