test linear effect of height

[teixeirak]

@mcgregorian1, could you please see what the model gives if you use height instead of height_ln? (I ask because theoretical considerations might make this more appropriate-- e.g., McDowell & Allen). Please show coefficients, including for canopy position, as that interacts.

[mcgregorian1]

Do you mean our full (best) model but using position*height?

[teixeirak]

For now, whatever's easiest.

[teixeirak]

I just want to know whether height or height_ln is a stronger predictor.

[mcgregorian1]

With height_ln, the best model include position, tlp, rp, elev, and height_ln as we saw earlier.

With using absolute height, the best model includes the same as above plus dbh (so basically the same).

This looks to me like we're getting completely opposite reactions, but height_ln has a stronger negative reaction than normal height has a positive reaction

[teixeirak]

Please take dbh out of the model so we can compare directly.

[mcgregorian1]

Yep I was about to do that.

With height_ln dAIC of the null model without height_ln = 54.8

With height. dAIC for the null model without height = 23.54

[teixeirak]

Thanks! And what’s the dAIC for height on each?

[mcgregorian1]

I've updated the first comment with the values

[teixeirak]

Okay, so height_ln seems to be the better predictor. Closing this.