RL Regressions
2026-07-29
Setup
## [1] 165## [1] 54## [1] 89## [1] 26## [1] 7## [1] 26## [1] 6## [1] 3## [1] 7Results
All values are z-scored
Final Major mean + std
Tier 1 Drop mean + std
Final Major mean * std
Tier 1 Drop mean * std
Final Major mean * std B/W
Tier 1 Drop mean * std B/W
GLMs
DTZ ~ Alpha
##
## Call:
## glm(formula = "tier1_STEM_drop_tozero ~ alpha_plus_mean_base",
## family = binomial, data = jrk_ABG)
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.52473 0.09227 -16.525 <2e-16 ***
## alpha_plus_mean_base 0.15556 0.08334 1.867 0.0619 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 759.64 on 805 degrees of freedom
## Residual deviance: 756.34 on 804 degrees of freedom
## (175 observations deleted due to missingness)
## AIC: 760.34
##
## Number of Fisher Scoring iterations: 4## Data were 'prettified'. Consider using `terms="alpha_plus_mean_base
## [all]"` to get smooth plots.
##
## Call:
## glm(formula = "final_major_t1_STEM ~ alpha_plus_mean_base", family = binomial,
## data = jrk_ABG)
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.15924 0.07099 2.243 0.02489 *
## alpha_plus_mean_base -0.19520 0.07413 -2.633 0.00846 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1112.3 on 805 degrees of freedom
## Residual deviance: 1105.0 on 804 degrees of freedom
## (175 observations deleted due to missingness)
## AIC: 1109
##
## Number of Fisher Scoring iterations: 4## Data were 'prettified'. Consider using `terms="alpha_plus_mean_base
## [all]"` to get smooth plots.
GLM_FM~alpha_and_other_stuff
##
## Call:
## glm(formula = "final_major_t1_STEM ~ alpha_minus_mean_base + alpha_minus_std_1b + alpha_plus_std_1b + alpha_plus_mean_base",
## family = binomial, data = jrk_ABG)
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.12337 0.08801 1.402 0.1610
## alpha_minus_mean_base 0.36998 0.16758 2.208 0.0273 *
## alpha_minus_std_1b -0.04686 0.12647 -0.371 0.7110
## alpha_plus_std_1b 0.06068 0.10645 0.570 0.5687
## alpha_plus_mean_base -0.16320 0.11517 -1.417 0.1565
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 921.33 on 665 degrees of freedom
## Residual deviance: 907.58 on 661 degrees of freedom
## (315 observations deleted due to missingness)
## AIC: 917.58
##
## Number of Fisher Scoring iterations: 4## Data were 'prettified'. Consider using `terms="alpha_minus_mean_base
## [all]"` to get smooth plots.
##
## Call:
## glm(formula = "final_major_t1_STEM ~ alpha_plus_std_1b + alpha_plus_mean_base + beta_2a_mean_base + beta_2a_std",
## family = binomial, data = jrk_ABG)
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.15982 0.07102 2.250 0.02442 *
## alpha_plus_std_1b -0.04640 0.07714 -0.601 0.54753
## alpha_plus_mean_base -0.21202 0.07826 -2.709 0.00675 **
## beta_2a_mean_base 0.02546 0.10539 0.242 0.80911
## beta_2a_std 0.04733 0.10764 0.440 0.66017
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1112.3 on 805 degrees of freedom
## Residual deviance: 1104.5 on 801 degrees of freedom
## (175 observations deleted due to missingness)
## AIC: 1114.5
##
## Number of Fisher Scoring iterations: 4## Data were 'prettified'. Consider using `terms="alpha_plus_mean_base
## [all]"` to get smooth plots.
##
## Call:
## glm(formula = "final_major_t1_STEM ~ alpha_minus_mean_base + alpha_minus_std_1b + alpha_plus_std_1b + alpha_plus_mean_base + beta_2a_mean_base + beta_2a_std",
## family = binomial, data = jrk_ABG)
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.12691 0.08829 1.437 0.1506
## alpha_minus_mean_base 0.40704 0.17589 2.314 0.0207 *
## alpha_minus_std_1b -0.04734 0.12679 -0.373 0.7089
## alpha_plus_std_1b 0.06923 0.10821 0.640 0.5223
## alpha_plus_mean_base -0.15703 0.11855 -1.325 0.1853
## beta_2a_mean_base 0.04194 0.11446 0.366 0.7141
## beta_2a_std -0.02205 0.11877 -0.186 0.8527
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 921.33 on 665 degrees of freedom
## Residual deviance: 907.05 on 659 degrees of freedom
## (315 observations deleted due to missingness)
## AIC: 921.05
##
## Number of Fisher Scoring iterations: 4## Data were 'prettified'. Consider using `terms="alpha_plus_mean_base
## [all]"` to get smooth plots.
GLM_FM ~ semester study
##
## Call:
## glm(formula = "final_major_t1_STEM ~ alpha_plus_mean_uncty * alpha_minus_mean_uncty + semester_study + UM_credits_at_study",
