[R] summary.manova rank deficiency error + data

[Pedro Mardones]

Dear R-users;

Previously I posted a question about the problem of rank deficiency in
summary.manova. As somebody suggested, I'm attaching a small part of
the data set.

#***************************************************

"test" <-

structure(.Data = list(structure(.Data = c(rep(1,3),rep(2,18),rep(3,10)),
levels = c("1", "2", "3"),
class = "factor")

,c(0.181829,0.090159,0.115824,0.112804,0.134650,0.249136,0.163144,0.122012,0.157554,0.126283,
0.105344,0.125125,0.126232,0.084317,0.092836,0.108546,0.159165,0.121620,0.142326,0.122770,
0.117480,0.153762,0.156551,0.185058,0.161651,0.182331,0.139531,0.188101,0.103196,0.116877,0.113733)

,c(0.181445,0.090254,0.115840,0.112863,0.134610,0.249003,0.163116,0.122135,0.157206,0.126129,
0.105302,0.124917,0.126243,0.084455,0.092818,0.108458,0.158769,0.121244,0.141981,0.122595,
0.117556,0.153507,0.156308,0.184644,0.161421,0.181999,0.139376,0.187708,0.103126,0.116615,0.113746)

,c(0.181058,0.090426,0.115926,0.113022,0.134632,0.248845,0.163140,0.122331,0.156871,0.126023,
0.105335,0.124757,0.126325,0.084690,0.092885,0.108455,0.158386,0.120913,0.141676,0.122492,
0.117707,0.153293,0.156095,0.184242,0.161214,0.181670,0.139271,0.187318,0.103129,0.116421,0.113826)

,c(0.180692,0.090704,0.116110,0.113319,0.134745,0.248678,0.163256,0.122637,0.156581,0.125998,
0.105479,0.124686,0.126514,0.085066,0.093088,0.108587,0.158040,0.120674,0.141446,0.122488,
0.117972,0.153150,0.155954,0.183885,0.161063,0.181383,0.139251,0.186956,0.103232,0.116351,0.114001)

,c(0.180353,0.091088,0.116392,0.113753,0.134965,0.248520,0.163475,0.123046,0.156354,0.126067,
0.105726,0.124713,0.126821,0.085584,0.093432,0.108858,0.157742,0.120533,0.141309,0.122595,
0.118340,0.153088,0.155897,0.183582,0.160975,0.181143,0.139314,0.186636,0.103449,0.116415,0.114275)
)
,names = c("GROUP", "Y1", "Y2", "Y3", "Y4","Y5")
,row.names = seq(1:31)
,class = "data.frame"
)

summary(manova(cbind(Y1,Y2,Y3,Y4,Y5)~GROUP, test), test = "Wilks")

#Error in summary.manova(manova(cbind(Y1, Y2, Y3, Y4, Y5) ~ GROUP, test),  :
  residuals have rank 3 < 5

#***************************************************

What I don't understand is why SAS returns no errors using PROC GLM
for the same data set. Is because PROC GLM doesn't take into account
problems of rank deficiency? So, should I trust manova instead of PROC
GLM output? I know it can be a touchy question but I would like to
receive some insights.
Thanks
PM