다항 로지스틱 회귀모형을 이용한 서울시민의 정치성향 요인 분석

This paper analyzes the influence of factors such as income, age, gender, education level, and residential area on the political orientation of Seoul citizens using multinomial logistic regression models. The marginal effect of income shows that there is little change across most income brackets, but the probability of being conservative increases sharply in the highest income bracket, while the probabilities of being moderate and liberal decrease. The effect of age indicates that, as expected, the probability of becoming conservative increases and the probability of becoming liberal decreases as age increases. The impact of education is not linear; graduating from college increases the probability of being liberal and decreases the probability of being conservative. However, obtaining a graduate degree, such as a master's or doctorate, results in an increase in both liberal and conservative probabilities, while the probability of being moderate decreases. It was also found that women are more likely to be liberal than men and show very different patterns in the marginal effects of income, age, and education compared to men.