## family = binomial, data = jrk_ABG)
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) 0.851660 0.164503 5.177
## alpha_plus_mean_uncty -0.239269 0.125269 -1.910
## alpha_minus_mean_uncty 0.281910 0.132591 2.126
## semester_study -0.514198 0.298970 -1.720
## UM_credits_at_study 0.005649 0.019891 0.284
## alpha_plus_mean_uncty:alpha_minus_mean_uncty -0.138183 0.146430 -0.944
## Pr(>|z|)
## (Intercept) 2.25e-07 ***
## alpha_plus_mean_uncty 0.0561 .
## alpha_minus_mean_uncty 0.0335 *
## semester_study 0.0855 .
## UM_credits_at_study 0.7764
## alpha_plus_mean_uncty:alpha_minus_mean_uncty 0.3453
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 921.33 on 665 degrees of freedom
## Residual deviance: 877.51 on 660 degrees of freedom
## (315 observations deleted due to missingness)
## AIC: 889.51
##
## Number of Fisher Scoring iterations: 4## Data were 'prettified'. Consider using `terms="alpha_plus_mean_uncty
## [all]"` to get smooth plots.
##
## Call:
## glm(formula = "final_major_t1_STEM ~ alpha_plus_mean_0b_uncty * alpha_minus_mean_0b_uncty + semester_study + UM_credits_at_study",
## family = binomial, data = jrk_ABG)
##
## Coefficients:
## Estimate Std. Error z value
## (Intercept) 0.851660 0.164503 5.177
## alpha_plus_mean_0b_uncty -0.239269 0.125269 -1.910
## alpha_minus_mean_0b_uncty 0.281910 0.132591 2.126
## semester_study -0.514198 0.298970 -1.720
## UM_credits_at_study 0.005649 0.019891 0.284
## alpha_plus_mean_0b_uncty:alpha_minus_mean_0b_uncty -0.138183 0.146430 -0.944
## Pr(>|z|)
## (Intercept) 2.25e-07 ***
## alpha_plus_mean_0b_uncty 0.0561 .
## alpha_minus_mean_0b_uncty 0.0335 *
## semester_study 0.0855 .
## UM_credits_at_study 0.7764
## alpha_plus_mean_0b_uncty:alpha_minus_mean_0b_uncty 0.3453
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 921.33 on 665 degrees of freedom
## Residual deviance: 877.51 on 660 degrees of freedom
## (315 observations deleted due to missingness)
## AIC: 889.51
##
## Number of Fisher Scoring iterations: 4## Data were 'prettified'. Consider using `terms="alpha_plus_mean_0b_uncty
## [all]"` to get smooth plots.
GLM_FM ~ alpha_plus mean * std (base model)
##
## Call:
## glm(formula = "final_major_t1_STEM ~ alpha_plus_std_1b + alpha_plus_mean_base",
## family = binomial, data = jrk_ABG)
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.15981 0.07101 2.251 0.02441 *
## alpha_plus_std_1b -0.03909 0.07537 -0.519 0.60400
## alpha_plus_mean_base -0.20560 0.07660 -2.684 0.00727 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1112.3 on 805 degrees of freedom
## Residual deviance: 1104.7 on 803 degrees of freedom
## (175 observations deleted due to missingness)
## AIC: 1110.7
##
## Number of Fisher Scoring iterations: 4## Data were 'prettified'. Consider using `terms="alpha_plus_mean_base
## [all]"` to get smooth plots.
##
## Call:
## glm(formula = "final_major_t1_STEM ~ alpha_plus_std_1b * alpha_plus_mean_base",
## family = binomial, data = jrk_ABG)
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.14811 0.07176 2.064 0.03901 *
## alpha_plus_std_1b 0.01125 0.08570 0.131 0.89559
## alpha_plus_mean_base -0.28536 0.10297 -2.771 0.00558 **
## alpha_plus_std_1b:alpha_plus_mean_base -0.04196 0.03620 -1.159 0.24640
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1112.3 on 805 degrees of freedom
## Residual deviance: 1103.3 on 802 degrees of freedom
## (175 observations deleted due to missingness)
## AIC: 1111.3
##
## Number of Fisher Scoring iterations: 4## Data were 'prettified'. Consider using `terms="alpha_plus_mean_base
## [all]"` to get smooth plots.
GLM_FM ~ alpha_plus mean * std (uncertainty model)
##
## Call:
## glm(formula = "final_major_t1_STEM ~ alpha_plus_std_uncty + alpha_plus_mean_uncty",
## family = binomial, data = jrk_ABG)
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.15972 0.07099 2.250 0.02445 *
## alpha_plus_std_uncty -0.03282 0.07384 -0.444 0.65670
## alpha_plus_mean_uncty -0.19332 0.07502 -2.577 0.00997 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1112.3 on 805 degrees of freedom
## Residual deviance: 1105.3 on 803 degrees of freedom
## (175 observations deleted due to missingness)
## AIC: 1111.3
##
## Number of Fisher Scoring iterations: 4## Data were 'prettified'. Consider using `terms="alpha_plus_mean_uncty
## [all]"` to get smooth plots.
##
## Call:
## glm(formula = "final_major_t1_STEM ~ alpha_plus_std_uncty + alpha_plus_mean_uncty",
## family = binomial, data = jrk_ABG)
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.15972 0.07099 2.250 0.02445 *
## alpha_plus_std_uncty -0.03282 0.07384 -0.444 0.65670
## alpha_plus_mean_uncty -0.19332 0.07502 -2.577 0.00997 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 1112.3 on 805 degrees of freedom
## Residual deviance: 1105.3 on 803 degrees of freedom
## (175 observations deleted due to missingness)
## AIC: 1111.3
##
## Number of Fisher Scoring iterations: 4## Data were 'prettified'. Consider using `terms="alpha_plus_mean_uncty
## [all]"` to get smooth plots.
Extra Visualizations
n_negative updates
## `geom_smooth()` using formula = 'y ~ x'
n_positive updates
## `geom_smooth()` using formula = 'y ~ x'
n_any updates
## `geom_smooth()` using formula = 'y ~ x'