# [R] posting a question in the R-help forum

Michael Dewey info at aghmed.fsnet.co.uk
Tue Jan 8 12:39:24 CET 2013

```At 14:14 07/01/2013, Violet Swakman wrote:
>Hello,
>I wanted to post this question below, on the R-help forum, but I'm not sure
>I succeeded because it said that I wasn't subscribed to the mailing list
>yet.
>Now I am subscribed, but will my question be accepted now automatically, or
>should I submit it again?
>Violet Swakman

Violet
You have apparently 20 groups but only 23 observations. I think your
model would work fine with more data. Note also the warning sign of
the correlation of -0.999 for intercept and tarsus length.

>Hello everyone,
>
>I'm having trouble understanding my output from a linear mixed effects
>model (nlme :: lme), I hope someone can help me.
>Say I'm interested in the effect of Tarsus length on the Bar length of
>feathers.
>I used an lme since some birds were living in the same territory, so
>territory was included as random effect.
>Both Bar length and Tarsus length are seen as numerical values, Territory
>is seen as a factor.
>
>#########################
>m1 <- lme(Bar_length~Tarsus_length, random = ~ 1|Territory, data=data)
>
> > summary(m1)
>Linear mixed-effects model fit by REML
>  Data: min_s12
>         AIC       BIC   logLik
>   -104.1593 -99.98116 56.07963
>
>Random effects:
>  Formula: ~1 | as.factor(Territory)
>         (Intercept)   Residual
>StdDev:  0.01023884 0.01072872
>
>Fixed effects: Av_bar_length ~ Tarsus_av
>                   Value              Std.Error       DF    t-value p-value
>(Intercept)  0.22391092 0.08472658 19  2.6427470  0.0160
>Tarsus_av   -0.00048219 0.00338510  2 -0.1424453  0.8998
>  Correlation:
>           (Intr)
>Tarsus_av -0.999
>
>Standardized Within-Group Residuals:
>         Min                 Q1                Med
>Q3             Max
>-2.02920116 -0.49093095  0.05736504  0.47632005  1.15944871
>
>Number of Observations: 23
>Number of Groups: 20
>
>######################
>
>I do not understand why the model needs 17 degrees of freedom to calculate
>1 intercept and slope (just 1 numerical explanatory variable). Could anyone
>maybe explain this to me?
>
>When I use Season, a factor with 2 levels, as an explanatory variable the
>same thing happens, the model takes 17 DF's to calculate the effect of
>Season.
>