We conducted a field trail, where a harvesting machine repeatedly passed numerous study plots (randomly located in a forest stand). After each pass, rut depth increment was measured.
Number of study plots: 30
8 machine passes on each plot
rutting: rut depth increment per machine pass
so a lmer {lme4} can be fitted, like this:
lmer(rutting ~ 1 + (1|plot), data)
to test intraclass correlation (ICC).
In the given trail, ICC amounts 32%.
Although we have to assume, that the samples are non-independent on each measuring plot, it can not be shown by the ICC, as the variance of the random effect only explains 32% of total variance (random effect + residuals). Hence, variations of i.e. machines' performance might be higher, compared to spatial variance.
In addition, the linear model, without random effects, shows better results regarding desired mean differences.
Is it legit to exclude the random effect, given in plot ?
Dear Marian Schönauer,
I suggest you to see links and attached files on topic.
https://arxiv.org/pdf/1608.01828.pdf
https://www.nature.com/articles/s41467-017-02411-5
https://edge.seas.harvard.edu/files/edge/files/zad-18-upenn.pdf
https://core.ac.uk/download/pdf/81581476.pdf
https://upcommons.upc.edu/bitstream/handle/2117/101278/TVT2601927.pdf?sequence=1&isAllowed=y
Bzst regards
Dear Mohamed-Mourad Lafifi,
thanks for the answer, but the machines mentioned in the questions are not UAVs (?)
Kind regards, Marian
Dear Marian Schönauer,
I regret sorry I was wrong, for your question i suggest you to see links and attached files on subject.
https://stats.idre.ucla.edu/spss/seminars/spss-mixed-command/
https://strengejacke.github.io/mixed-models-snippets/random-effects-within-between-effects-model.html
Best regards
- Badges
- Science topic
- Similar topics
- Automotive Engineering
- Machines