REPLICATIONMARKDOWN
2024-05-08
Analyze only whites
anes_white <- anes_2008 %>% filter(anes_2008$V081102 ==1)variables for scores
anes_white <- anes_2008 %>% filter(anes_2008$V081102 ==1)variables for scores
#variable for score for whites
anes_white$V085174A_clean <- ifelse(anes_white$V085174A %in% c(-2, -8, -9), NA, anes_white$V085174A)
anes_white$V085175A_clean <- ifelse(anes_white$V085175A %in% c(-2, -8, -9), NA, anes_white$V085175A)
anes_white$white_score <- anes_white$V085174A_clean + anes_white$V085175A_clean
anes_white$white_score <- anes_white$white_score/2
# nonwhite variables
anes_white$V085174B_clean <- ifelse(anes_white$V085174B %in% c(-2, -8, -9), NA, anes_white$V085174B)
anes_white$V085174C_clean <- ifelse(anes_white$V085174C %in% c(-2, -8, -9), NA, anes_white$V085174C)
anes_white$V085174D_clean <- ifelse(anes_white$V085174D %in% c(-2, -8, -9), NA, anes_white$V085174D)
anes_white$V085175B_clean <- ifelse(anes_white$V085175B %in% c(-2, -8, -9), NA, anes_white$V085175B)
anes_white$V085175C_clean <- ifelse(anes_white$V085175C %in% c(-2, -8, -9), NA, anes_white$V085175C)
anes_white$V085175D_clean <- ifelse(anes_white$V085175D %in% c(-2, -8, -9), NA, anes_white$V085175D)
anes_white$nonwhite_score <- anes_white$V085174B_clean+anes_white$V085174C_clean+anes_white$V085174D_clean+anes_white$V085175B_clean+anes_white$V085175C_clean+anes_white$V085175D_clean
anes_white$nonwhite_score <- anes_white$nonwhite_score/6
hist(anes_white$white_score - anes_white$nonwhite_score)
# Creating the difference variable
anes_white$diff_score <- anes_white$white_score - anes_white$nonwhite_score
# Normalizing to a scale of -1 to 1
min_diff <- min(anes_white$diff_score, na.rm = TRUE)
max_diff <- max(anes_white$diff_score, na.rm = TRUE)
#preventing division by zero
if (max_diff != min_diff) {
anes_white$normalized_diff <- 2 * ((anes_white$diff_score - min_diff) / (max_diff - min_diff)) - 1
} else {
anes_white$normalized_diff <- 0 # Or some other constant value since all diffs are the same
}
summary(anes_white$normalized_diff)## Min. 1st Qu. Median Mean 3rd Qu. Max. NA's
## -1.00000 0.00000 0.17391 0.08975 0.17391 1.00000 181#obama vote
# Recode 'V085044a' to create 'voteobama'
anes_white <- anes_white %>%
mutate(
voteobama = case_when(
V085044A == 1 ~ 0, # Obama
V085044A == 3 ~ 1, # McCain
V085044A == -1 ~ NA, # Abstained
TRUE ~ NA_real_ # Other cases
)
)
table(anes_white$voteobama)##
## 0 1
## 468 486Note that the echo = FALSE parameter was added to the
code chunk to prevent printing of the R code that generated the
plot.
