Testing contribution of salience in chapter 9.2.7

[Diesfeldt]

The first model omits the random effect for meaning, and includes the random effect of salience; the second model includes random effects for meaning and salience; comparing both models tests the random effect of meaning (which we know already to be very small), not salience. So, my proposal: the first model omits the random effect for meaning (which is a nuisance by a variance nearing zero), and also omits salience in order to find out whether salience is worth it to estimate; the second model includes the random effect for salience only; comparing both models tests the random effect of salience.

anova(l1_random_csal, l1_random2) Data: data Models: l1_random2: lg_rt ~ 1 + c_mean + c_sal + (1 | id) l1_random_csal: lg_rt ~ 1 + c_mean + c_sal + (1 | id) + (0 + c_sal | id) npar AIC BIC logLik deviance Chisq Df Pr(>Chisq) l1_random2 5 16526 16561 -8257.9 16516
l1_random_csal 6 16527 16569 -8257.7 16515 0.5202 1 0.4708