The Association between Trust in Media and Final Levels of Education across Presidential Voting

Live Poster Session: https://wesleyan.zoom.us/j/97834235378

Abstract: In the past decade, research on voter’s trust in media networks has increased. This study aims to investigate the relationship between voter trust in media and educational attainment across differing presidential voting. Primary variables of media trust, education, and presidential vote were derived from survey responses from the 2024 American National Election Survey. Multivariate regression moderating for the impact of presidential voting showed that while education was statistically significant in prior linear tests, presidential voting in 2024 had a significant relationship to voter’s trust in media. These results indicate the need for further study on the impact of political polarization on trust in institutions such as media across other presidential elections.

poster-2

#set up
library(tidyverse)
library(ggplot2)
library(descr) 
library(dplyr)

load("P:/QAC/qac201/Studies and Codebooks/ANES/Data/anes_2024.RData")

#loading in variables and making my own dataset
var.to.keep<-c("V241334","V241335","V242423","V241012","V241013","V241025","V241463","V241465x","V241466","V242096x")
myData <- anes_2024[,var.to.keep]

#creation of my variables

#media trust related: freq tables, data management, graph
myData$V241335[myData$V241335 %in% c("-9","-8","-1")]<-NA
names(myData)[names(myData)== "V241335"] <- "mediatrust"
freq(myData$ mediatrust)

myData$mediatrust <- factor(myData$mediatrust)
levels(myData$mediatrust) <- c("none", "a little", "a moderate amount", "a lot","a great deal")

myData$quantmediatrust[myData$mediatrust=="none"]<-1
myData$quantmediatrust[myData$mediatrust=="a little"]<-2
myData$quantmediatrust[myData$mediatrust=="a moderate amount"]<-3
myData$quantmediatrust[myData$mediatrust=="a lot"]<-4
myData$quantmediatrust[myData$mediatrust=="a great deal"]<-5

ggplot(data=myData)+
  geom_histogram(aes(x=quantmediatrust))+
  ggtitle("Amount of trust voters have in news outlets")

# party reg related: freq tables 
# NOT PRIMARY VARIABLE
myData$V241025[myData$V241025 %in% c("-9","-8","-1")]<-NA
names(myData)[names(myData)== "V241025"] <- "partyreg"
freq(myData$partyreg)

myData$partyreg <- factor(myData$partyreg)
levels(myData$partyreg) <- c("dem", "gop", "none/indep", "other")

ggplot(data=myData)+
  geom_bar(aes(x=partyreg))+
  ggtitle("amount of voters registered per party")

# voting related: freq table, data management
myData$V242096x[myData$V242096x %in% c("-7","-6","-5","-2")]<-NA 
myData$V242096x[myData$V242096x %in% c("3","4","5")]<-"6"
myData$V242096x[myData$V242096x %in% c("6")] <- "Other"
names(myData)[names(myData)== "V242096x"] <- "presvote"
freq(myData$ presvote)

myData$presvote <- factor(myData$presvote)
levels(myData$presvote) <- c("Harris", "Trump","Other")

ggplot(data=myData)+
  geom_bar(aes(x=presvote))+
  ggtitle("presidential voting in the 2024 election")

# educationn related: freq table,data mangment, graph
myData$V241465x[myData$V241465x %in% c("-9","-8","-4","-2")]<-NA 
names(myData)[names(myData)== "V241465x"] <- "education"
freq(myData$ education)

myData$education <- factor(myData$education)
levels(myData$education) <- c("less than HS", "HS", "some post HS", "BA","MA+")

ggplot(data=myData)+
  geom_bar(aes(x=education))+
  ggtitle("Highest level of education pursured by voters")


# bivariate analysis
# numeric values
by(myData$quantmediatrust, myData$education, mean, na.rm = TRUE)
by(myData$quantmediatrust, myData$education, sd, na.rm = TRUE) 
by(myData$quantmediatrust, myData$education, length) 

by(myData$quantmediatrust, myData$presvote, mean, na.rm = TRUE)
by(myData$quantmediatrust, myData$presvote, sd, na.rm = TRUE) 
by(myData$quantmediatrust, myData$presvote, length) 

# hypothesis testing: ANOVA
myAnovaResults <- aov(quantmediatrust ~ education, data = myData) 
summary(myAnovaResults)

TukeyHSD(myAnovaResults)

#graphs
sub_1 <- filter(myData, is.na(education) != TRUE & is.na(quantmediatrust) != TRUE
ggplot(data=sub_1)+
 geom_boxplot(aes(x=education, y=quantmediatrust)) +
  ggtitle("trust in media across differing levels of education") 


sub_2 <-  filter(myData, is.na(education) != TRUE & is.na(presvote) != TRUE) 
ggplot(data=sub_2, aes(x=education, fill=presvote)) +
  geom_bar(position= "dodge") +
  ylab("vote cast") +
  ggtitle("presidential vote cast across differing levels of education") 

sub_3 <-  filter(myData, is.na(quantmediatrust) != TRUE & is.na(presvote) != TRUE) 
ggplot(data=sub_3)+
  geom_boxplot(aes(x=presvote, y=quantmediatrust)) +
  ggtitle("media trust across differing presidential political votes") 



sub_4 <-  filter(myData, is.na(presvote) != TRUE & is.na(partyreg) != TRUE)
ggplot(data=sub_4, aes(x=partyreg, fill=presvote)) +
  geom_bar(position= "dodge") +
  ylab("vote cast") +
  ggtitle("presidential vote cast across different parties of registration")


sub_5 <-  filter(myData, is.na(education) != TRUE & is.na(quantmediatrust) != TRUE)
ggplot(data=sub_5)+
  geom_boxplot(aes(x=education, y=quantmediatrust))+
  facet_grid(.~presvote)+
  ggtitle("trust in media across different levels of education divided by presidental vote")


#linear regression
my.lm <- lm(quantmediatrust ~ education, data = myData) 
summary(my.lm)
confint(my.lm)

my.lm <- lm(quantmediatrust ~ factor(presvote), data = myData) 
summary(my.lm)
confint(my.lm)

#multiple linear regression
my.lm <- lm(quantmediatrust ~ education + factor(presvote), data = myData) 
summary(my.lm)

my.lm <- lm(quantmediatrust ~ education + factor(presvote)*education, data = myData) 
summary(my.lm)