sorry for spaming, but IU just had an idea not sure if that may be a way of
doing it:
1. calculate the standrad error of e.g. intercept as mean of the standards
errors obtained from the upper/lower
confidence intervals of the intercept
error_intercept1 = (allo.lmodel2$confidence.intervals[4,2] -
mean(log(biomass_data$BM_roots)))/1.96
error_intercept2 = (allo.lmodel2$confidence.intervals[4,3] -
mean(log(biomass_data$BM_roots)))/1.96
stderr_intercept = round((error_intercept1+error_intercept2)/2,
digits=8)
2. Calculate the t-value as intercept estimate divided by the standard error
from 1. and using
the following for calculating a two-tailed p-value
p_intercept = 2 * (1 - pt(abs(intercept/stderr_intercept),
df=length(biomass_data)-1))
Might this a reasonable approach for a 'rough' estimation of an p-value?
I
glad for every suggestion...
2009/7/20 Katharina May <may.katharina@googlemail.com>
> Hi *,
>
> is there a way to obtain some kind of p-value for a model fitted with RMA
> using the lmodel2 package?
> I know that p-values are discussed and criticized a lot and as you can
> image from my question I'm not
> very much of a statistican (only writing my bachelor thesis).
>
> As fare as I understood the confidence interval statistic correctly, a
> coefficient is regarded as statistically
> significant if the corresponding CI does not include 0 (null hypothesis).
> But can I obtain some kind of a
> p-value to say that it is highly significant (< 0.01), significant
> (0.05),... like in the output of lm?
>
> Sorry for bothering everybody with this, well, probably rather idiotic
> question, but I don't know where to
> continue from this point...
>
> Thanks,
>
> Katharina
>
>
> Here the output of my lmodel2 regression:
>
>
> Model II regression
>
> Call: lmodel2(formula = log(AGB) ~ log(BM_roots), data = biomass_data,
> range.y = "interval", range.x = "interval", nperm = 99)
>
> n = 1969 r = 0.9752432 r-square = 0.9510993
> Parametric P-values: 2-tailed = 0 1-tailed = 0
> Angle between the two OLS regression lines = 1.433308 degrees
>
> Permutation tests of OLS, MA, RMA slopes: 1-tailed, tail corresponding to
> sign
> A permutation test of r is equivalent to a permutation test of the OLS
> slope
> P-perm for SMA = NA because the SMA slope cannot be tested
>
> Regression results
> Method Intercept Slope Angle (degrees) P-perm (1-tailed)
> 1 OLS 0.6122146 1.038792 46.09002 0.01
> 2 MA 0.5787299 1.066868 46.85300 0.01
> 3 SMA 0.5807645 1.065162 46.80725 NA
> 4 RMA 0.5792123 1.066463 46.84216 0.01
>
> Confidence intervals
> Method 2.5%-Intercept 97.5%-Intercept 2.5%-Slope 97.5%-Slope
> 1 OLS 0.5779465 0.6464828 1.028376 1.049207
> 2 MA 0.5659033 0.5914203 1.056227 1.077622
> 3 SMA 0.5682815 0.5931260 1.054797 1.075628
> 4 RMA 0.5663916 0.5918989 1.055826 1.077213
>
> Eigenvalues: 19.83213 0.2475542
>
> H statistic used for computing C.I. of MA: 2.502866e-05
>
>
>
--
Time flies like an arrow, fruit flies like bananas.
[[alternative HTML version deleted]]