Change in prediction (second prediction - first prediction)

Predictions are collected immediately following the completion of an exam, and immediately before the exam grade is revealed (several days later). This distribution shows the difference between these predictions

Change in prediction conditioned on PHQ score (median split)

Change in prediction between exams (updating)

Multilevel model Summary

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: pred_delta ~ PE_lag1 + (1 + PE_lag1 | id)
##    Data: df
## 
## REML criterion at convergence: 11480.4
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -5.3640 -0.4760 -0.0148  0.5107  5.2881 
## 
## Random effects:
##  Groups   Name        Variance Std.Dev. Corr 
##  id       (Intercept)   0.1942  0.4407       
##           PE_lag1       0.0467  0.2161  -1.00
##  Residual             194.4667 13.9451       
## Number of obs: 1407, groups:  id, 519
## 
## Fixed effects:
##              Estimate Std. Error        df t value Pr(>|t|)    
## (Intercept) 1.348e+00  3.847e-01 1.365e+03   3.503 0.000474 ***
## PE_lag1     3.626e-01  2.784e-02 3.992e+02  13.027  < 2e-16 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##         (Intr)
## PE_lag1 -0.096
## convergence code: 0
## boundary (singular) fit: see ?isSingular

Raw data with regression line

Plotting random effects