CBZS: ANES Data
Select variables of interest
V241221 (party id); V241177 (ideology); V241551 (gender); V241501x (race); V242337 (social class); V241465x (education); V241566x (household income); V241458x (age); V242305 (Because of the rich and powerful, it becomes difficult for the rest of us to get ahead); V242350 (minimum wage increase); V242316b (tax on millionaires); V242253x (government should reduce income inequality).
df_anes <- df_anes_raw %>%
rename(party_id = V241221,
ideo = V241177,
gender = V241551,
race = V241501x,
class = V242337,
class_mid = V242338,
edu = V241465x,
income = V241566x,
age = V241458x,
class_zs = V242305,
minwage = V242350,
taxrich = V242316b,
reduceineq = V242253x) %>%
select(party_id,
ideo,
gender,
race,
class,
class_mid,
edu,
income,
age,
class_zs,
minwage,
reduceineq)clean data
df_anes <- df_anes %>%
mutate(party_id = case_when(party_id == 1 ~ "dem",
party_id == 2 ~ "rep",
party_id == 3 ~ "ind",
party_id == 5 ~ "other",
.default = NA),
ideo = case_when(ideo < 0 ~ NA,
ideo == 99 ~ NA,
.default = as.numeric(ideo)),
gender = case_when(gender == 1 ~ "man",
gender == 2 ~ "woman",
gender == 3 ~ "nonbinary",
gender == 4 ~ "other",
.default = NA),
race = case_when(race == 1 ~ "white,non-hisp",
race == 2 ~ "black,non-hisp",
race == 3 ~ "hisp",
race == 4 ~ "asian,non-hisp",
race == 5 ~ "native,non-hisp",
race == 6 ~ "multiracial,non-hisp",
.default = NA),
class = case_when(class == 1 ~ "lower",
class == 2 ~ "working",
class == 3 ~ "middle",
class == 4 ~ "upper",
.default = NA),
class = case_when(class == "middle" & class_mid == 1 ~ "middle-lower",
class == "middle" & class_mid == 2 ~ "middle-mid",
class == "middle" & class_mid == 3 ~ "middle-upper",
.default = as.character(class)),
edu = case_when(edu < 0 ~ NA,
.default = as.numeric(edu)),
income = case_when(income < 0 ~ NA,
.default = as.numeric(income)),
age = case_when(age < 0 ~ NA,
.default = as.numeric(age)),
class_zs = case_when(class_zs < 0 ~ NA,
.default = as.numeric(class_zs)),
minwage = case_when(minwage < 0 ~ NA,
.default = as.numeric(minwage)),
reduceineq = case_when(reduceineq < 0 ~ NA,
.default = as.numeric(reduceineq))) %>%
mutate(minwage_R = 5 - minwage,
reduceineq_R = 8 - reduceineq,
class_num = as.numeric(factor(class,levels = c("lower",
"working",
"middle-lower",
"middle-mid",
"middle-upper",
"upper"))))Lemme see if i can get an objective identity of working class (I’ll just do with vs. without a bachelor’s degree) and a subjective identity of lower and working class (I’ll do lower and working as 1 vs. anything else as 0)
df_anes <- df_anes %>%
mutate(wrkclass_obj = ifelse(edu > 3,0,1),
class_subjbin = ifelse(class == "lower" | class == "working",1,0))Descriptives
Alright, let’s take a look at our sample.
