03_confirm.R

library(ggplot2)
library(dplyr)

d_mean_only <- read.csv(file="glm_means.csv") %>%
   mutate(ageC=age-mean(age),
          ageC2=ageC^2,
          invageC=1/ageC)

ggplot(d_mean_only) +
   aes(y=AScorr, x=age, color=sex, group=`id`) +
   #geom_point() + geom_line() +
   geom_smooth(aes(group=NULL)) +
   facet_wrap(~roinum) +
   see::theme_modern()


phack <- d_mean_only |> split(d_mean_only$roinum) |> sapply(function(x) coef(summary(lm(ASerrorCo~ageC,data=x)))[2,"Pr(>|t|)"]) 
#          1           2           3           4           5           6 
#0.084942343 0.035268150 0.020171441 0.008404087 0.048423944 0.080818894 
#          7           8           9          10          11          12 
#0.882050746 0.505183076 0.116633781 0.417744784 0.008076829 0.001209913 
#         13          15          16          17          18 
#0.008943929 0.027526725 0.025300533 0.050947150 0.151677015 


# https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3828464/
# >Mean growth curve for error processing (dACC during corrected error trials) indicates increases in the percentage of signal change with age.
FEF_L <- d_mean_only |> filter(roinum==2) # > p=.029 in ordez 2013
dACC <- d_mean_only |> filter(roinum==18) # > p=.000 in ordez 2013
fml <- ASerrorCo ~ ageC + (1|id)
m <- lme4::lmer(fml,data=dACC)

summary(lmerTest::lmer(fml,data=dACC))
#Fixed effects:
#             Estimate Std. Error        df t value Pr(>|t|)    
#(Intercept) 1.570e-02  3.649e-03 7.235e+01   4.302 5.22e-05 ***
#ageC        1.253e-03  8.946e-04 1.326e+02   1.401    0.164    
#Correlation of Fixed Effects:
#     (Intr)
#ageC 0.002 

anova(m)
#Analysis of Variance Table
#     npar    Sum Sq   Mean Sq F value
#ageC    1 0.0074754 0.0074754   1.962