Go through ESS 9 codebook, identify variables that allow you to generate an fairness of the political system scale. In particular identify variables measuring the aspects listed below. Use the set up of my slides, focus on the same four countries, and add these variables. Use the set up of my slides and add these variables. Conduct a PCA; does it identify a single underlying latent construct?
Political system in country ensures everyone fair chance to participate in politics,
Every one in country fair chance achieve level of education they seek,
Every one in country fair chance get job they seek,
Differences in wealth in country, how fair.
Influence decision to recruit in country: person's gender
ESS <- ESS %>% # Import the ESS, then (pipe result into)
transmute( # Create new variables and keep only those
cntry = as_factor(cntry), # Country of interview
gndr = as_factor(gndr),
facntr = as_factor(facntr), # Father born in cntry
mocntr = as_factor(mocntr), # Mother born in cntry
# Homophobia
freehms = zap_labels(freehms),
hmsfmlsh = max(hmsfmlsh, na.rm = TRUE) - zap_labels(hmsfmlsh),
hmsacld = zap_labels(hmsacld),
# Xenophobia
imbgeco = max(imbgeco, na.rm = TRUE) - zap_labels(imbgeco),
imueclt = max(imueclt, na.rm = TRUE) - zap_labels(imueclt),
imwbcnt = max(imwbcnt, na.rm = TRUE) - zap_labels(imwbcnt),
# Fairness
frprtpl = zap_labels(frprtpl),
evfredu = zap_labels(evfredu),
evfrjob = zap_labels(evfrjob),
wltdffr = zap_labels(wltdffr),
recgndr = zap_labels(recgndr),
pspwght = zap_label(pspwght),
eduyrs = case_when( # Education
eduyrs > 21 ~ 21, # Recode to max 21 years of edu.
eduyrs < 9 ~ 9, # Recode to min 9 years of edu.
TRUE ~ zap_labels(eduyrs))) %>% # Make it numeric, then
dplyr::filter(
# Keep only respondents with native-born parents,
facntr == "Yes" & mocntr == "Yes",
# filter cases from four countries
cntry == "Denmark" | cntry == "Germany" | cntry == "Sweden" | cntry == "Norway")
# Conduct PCA of society's fairness
(fairness_pca <- ESS %>% prcomp(
formula = ~ frprtpl + evfredu + evfrjob + wltdffr + recgndr,
data = ., na.action = na.exclude,
center = TRUE, scale = TRUE))
# Standard deviations (1, .., p=5):
# [1] 1.39 0.99 0.95 0.91 0.59
#
# Rotation (n x k) = (5 x 5):
# PC1 PC2 PC3 PC4 PC5
# frprtpl -0.38 0.24 -0.00091 -0.890 0.045
# evfredu -0.61 0.11 0.20454 0.256 -0.715
# evfrjob -0.60 0.11 0.18934 0.324 0.698
# wltdffr 0.19 0.94 -0.22187 0.176 -0.015
# recgndr 0.30 0.18 0.93439 -0.081 0.012
# Show importance of single components.
summary(fairness_pca)
# Importance of components:
# PC1 PC2 PC3 PC4 PC5
# Standard deviation 1.393 0.987 0.952 0.911 0.5922
# Proportion of Variance 0.388 0.195 0.181 0.166 0.0701
# Cumulative Proportion 0.388 0.583 0.764 0.930 1.0000
# Visualize the results
fviz_pca_var(fairness_pca)
Extract a fariness of the system variable from a PCA of the three items that work together well. Make a plot that shows whether fairness is predicted by years of education. What is the strength of the correlation (rounded to two digits)?
# Conduct PCA of society's fairness
(fairness_pca <- ESS %>% prcomp(
formula = ~ frprtpl + evfredu + evfrjob,
data = ., na.action = na.exclude, scale = TRUE))
# Standard deviations (1, .., p=3):
# [1] 1.35 0.91 0.59
#
# Rotation (n x k) = (3 x 3):
# PC1 PC2 PC3
# frprtpl -0.41 -0.91 0.045
# evfredu -0.65 0.26 -0.715
# evfrjob -0.64 0.32 0.697
# Make a new "fairness_pc" variable and fill it with PC1.
ESS$fairness_pc <- fairness_pca$x[ , "PC1"]
# Plot association with education.
ggplot(data = ESS, aes(y = fairness_pc, x = eduyrs)) +
geom_jitter(alpha = 1/5) +
geom_smooth() +
labs(y = "Perception that society is fair",
x = "Years of education") +
theme_minimal()
# Estimate correlation between the two variables.
ESS %>%
select(fairness_pc, eduyrs) %>%
drop_na() %>% cor()
# fairness_pc eduyrs
# fairness_pc 1.000 -0.026
# eduyrs -0.026 1.000
Perform a large PCA of all xenophobia, homophobia, and the three fairness variables. Can you identify three distinct latent concepts based on all those variables: Xenophobia, Homophobia, and Fairness?
Tip: Use rotation methods to extract meaningful latent variables. Note, with the argument axes = c(1, 2), you can change, which PCs fviz_pca_var() visualizes.