Race
df_anes %>%
group_by(race) %>%
summarise(N = n()) %>%
ungroup() %>%
mutate(Perc = round(100*(N/sum(N)),2)) %>%
ungroup() %>%
arrange(desc(N)) %>%
kbl() %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| race | N | Perc |
|---|---|---|
| white,non-hisp | 3946 | 71.47 |
| hisp | 582 | 10.54 |
| black,non-hisp | 508 | 9.20 |
| asian,non-hisp | 197 | 3.57 |
| multiracial,non-hisp | 188 | 3.41 |
| NA | 67 | 1.21 |
| native,non-hisp | 33 | 0.60 |
Gender
df_anes %>%
group_by(gender) %>%
summarise(N = n()) %>%
ungroup() %>%
mutate(Perc = round(100*(N/sum(N)),2)) %>%
ungroup() %>%
arrange(desc(N)) %>%
kbl() %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| gender | N | Perc |
|---|---|---|
| woman | 2773 | 50.23 |
| man | 2397 | 43.42 |
| NA | 287 | 5.20 |
| nonbinary | 46 | 0.83 |
| other | 18 | 0.33 |
Education
- Less than high-school
- High school
- Some post high-school
- Bachelor’s degree
- Graduate degree
df_anes %>%
group_by(edu) %>%
summarise(N = n()) %>%
ungroup() %>%
mutate(Perc = round(100*(N/sum(N)),2)) %>%
ungroup() %>%
arrange(edu) %>%
kbl() %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| edu | N | Perc |
|---|---|---|
| 1 | 282 | 5.11 |
| 2 | 971 | 17.59 |
| 3 | 1726 | 31.26 |
| 4 | 1374 | 24.89 |
| 5 | 1081 | 19.58 |
| NA | 87 | 1.58 |
Income
1 = Under 5000 to 28 = Over 250,000
df_anes %>%
ggplot(aes(x = income)) +
geom_histogram(fill = "lightblue",
color = NA,
binwidth = 1) +
scale_x_continuous(breaks = seq(1,28,3),
limits = c(0,29)) +
geom_vline(xintercept = median(df_anes$income,na.rm = T),
color = "grey15",
size = 1,
linetype = "dashed") +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.ticks = element_blank(),
axis.line = element_line(color = "grey66"),
axis.text.y = element_text(color = "black"),
axis.text.x = element_text(color = "black",
face = "bold"),
axis.title.x = element_text(color = "black",
face = "bold"))
Class
df_anes %>%
group_by(class) %>%
summarise(N = n()) %>%
ungroup() %>%
mutate(Perc = round(100*(N/sum(N)),2)) %>%
ungroup() %>%
kbl() %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| class | N | Perc |
|---|---|---|
| lower | 418 | 7.57 |
| middle | 96 | 1.74 |
| middle-lower | 540 | 9.78 |
| middle-mid | 1233 | 22.33 |
| middle-upper | 651 | 11.79 |
| upper | 263 | 4.76 |
| working | 1707 | 30.92 |
| NA | 613 | 11.10 |
Class and education
I just want to square them to see how many people identify as working class even when they have a bachelor’s degree
df_anes %>%
group_by(edu,class) %>%
summarise(N = n()) %>%
ungroup() %>%
kbl() %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")## `summarise()` has grouped output by 'edu'. You can override using the `.groups`
## argument.| edu | class | N |
|---|---|---|
| 1 | lower | 45 |
| 1 | middle | 1 |
| 1 | middle-lower | 22 |
| 1 | middle-mid | 40 |
| 1 | middle-upper | 7 |
| 1 | upper | 6 |
| 1 | working | 104 |
| 1 | NA | 57 |
| 2 | lower | 153 |
| 2 | middle | 19 |
| 2 | middle-lower | 72 |
| 2 | middle-mid | 139 |
| 2 | middle-upper | 40 |
| 2 | upper | 18 |
| 2 | working | 398 |
| 2 | NA | 132 |
| 3 | lower | 155 |
| 3 | middle | 24 |
| 3 | middle-lower | 158 |
| 3 | middle-mid | 352 |
| 3 | middle-upper | 112 |
| 3 | upper | 26 |
| 3 | working | 695 |
| 3 | NA | 204 |
| 4 | lower | 43 |
| 4 | middle | 27 |
| 4 | middle-lower | 171 |
| 4 | middle-mid | 364 |
| 4 | middle-upper | 228 |
| 4 | upper | 81 |
| 4 | working | 339 |
| 4 | NA | 121 |
| 5 | lower | 13 |
| 5 | middle | 21 |
| 5 | middle-lower | 113 |
| 5 | middle-mid | 316 |
| 5 | middle-upper | 259 |
| 5 | upper | 131 |
| 5 | working | 148 |
| 5 | NA | 80 |
| NA | lower | 9 |
| NA | middle | 4 |
| NA | middle-lower | 4 |
| NA | middle-mid | 22 |
| NA | middle-upper | 5 |
| NA | upper | 1 |
| NA | working | 23 |
| NA | NA | 19 |
Subjective and objective class
I binarized education and subjective class, to see in an interaction
later on.
