Hello all,
I am unsure of how to interpret the output from a Generalized Estimating
Equation analysis of an ordinal response. I hope someone can enlighten
me. The analysis was done using package 'repolr'. The data consists of
a Score on a 3-point scale from 56 Subjects after repeatedly washing
their hands with soap. Two soap Products were tested, each panelist
washed 10 times = 10 Applications.
I'm puzzled by some aspects of the model output below from repolr---
* correlation structure changed from AR1 to Fixed
* additional factors = "cuts" 1 and 2 added to model ( is this
testing
for the ability of GLM to detect cutpoints for the assumed latent
effect?)
* assume that "factor(Product)[T.2]" refers to the 2nd level of the
Product factor (coded "869"), for comparison to baseline of 1st level
(coded "143")
Any insight is much appreciated. Thanks, Paul
================ Model ========================
This is the model that was fit, using the "repolr" package:
soapfeel.mod <- repolr(formula = Score ~ factor(Product) * Application ,
subjects = "Subject" , data = soap.data , categories = 3 , times
c(1,2,3,4,5,6,7,8,9,10) , corstr = "ar1", tol = 0.001, scalevalue = 1,
alpha = 0.5,po.test=TRUE, fixed=FALSE)
================ Output========================> summary(soapfeel.mod[["gee"]])
GEE: GENERALIZED LINEAR MODELS FOR DEPENDENT DATA
gee S-function, version 4.13 modified 98/01/27 (1998)
Model:
Link: Logit
Variance to Mean Relation: Binomial
Correlation Structure: Fixed
Call:
ogee(formula = formula, id = exdata$exdata$subjects, data exdata$exdata,
R = R_mat, b = as.numeric(coeffs), maxiter = 10, family
"binomial",
corstr = "fixed", silent = TRUE, scale.fix = TRUE, scale.value
scalevalue)
Summary of Residuals:
Min 1Q Median 3Q Max
-0.32791175 -0.13163165 -0.05841431 -0.02337869 0.97076089
Coefficients:
Estimate Naive S.E. Naive z Robust
S.E.
factor(cuts)1 -3.0093220 0.47829379 -6.291786
0.51860442
factor(cuts)2 -1.3082469 0.41257539 -3.170928
0.40572065
factor(Product)[T.2] -0.6136790 0.58810180 -1.043491
0.69995927
Application -0.1445904 0.07672638 -1.884494
0.09077013
factor(Product)[T.2]:Application 0.2650185 0.09867353 2.685811
0.10336124
Robust z
factor(cuts)1 -5.8027310
factor(cuts)2 -3.2245017
factor(Product)[T.2] -0.8767352
Application -1.5929293
factor(Product)[T.2]:Application 2.5640026
Estimated Scale Parameter: 1
Number of Iterations: 1
Working Correlation
[,1] [,2] [,3] [,4] [,5]
[1,] 1.000000e+00 4.271812e-01 2.375569e-01 1.014798e-01 5.643328e-02
...
...
====================Data frame===================> str(soap.data)
'data.frame': 560 obs. of 6 variables:
$ Subject : int 1 1 1 1 1 1 1 1 1 1 ...
$ Product : int 869 869 869 869 869 869 869 869 869 869 ...
$ Question : int 2 2 2 2 2 2 2 2 2 2 ...
$ Application: int 1 2 3 4 5 6 7 8 9 10 ...
$ Score : int 3 3 3 3 3 3 3 3 3 3 ...
=====================sessionInfo()=================R version 2.8.1 (2008-12-22)
i386-pc-mingw32
locale:
LC_COLLATE=English_United States.1252;LC_CTYPE=English_United
States.1252;LC_MONETARY=English_United
States.1252;LC_NUMERIC=C;LC_TIME=English_United States.1252
attached base packages:
[1] tcltk grid stats graphics grDevices datasets utils
[8] methods base
other attached packages:
[1] Rcmdr_1.4-7 car_1.2-12 repolr_1.0 hints_1.0.1-1
[5] gee_4.13-13 effects_2.0-3 nnet_7.2-45 MASS_7.2-45
[9] lattice_0.17-20 geepack_1.0-16
loaded via a namespace (and not attached):
[1] boot_1.2-35 lme4_0.999375-28 Matrix_0.999375-20 tools_2.8.0
[5] Zelig_3.4-1
Paul Prew
Statistician
Ecolab
ESC F44, 655 Lone Oak Drive
Eagan, MN 55123
CONFIDENTIALITY NOTICE: \ This e-mail communication an...{{dropped:11}}
