Summarize felm model fits
summary method for class
# S3 method for felm summary(object, ..., robust = !is.null(object$clustervar), lhs = NULL)
an object of class
"felm", a result of a call to
further arguments passed to or from other methods.
logical. Use robust standard errors. See notes.
character. If multiple left hand sides, specify the name of the one to obtain a summary for.
summary.felm returns an object of
"summary.felm". It is quite similar to en
object, but not entirely compatible.
"summary.felm" object is a list containing the following fields:
a numerical vector. The residuals of the full system, with dummies.
an integer. The total number of coefficients, including those projected out.
an integer. The number of coefficients, excluding those projected out.
a Pp x 4 matrix with columns for the estimated coefficients, their standard errors, t-statistic and corresponding (two-sided) p-value.
residual standard error.
R^2, the fraction of variance explained by the model.
Projected F-statistic. The result of a
list of factors. A list of the terms in the second part of the model.
object is the result of an estimation with multiple left hand sides,
the actual argument
lhs will be copied to this field.
F-statistic for excluded instruments in 1. step IV, see
The standard errors are adjusted for the reduced degrees of freedom coming from the dummies which are implicitly present. They are also small-sample corrected.
robust parameter is
FALSE, the returned object will
contain ordinary standard errors. If the
robust parameter is
TRUE, clustered standard errors are reported if a cluster was
specified in the call to
felm; if not, heteroskedastic robust
standard errors are reported.
Several F-statistics reported. The
P.fstat is for the projected
system. I.e. a joint test on whether all the
Pp coefficients in
coefficients are zero. Then there are
which is a test on all the coefficients including the ones projected out,
except for an intercept. This statistic assumes i.i.d. errors and is not
reliable for robust or clustered data.
For a 1st stage IV-regression, an F-statistic against the model with excluded instruments is also computed.