wrkclass_obj: 0 = anyone with a bachelor’s degree and
higher; 1 = anyone with less than a bachelor’s degree.
class_subjbin: = 0 = anyone who identified as
lower-middle, middle, upper-middle, or upper; 1 = anyone who identified
as lower or working.
df_anes %>%
group_by(wrkclass_obj,
class_subjbin) %>%
summarise(N = n()) %>%
ungroup() %>%
kbl() %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")## `summarise()` has grouped output by 'wrkclass_obj'. You can override using the
## `.groups` argument.| wrkclass_obj | class_subjbin | N |
|---|---|---|
| 0 | 0 | 1711 |
| 0 | 1 | 543 |
| 0 | NA | 201 |
| 1 | 0 | 1036 |
| 1 | 1 | 1550 |
| 1 | NA | 393 |
| NA | 0 | 36 |
| NA | 1 | 32 |
| NA | NA | 19 |
Age
Note that they just assigned 80 to anyone over 80, so it looks skewed.
df_anes %>%
ggplot(aes(x = age)) +
geom_histogram(fill = "lightblue",
color = NA,
binwidth = 1) +
# scale_x_continuous(breaks = seq(1,28,3),
# limits = c(0,29)) +
geom_vline(xintercept = mean(df_anes$age,na.rm = T),
color = "grey15",
size = 1,
linetype = "dashed") +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.ticks = element_blank(),
axis.line = element_line(color = "grey66"),
axis.text.y = element_text(color = "black"),
axis.text.x = element_text(color = "black",
face = "bold"),
axis.title.x = element_text(color = "black",
face = "bold"))
Party ID
df_anes %>%
group_by(party_id) %>%
summarise(N = n()) %>%
ungroup() %>%
mutate(Perc = round(100*(N/sum(N)),2)) %>%
ungroup() %>%
arrange(desc(N)) %>%
kbl() %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| party_id | N | Perc |
|---|---|---|
| dem | 1865 | 33.78 |
| rep | 1652 | 29.92 |
| ind | 1553 | 28.13 |
| NA | 298 | 5.40 |
| other | 153 | 2.77 |
Political ideology
1 = Extremely Liberal to 7 = Extremely Conservative
df_anes %>%
ggplot(aes(x = ideo)) +
geom_histogram(fill = "lightblue",
color = NA,
binwidth = 1) +
scale_x_continuous(breaks = seq(1,7,1),
limits = c(0,8)) +
geom_vline(xintercept = mean(df_anes$ideo,na.rm = T),
color = "grey15",
size = 1,
linetype = "dashed") +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.ticks = element_blank(),
axis.line = element_line(color = "grey66"),
axis.text.y = element_text(color = "black"),
axis.text.x = element_text(color = "black",
face = "bold"),
axis.title.x = element_text(color = "black",
face = "bold"))
Class Zero-Sum
How well does the following statement describe your view? (1 =
Not at all well to 5 = Extremely well)
Because of the rich and powerful, it becomes difficult for the rest
of us to get ahead.
df_anes %>%
ggplot(aes(x = class_zs)) +
geom_histogram(fill = "lightblue",
color = NA,
binwidth = 1) +
scale_x_continuous(breaks = seq(1,5,1),
limits = c(0,6)) +
geom_vline(xintercept = mean(df_anes$class_zs,na.rm = T),
color = "grey15",
size = 1,
linetype = "dashed") +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.ticks = element_blank(),
axis.line = element_line(color = "grey66"),
axis.text.y = element_text(color = "black"),
axis.text.x = element_text(color = "black",
face = "bold"),
axis.title.x = element_text(color = "black",
face = "bold"))
Minimum wage increase
Should the federal minimum wage be [raised (4), kept the same (3),lowered but not eliminated (2), or eliminated altogether (1)]?