#Party ID
# Recoding V083097 to create pid308
anes_white <- anes_white %>%
mutate(
pid308 = case_when(
V083097 == 1 ~ 1,
V083097 == 2 ~ 0,
V083097 %in% c(3, 4, 5) ~ 0.5,
TRUE ~ NA_real_
)
)
# Generating pid708 with conditional replacements
anes_white <- anes_white %>%
mutate(
pid708 = NA_real_, # Initialize with NA
pid708 = ifelse(pid308 == 0 & V083098A == 1, 0, pid708),
pid708 = ifelse(pid308 == 0 & V083098A == 5, 0.17, pid708),
pid708 = ifelse(V083098B == 1, 0.33, pid708),
pid708 = ifelse(V083098B == 3, 0.5, pid708),
pid708 = ifelse(V083098B == 5, 0.67, pid708),
pid708 = ifelse(pid308 == 1 & V083098A == 5, 0.83, pid708),
pid708 = ifelse(pid308 == 1 & V083098A == 1, 1, pid708)
)
# Creating pid_new based on the given formula: pid708 * 2 - 1
anes_white <- anes_white %>%
mutate(pid_new = pid708 * 2 - 1)
table(anes_white$pid_new)##
## -1 -0.66 -0.34 0 0.34 0.66 1
## 208 182 182 172 246 211 223#issues
# Recoding Spending and Services
anes_white <- anes_white %>%
mutate(
spdsvc7a = case_when(
V083105 == 1 ~ 0,
V083105 == 2 ~ 0.17,
V083105 == 3 ~ 0.33,
V083105 %in% c(4, -7, -8) ~ 0.5,
V083105 == 5 ~ 0.67,
V083105 == 6 ~ 0.83,
V083105 == 7 ~ 1,
TRUE ~ NA_real_ # For other cases
),
spdsvc7b = case_when(
V083108X == 7 ~ 0,
V083108X == 6 ~ 0.17,
V083108X == 5 ~ 0.33,
V083108X %in% c(4, -7, -8) ~ 0.5,
V083108X == 3 ~ 0.67,
V083108X == 2 ~ 0.83,
V083108X == 1 ~ 1,
TRUE ~ NA_real_
),
spdsvcnew = case_when(
spdsvc7a >= 0 & spdsvc7a <= 2 ~ 0,
spdsvc7b >= 0 & spdsvc7b <= 2 ~ 1,
TRUE ~ NA_real_
),
spdsvc708 = ifelse(spdsvcnew == 1, spdsvc7b, spdsvc7a)
)
# Recode Defense Spending
anes_white <- anes_white %>%
mutate(
spddef7a = case_when(
V083112 == 1 ~ 1,
V083112 == 2 ~ 0.83,
V083112 == 3 ~ 0.67,
V083112 %in% c(4, -7, -8) ~ 0.5,
V083112 == 5 ~ 0.33,
V083112 == 6 ~ 0.17,
V083112 == 7 ~ 0,
TRUE ~ NA_real_
),
spddef7b = case_when(
V083115X == 7 ~ 1,
V083115X == 6 ~ 0.83,
V083115X == 5 ~ 0.67,
V083115X %in% c(4, -7, -8) ~ 0.5,
V083115X == 3 ~ 0.33,
V083115X == 2 ~ 0.17,
V083115X == 1 ~ 0,
TRUE ~ NA_real_
),
spddefnew = case_when(
spddef7a >= 0 & spddef7a <= 2 ~ 0,
spddef7b >= 0 & spddef7b <= 2 ~ 1,
TRUE ~ NA_real_
),
spddef708 = ifelse(spddefnew == 1, spddef7b, spddef7a)
)
# Guaranteed Jobs
anes_white <- anes_white %>%
mutate(
jobsol708 = case_when(
V083128 == 1 ~ 1,
V083128 == 2 ~ 0.83,
V083128 == 3 ~ 0.67,
V083128 %in% c(4, -7, -8) ~ 0.5,
V083128 == 5 ~ 0.33,
V083128 == 6 ~ 0.17,
V083128 == 7 ~ 0,
TRUE ~ NA_real_
)
)
# Health Insurance
anes_white <- anes_white %>%
mutate(
hlthins08 = case_when(
V083119 == 1 ~ 1,
V083119 == 2 ~ 0.83,
V083119 == 3 ~ 0.67,
V083119 %in% c(4, -7, -8) ~ 0.5,
V083119 == 5 ~ 0.33,
V083119 == 6 ~ 0.17,
V083119 == 7 ~ 0,
TRUE ~ NA_real_
)
)
