...and I can't figure out why.:confused:

I have been trying to run a robust linear regression on a large (2000 row) dataset with categorical and continuous predictors using the robust library GUI. I can run the SPLUS tutorial with their datasets (with both factor and numeric data) just fine.

I have tried the following:
-reimporting my dataset from excel and from csv
-renaming my columns
-reordering my data
-using a smaller dataset
-using just a few variables
-running the analysis using the command line as specified in the robust library users guide
-switching computers
-altering the tutorial datasets to increase their size, changing their coding (they still run)

In my dataset, if I just select my dependent and one independent variable, I can get SPLUS to run. I get messages about there being a lot of bias, etc (i.e. messages about the stats themselves) and I also get "Problem in names(ans) <- list.names: supplied 2 names for a vector of length 1
Use traceback() to see the call stack".

If I select more than one independent variable, I either get (A)the same messages but no output, (B) I get just the message about the 2 names for one vector but no output, or (C) I get a run time error and the whole thing crashes.

Using the command line, I can't get the robust library function to run both the lm and lmRob together (fit.models) to work at all. I've tried a few other bits and pieces with varied success (test 6 with just numeric varibles seems ok; the others with categorical variables don't... and how come suddenly in test 8 it can't find my dependent variable anymore?)

> test6<-lmRobMM(RENT.AE~M.TB+n.of.tr, data=t50.res)
> test6
Initial S-estimates.

Call:
lmRobMM(formula = RENT.AE ~ M.TB + n.of.tr, data = t50.res)

Coefficients:
(Intercept) M.TB n.of.tr
2135.373 -27.33491 -54.97515

Degrees of freedom: 2043 total; 2040 residual
Residual scale estimate: 1125.368
Significant test at level 90%. The bias is high, and inference based on
final estimates is not recommended. Use initial estimates as
exploratory tools. in: print(object)
> test7<-lmRob(RENT.AE~M.TB+n.of.tr+GAR, data=t50.res)
Problem in lmRob.fit.compute(x2, y, x1 = x1, x1.idx..: x is singular, rank(x)= 4
Use traceback() to see the call stack

> test8<-aovRob(RENT.AE~GAR+MIX.USE)
Problem: Object "RENT.AE" not found
Use traceback() to see the call stack



So, what do I do?

My full data set is attached. Ideally, I'd like to run the following formula looking at the impact of housing variables and accessibility on rent:
RENT.AE~GAR+MIX.USE+NO.OF.FLAT+M.TB+no.of.tr+jFARE .AE+T50Dj

thanks a million,
Angela