df_anes %>%
ggplot(aes(x = minwage_R)) +
geom_histogram(fill = "lightblue",
color = NA,
binwidth = 1) +
scale_x_continuous(breaks = seq(1,4,1),
limits = c(0,5)) +
geom_vline(xintercept = mean(df_anes$minwage_R,na.rm = T),
color = "grey15",
size = 1,
linetype = "dashed") +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.ticks = element_blank(),
axis.line = element_line(color = "grey66"),
axis.text.y = element_text(color = "black"),
axis.text.x = element_text(color = "black",
face = "bold"),
axis.title.x = element_text(color = "black",
face = "bold"))
Reduce inequality
Do you favor or oppose, or neither favor nor oppose the government trying to reduce the difference in incomes between the richest and poorest households? (1 = Oppose a great deal to 7 = Favor a great deal)
df_anes %>%
ggplot(aes(x = reduceineq_R)) +
geom_histogram(fill = "lightblue",
color = NA,
binwidth = 1) +
scale_x_continuous(breaks = seq(1,7,1),
limits = c(0,8)) +
geom_vline(xintercept = mean(df_anes$reduceineq_R,na.rm = T),
color = "grey15",
size = 1,
linetype = "dashed") +
theme(panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank(),
axis.ticks = element_blank(),
axis.line = element_line(color = "grey66"),
axis.text.y = element_text(color = "black"),
axis.text.x = element_text(color = "black",
face = "bold"),
axis.title.x = element_text(color = "black",
face = "bold"))
Analysis
Correlation matrix
df_anes %>%
select(class_zs,minwage_R,reduceineq,ideo,income,edu,age,class_num) %>%
corPlot(upper = TRUE,stars = TRUE,xsrt = 270)
The “reducing inequality” item looks very weird. Feels like they made a coding error, or the codebook is wrong. I think it’s the opposite of what they’re saying. Hmm. The rest makes sense.
Linear models
Linear model 1A
Outcome: Minimum wage support; predictor: class zero-sum.
m1 <- lm(minwage_R ~ class_zs,data = df_anes)
apa_lm <- apa_print(m1)
kbl(apa_lm$table) %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| term | estimate | conf.int | statistic | df | p.value |
|---|---|---|---|---|---|
| Intercept | 2.81 | [2.74, 2.88] | 82.54 | 4854 | < .001 |
| Class zs | 0.19 | [0.17, 0.21] | 19.15 | 4854 | < .001 |
Linear model 1B
Outcome: Minimum wage support;
Predictor: class zero-sum;
Controls: Party ID and political ideology
m1 <- lm(minwage_R ~ class_zs + party_id + ideo,data = df_anes)
apa_lm <- apa_print(m1)
kbl(apa_lm$table) %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| term | estimate | conf.int | statistic | df | p.value |
|---|---|---|---|---|---|
| Intercept | 3.79 | [3.68, 3.90] | 66.20 | 4045 | < .001 |
| Class zs | 0.10 | [0.08, 0.12] | 8.91 | 4045 | < .001 |
| Party idind | -0.14 | [-0.20, -0.07] | -4.21 | 4045 | < .001 |
| Party idother | -0.47 | [-0.63, -0.31] | -5.67 | 4045 | < .001 |
| Party idrep | -0.29 | [-0.37, -0.21] | -7.04 | 4045 | < .001 |
| Ideo | -0.13 | [-0.15, -0.11] | -13.26 | 4045 | < .001 |
Linear model 1C
Outcome: Minimum wage support;
Predictor: class zero-sum;
Controls: Party ID, political ideology, race, gender, income, education,
and age