# Calculating missing issue scale
anes_white <- anes_white %>%
mutate(
issuemis = rowSums(is.na(cbind(spdsvc708, spddef708, jobsol708, hlthins08))),
spdsvc708a = ifelse(is.na(spdsvc708), 0, spdsvc708),
spddef708a = ifelse(is.na(spddef708), 0, spddef708),
jobsol708a = ifelse(is.na(jobsol708), 0, jobsol708),
hlthins08a = ifelse(is.na(hlthins08), 0, hlthins08),
issuescale = (spdsvc708a + spddef708a + jobsol708a + hlthins08a) / (4 - issuemis),
issuescale = ifelse(issuemis < 3, issuescale, NA),
issue_new = issuescale * 2 - 1
)
# Summarize the issue scale
hist(anes_white$issuescale)
#economic assessement
# Recoding variables
anes_white <- anes_white %>%
mutate(
ecprog08 = case_when(
V083083X == 1 ~ 1,
V083083X == 2 ~ 0.75,
V083083X == 3 ~ 0.5,
V083083X == 4 ~ 0.25,
V083083X == 5 ~ 0,
TRUE ~ NA_real_ # For other cases
),
ueprog08 = case_when(
V083087X == 1 ~ 1,
V083087X == 2 ~ 0.75,
V083087X == 3 ~ 0.5,
V083087X == 4 ~ 0.25,
V083087X == 5 ~ 0,
TRUE ~ NA_real_
),
inprog08 = case_when(
V083089X == 1 ~ 1,
V083089X == 2 ~ 0.75,
V083089X == 3 ~ 0.5,
V083089X == 4 ~ 0.25,
V083089X == 5 ~ 0,
TRUE ~ NA_real_
)
)
# Creating a data frame with only the required variables
alpha_data <- anes_white %>%
select(ecprog08, ueprog08, inprog08)
# Computing alpha reliability
alpha_result <- alpha(alpha_data, na.rm = TRUE)
# Outputing the alpha result
print(alpha_result)##
## Reliability analysis
## Call: alpha(x = alpha_data, na.rm = TRUE)
##
## raw_alpha std.alpha G6(smc) average_r S/N ase mean sd median_r
## 0.56 0.56 0.46 0.3 1.3 0.02 0.16 0.15 0.28
##
## 95% confidence boundaries
## lower alpha upper
## Feldt 0.52 0.56 0.6
## Duhachek 0.52 0.56 0.6
##
## Reliability if an item is dropped:
## raw_alpha std.alpha G6(smc) average_r S/N alpha se var.r med.r
## ecprog08 0.42 0.42 0.26 0.26 0.72 0.031 NA 0.26
## ueprog08 0.44 0.44 0.28 0.28 0.80 0.029 NA 0.28
## inprog08 0.52 0.52 0.35 0.35 1.09 0.025 NA 0.35
##
## Item statistics
## n raw.r std.r r.cor r.drop mean sd
## ecprog08 1436 0.73 0.75 0.54 0.40 0.11 0.18
## ueprog08 1411 0.76 0.74 0.52 0.38 0.19 0.21
## inprog08 1411 0.71 0.71 0.44 0.33 0.18 0.20
##
## Non missing response frequency for each item
## 0 0.25 0.5 0.75 1 miss
## ecprog08 0.66 0.24 0.08 0.01 0 0.00
## ueprog08 0.47 0.34 0.16 0.03 0 0.02
## inprog08 0.48 0.34 0.17 0.01 0 0.02# Creating ecprog08scale and ecprog_new
anes_white <- anes_white %>%
mutate(
ecprog08scale = (ecprog08 + ueprog08 + inprog08) / 3, # Calculate the average
ecprog_new = ecprog08scale * 2 - 1 # Transform to the desired scale
)
hist(anes_white$ecprog08scale)
#political information
# Initializing and replacing 'ptycon' based on conditions
anes_white <- anes_white %>%
mutate(
ptycon = NA_real_, # Initialize with NA
