summary.manova(object, univariate=F, test="pillai", alias=F, intercept=F)
"manova"
or
"maov"
.
TRUE
, then a univariate anova table is produced for
each response.
If
FALSE
, then a manova table is produced.
"pillai"
,
"wilks lambda"
,
"hotelling-lawley"
, or
"roy largest"
(a partial match is sufficient).
This is ignored if
univariate
is
TRUE
.
TRUE
, then the intercept term is included in the table.
"summary.manova"
, which is a list containing components:
test
.
This is the method for objects of class
"manova"
of the generic function
summary
.
It can be called directly on objects of class
"manova"
or
"maov"
.
There is a choice of four tests, Pillai-Bartlett trace, Wilks' lambda,
Hotelling-Lawley trace, and Roy's largest eigenvalue.
These tests are transformed to an approximate F statistic (which is
exact under some circumstances).
The p-value for Roy's largest eigenvalue test is a lower bound for the
true p-value.
These tests are made under the assumptions of independent Gaussian errors
with common variance matrices.
The default test is the Pillai-Bartlett trace since it appears to be
slightly more robust to failure of these assumptions than the other tests,
and it is the most powerful under some reasonable conditions (see Hand
and Taylor (1987) page 76).
Hand, D. J. and Taylor, C. C. (1987).
Multivariate Analysis of Variance and Repeated Measures.
Chapman and Hall, London.
Mardia, K. V., Kent, J. T. and Bibby, J. M. (1979).
Multivariate Analysis.
Academic Press, London.
Seber, G. A. F., (1984).
Multivariate Observations.
Wiley, New York.
wafer.manova <- manova(cbind(pre.mean, post.mean) ~ maskdim + visc.tem + spinsp, wafer) summary(wafer.manova) # manova table with Pillai's trace summary(wafer.manova, univar=T) # univariate anova tables summary(wafer.manova, test="wilk") # manova table with Wilks' Lambda