s-news
[Top] [All Lists]

coefficient?

To: <s-news@lists.biostat.wustl.edu>
Subject: coefficient?
From: "B. Bakker" <bernhard.bakker@wolmail.nl>
Date: Fri, 22 Jun 2001 11:14:25 +0200
Importance: Normal
In-reply-to: <E09E527B56BE2D438A3D6A246DDD27A99CAE@Roper-CV.qld.cmis.csiro.au>
Keywords: Statistics
Reply-to: <bernhard.bakker@wolmail.nl>
Dear All,

When using the spatial linear regression model slm() it produces
"correlation of coefficient estimates" (see example below);

(1) what kind of correlation is this? a Pearson or spearman correlations
coefficient r?
(2) what would be the accepted way to get the slm() to produce
an R-squared?


Example summary:
-----------------------------------------------------------
Residuals:
Min     1Q      Median  3Q      Max
-12     2       8               13      30


Coefficients:
                Value           std error       t value Pr(>|t|)
(Intercept)       1.6           0.2364  6.23            0.0000
my.variable1  0.9               0.4927  4.29            0.0010
my.variable2  0.3               0.1038  7.93            0.0000

Residual standard error: 34.1745 on 96 degrees of freedom

Variance-Covariance matrix of Coefficient
                (Intercept)     my.variable1    my.variable2
(Intercept)             0.055           0.023                   0.003
my.variable1    0.001           0.001                   0.002
my.variable2    0.001           0.001                   0.001


Correlation of coefficient estimates
                (Intercept)     my.variable1    my.variable2
(Intercept)             1               ?                       ?
my.variable1    ?               1                       ?
my.variable2    ?               ?                       1

rho = 0.6434

Iterations = 9
gradient norm = 4.532e-7
Log-likelihood = -200.4

Convergence: RELATIVE FUNCTION CONVERGENCE
----------------end of example------------------------------

Thank you in advance for your consideration.

Bernhard



<Prev in Thread] Current Thread [Next in Thread>
  • coefficient?, B. Bakker <=