ptycon = ifelse(V085119A == 5, 1, ptycon), # If V085119a is 5, set to 1
ptycon = ifelse(V085119A %in% c(-8, -9, 1), 0, ptycon),
ptycon = ifelse(V085119 %in% c(5, -8, 9), 0, ptycon) # Ensure correct variable names
)
# Recoding V085066 into 'hormaj' and V085067 into 'senmaj'
anes_white <- anes_white %>%
mutate(
hormaj = case_when(
V085066 == 1 ~ 1,
V085066 == 5 ~ 0,
V085066 %in% c(-8, -9) ~ 0,
TRUE ~ NA_real_ # For other cases
),
senmaj = case_when(
V085067 == 1 ~ 1,
V085067 == 5 ~ 0,
V085067 %in% c(-8, -9) ~ 0,
TRUE ~ NA_real_
)
)
# Calculating 'infopty08' as the average of 'ptycon', 'hormaj', and 'senmaj'
anes_white <- anes_white %>%
mutate(
infopty08 = (ptycon + hormaj + senmaj) / 3
)
hist(anes_white$infopty08)
#education
anes_white <- anes_white %>%
mutate(
educ08 = ifelse(V083217 %in% c(-9, -8), 0, V083217) # Replace -9 and -8 with 0
)
# Tabulate 'educ08'
table(anes_white$educ08)##
## 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
## 4 2 1 4 2 4 14 6 28 31 38 65 428 143 209 72 216 175#Divide 'educ08' by 17
anes_white <- anes_white %>%
mutate(
educ08 = educ08 / 17 # Normalize or scale 'educ08'
)
hist(anes_white$educ08)
#gender
anes_white <- anes_white %>%
mutate(female08 = V081101 - 1)
table(anes_white$female08)##
## 0 1
## 627 815#household ecnomic conditions
anes_white <- anes_white %>%
mutate(
hhecon08 = case_when(
V083057X == 1 ~ 1, # Recode 1 to 1
V083057X == 2 ~ 0.75, # Recode 2 to 0.75
V083057X == 3 ~ 0.5, # Recode 3 to 0.5
V083057X == 4 ~ 0.25, # Recode 4 to 0.25
V083057X == 5 ~ 0, # Recode 5 to 0
TRUE ~ NA_real_ # For other cases, set to NA
),
# Creating 'hhecon_new' by transforming 'hhecon08'
hhecon_new = hhecon08 * 2 - 1 # Transform to a scale of -1 to 1
)
hist(anes_white$hhecon_new)
#MAKE TABLE
library(nnet) # For multinomial logistic regression
library(broom) # For tidy model results
mnl_model <- multinom(
voteobama ~ normalized_diff + hhecon_new + educ08 + infopty08 + ecprog08scale + issuescale + pid_new,
data = anes_white
)## # weights: 9 (8 variable)
## initial value 611.355813
## iter 10 value 317.554146
## iter 20 value 307.195637
## final value 307.194662
## converged# Display a summary of the model
summary(mnl_model)## Call:
## multinom(formula = voteobama ~ normalized_diff + hhecon_new +
## educ08 + infopty08 + ecprog08scale + issuescale + pid_new,
## data = anes_white)
##
## Coefficients:
## Values Std. Err.
## (Intercept) 0.67607185 0.7169357
## normalized_diff -1.40178634 0.4889157
## hhecon_new -0.04128192 0.1769155
## educ08 0.86171925 0.8102326
## infopty08 0.54429186 0.3030066
## ecprog08scale 0.36654037 0.8111269
## issuescale -3.02754187 0.5957933
## pid_new -2.74300780 0.1917186
##
## Residual Deviance: 614.3893
## AIC: 630.3893