I am aiming to estimate the breakpoint of a linear model of how animals are performing (proportion of trials correct, $propcorrect) on a causal reasoning test as they age ($agedays). The goal is to find when does the learning stop increasing and start to plateu? However, it seems not to be registering when I try to set the initial PSI estimate to run the segmented model.
I used the following code to run the lm:
causal.lm <- lm(propcorrect ~ agedays, data = causal_red_data)
summary(causal.lm)
I then used the following code to run the breakpoint analysis:
causal.seg <- segmented.lm(causal.lm,
seg.Z = ~agedays, psi=97))
summary(causal.seg)
The breakpoint looks to me as if it is around 97 days, so I set psi=97. However, it is finding a breakfpoint much earlier instead (probably due to lots of uncertainty around the early days so the line dips a big before it starts steadily increasing), and the output looks like this:
#> causal.seg$psi
#> Initial Est. St.Err
#> psi1.agedays NA 62 1.103706
Why does it say Initial PSI is NA, when I had set it to 97? In the example codes I looked at to learn how to do this, it always showed whatever the user had set psi equal to under "Initial". No matter what I set it to (closer to 62, farther), the Initial comes back as NA. Does this mean it is not reading/taking into to account the PSI estimate I am trying to set? Something wrong with my code?