Table 3.
Logistic regression and neural network models estimating the influence of pediatricians’ recommendation and participants’ socio-demographic characteristics and caregivers’ COVID-19 vaccine acceptance for children.
| Socio-demographic Predictors | Multinomial logistic regression | MLPNN | ||||
|---|---|---|---|---|---|---|
| −2 Log likelihood of reduced modela | χ 2 | p-value | Rank | Relative importance (%) | Rank | |
| Participants’ COVID Vaccination Status | 758.753 | 108.96 | <.001* | 1 | 20.95% | 1 |
| Pediatricians’ Recommendation Score | 758.143 | 108.35 | <.001* | 2 | 17.46% | 2 |
| Participants’ Post-Vaccination Side Effects | 686.176 | 36.38 | .014* | 3 | 9.50% | 3 |
| HHS Category | 680.39 | 30.60 | .032* | 4 | 8.14% | 5 |
| Race | 673.48 | 23.69 | .049* | 5 | 5.37% | 10 |
| Child's Influenza Vaccination Status | 668.5 | 18.71 | <.001* | 6 | 7.83% | 6 |
| Age | 667.909 | 18.12 | .020* | 7 | 6.12% | 8 |
| Financial Status | 659.542 | 9.75 | 0.136 | 8 | 8.70% | 4 |
| Healthcare Worker | 659.02 | 9.23 | .010* | 9 | 3.32% | 11 |
| Gender | 658.193 | 8.40 | 0.078 | 10 | 6.06% | 9 |
| Level of Education | 655.356 | 5.56 | 0.696 | 11 | 6.55% | 7 |
a
Model Fitting Criteria and the rank list of the predictors in regression model was computed based on the value of “−2 Log Likelihood of Reduced Model”, which estimated the degree of change in regression model if that variable was removed from the analysis. The ranking of the predictors in MLPNN model was determined by their relative importance.
*
p-value <0.05 was considered significant. MLPNN, multilayer perceptron neural network.