lnre.goodness.of.fit {UCS} | R Documentation |
Evaluate the goodness-of-fit of a LNRE model with a multivariate chi-squared test (Baayen, 2001, Sec. 3.3).
lnre.goodness.of.fit(model, m.max=15)
model |
an object representing a LNRE model whose parameters have
been estimated from observed word frequency data. Currently, the
Zipf-Mandelbrot (ZM, class "zm" ) and the finite
Zipf-Mandelbrot (fZM, class "fzm" ) models are
supported. |
m.max |
highest frequency rank to be included in the evaluation
(limited by the number of ranks stored in the model object). |
This function performs a multivariate chi-squared test to evaluate the goodness-of-fit of an LNRE model (Baayen 2001, p. 119-122).
All LNRE models that follow the UCS/R conventions are supported.
In particular, they must specify the number of parameters estimated
from the observed data (in the n.param
component), and they
must provide appropriate implementations of the EV
, EVm
,
and VV
methods. Currently available LNRE models are objects of class
"zm"
or "fzm"
. The model
object must include observed
frequency data (in components N
, V
, and spc
),
which is usually achieved by estimating the model parameters from the
observed frequency spectrum.
A data frame with one row and three columns:
X2 |
the value of the multi-variate χ^2 test statistic |
df |
the degrees of freedom of the approximate χ^2 distribution of the test statistic under the null hypothesis |
p |
the p-value for the test |
Baayen, R. Harald (2001). Word Frequency Distributions. Kluwer, Dordrecht.