# Table 6 Mixed-effects ordinal logistic regression model used to predict score for each problem on a scale of 0 to 4 (N = 230)

Predictor variables B (SE) Odds ratio (95% CI) z X2
Intercept for 0|1 − 2.90 (0.64) 0.05 (0.02, 0.19) − 4.56***
Intercept for 1|2 4.15 (0.60) 63.71 (19.69, 206.11) 6.93***
Intercept for 2|3 5.14 (0.62) 170.27 (51.01, 568.39) 8.35***
Intercept for 3|4 5.86 (0.63) 351.19 (102.48, 1203.49) 9.33***
Media literacy accuracy at pretest 1.40 (0.60) 4.05 (1.26, 12.98) 2.35* 5.56*
Lateral reading at pretest (No = 0) 0.89 (0.46) 2.44 (1.00, 5.98) 1.96 3.81
Instructor (Instructor 1 = 0)a 10.72*
Instructor 2 0.19 (0.36) 1.21 (0.59, 2.45) 0.52
Instructor 3 0.39 (0.35) 1.48 (0.75, 2.91) 1.12
Instructor 4 1.07 (0.36) 2.93 (1.45, 5.90) 3.01**
Problem type (sourcing evidence = 0)a 21.54***
Clickbait science and medical disinformation 0.02 (0.23) 1.02 (0.66, 1.60) 0.11
Fake news 0.79 (0.23) 2.19 (1.40, 3.43) 3.45***
Photographic evidence − 0.12 (0.23) 0.89 (0.56, 1.40) − 0.50
Condition (Control = 0) 1.73 (0.45) 5.66 (2.34, 13.68) 3.85*** 14.71***
Number of assignments attempted 0.48 (0.18) 1.62 (1.15, 2.29) 2.76** 7.63**
1. For instructor, post hoc comparisons with Tukey adjustment for multiple comparisons indicated that instructor 4’s students were more likely to score higher than instructor 1’s students (p = .014). The difference between instructor 4 and instructor 2’s students approached significance (p = .054). For problem type, post hoc comparisons with Tukey adjustment for multiple comparisons indicated that students were more likely to score higher on Fake News than Sourcing Evidence (p = .003), Clickbait Science and Medical Disinformation (p = .002), and Photo Evidence (p < .001)
2. p < .06, *p < .05, **p < .01, ***p < .001
3. aBaselines set based on the lowest number of problems read laterally and correctly assessed at posttest