Hello,
I would be pleased if you could help me in determining the standard error of the
estimate for ?place2L1.N2? from the following table:
Call:
glm(formula = y ~ places2 + shellheight, family = quasibinomial)
Deviance Residuals:
Min 1Q Median 3Q Max
-3.41415 -0.70817 0.01214 0.79700 3.18387
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) -1.60286 0.22383 -7.161 8.92e-12 ***
places2L1.N2 0.31922 0.08095 3.944 0.000104 ***
shellheight 0.08742 0.01135 7.703 3.12e-13 ***
---
Signif. codes: 0 ?***? 0.001 ?**? 0.01 ?*? 0.05 ?.? 0.1 ? ? 1
(Dispersion parameter for quasibinomial family taken to be 1.354491)
Null deviance: 441.61 on 252 degrees of freedom
Residual deviance: 355.52 on 250 degrees of freedom
AIC: NA
Number of Fisher Scoring iterations: 3
I know that the mean value is -1.60286+0.31922 and that the Std. Error (0.08095)
should be the standard error for the difference between ?(Intercept)? and
?places2L1.N2? (i.e. 0.31922).
But when I take this Std. Error as
Sqrt((sA2/nA)+(sB2/nB))
With
sqrt(sA2/nA)
being the Std. Error for the intercept (i.e. 0.22383), then
sqrt (sB2/nB))
should be the Std. Error for places2L1.N2.
But this calculation isn?t right. I don?t get the right Std. Error for the
(Intercept) and therefore also not for places2L1.N2. What?s wrong in this
calculation? Thank you for your help,
Alexander.
_______________________________________________________________________
DSL zum Nulltarif + 20 Euro Extrapr?mie bei Online-Bestellung ?ber die
DSL Freundschaftswerbung! http://dsl.web.de/?ac=OM.AD.AD008K15279B7069a