m1 <- lm(minwage_R ~ class_zs + party_id + ideo + race + gender + income + edu + age,data = df_anes)
apa_lm <- apa_print(m1)
kbl(apa_lm$table) %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| term | estimate | conf.int | statistic | df | p.value |
|---|---|---|---|---|---|
| Intercept | 3.76 | [3.53, 3.99] | 32.00 | 3676 | < .001 |
| Class zs | 0.10 | [0.08, 0.12] | 8.47 | 3676 | < .001 |
| Party idind | -0.10 | [-0.17, -0.03] | -2.90 | 3676 | .004 |
| Party idother | -0.45 | [-0.62, -0.27] | -5.09 | 3676 | < .001 |
| Party idrep | -0.27 | [-0.36, -0.19] | -6.34 | 3676 | < .001 |
| Ideo | -0.13 | [-0.16, -0.11] | -12.51 | 3676 | < .001 |
| Raceblack,non-hisp | 0.07 | [-0.10, 0.23] | 0.81 | 3676 | .415 |
| Racehisp | 0.10 | [-0.05, 0.26] | 1.30 | 3676 | .192 |
| Racemultiracial,non-hisp | 0.07 | [-0.12, 0.26] | 0.74 | 3676 | .459 |
| Racenative,non-hisp | 0.12 | [-0.28, 0.51] | 0.58 | 3676 | .562 |
| Racewhite,non-hisp | 0.00 | [-0.14, 0.13] | -0.03 | 3676 | .973 |
| Gendernonbinary | 0.13 | [-0.13, 0.38] | 0.97 | 3676 | .331 |
| Genderother | 0.05 | [-0.48, 0.58] | 0.19 | 3676 | .849 |
| Genderwoman | 0.14 | [0.09, 0.19] | 5.41 | 3676 | < .001 |
| Income | 0.00 | [-0.01, 0.00] | -1.21 | 3676 | .228 |
| Edu | -0.02 | [-0.05, 0.00] | -1.87 | 3676 | .061 |
| Age | 0.00 | [0.00, 0.00] | 1.39 | 3676 | .165 |
Linear model 1D
Outcome: Minimum wage support;
Predictor: class zero-sum;
Controls: Party ID, political ideology, race, gender, income, education,
age, and class
m1 <- lm(minwage_R ~ class_zs + party_id + ideo + race + gender + income + edu + age + class,data = df_anes)
apa_lm <- apa_print(m1)
kbl(apa_lm$table) %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| term | estimate | conf.int | statistic | df | p.value |
|---|---|---|---|---|---|
| Intercept | 3.76 | [3.52, 4.00] | 30.28 | 3668 | < .001 |
| Class zs | 0.10 | [0.08, 0.12] | 8.30 | 3668 | < .001 |
| Party idind | -0.10 | [-0.17, -0.04] | -2.97 | 3668 | .003 |
| Party idother | -0.45 | [-0.62, -0.27] | -5.10 | 3668 | < .001 |
| Party idrep | -0.27 | [-0.36, -0.19] | -6.35 | 3668 | < .001 |
| Ideo | -0.13 | [-0.16, -0.11] | -12.43 | 3668 | < .001 |
| Raceblack,non-hisp | 0.07 | [-0.10, 0.23] | 0.80 | 3668 | .423 |
| Racehisp | 0.11 | [-0.05, 0.26] | 1.32 | 3668 | .186 |
| Racemultiracial,non-hisp | 0.07 | [-0.12, 0.26] | 0.72 | 3668 | .471 |
| Racenative,non-hisp | 0.11 | [-0.29, 0.50] | 0.53 | 3668 | .596 |
| Racewhite,non-hisp | 0.00 | [-0.14, 0.14] | 0.01 | 3668 | .989 |
| Gendernonbinary | 0.12 | [-0.13, 0.37] | 0.93 | 3668 | .354 |
| Genderother | 0.04 | [-0.48, 0.57] | 0.16 | 3668 | .873 |
| Genderwoman | 0.14 | [0.09, 0.19] | 5.38 | 3668 | < .001 |
| Income | 0.00 | [0.00, 0.00] | -0.63 | 3668 | .530 |
| Edu | -0.02 | [-0.05, 0.01] | -1.40 | 3668 | .162 |
| Age | 0.00 | [0.00, 0.00] | 1.69 | 3668 | .090 |
| Classmiddle-lower | -0.07 | [-0.19, 0.06] | -1.03 | 3668 | .304 |
| Classmiddle-mid | -0.08 | [-0.20, 0.04] | -1.36 | 3668 | .174 |
| Classmiddle-upper | -0.08 | [-0.21, 0.05] | -1.26 | 3668 | .208 |
| Classupper | 0.00 | [-0.15, 0.15] | 0.00 | 3668 | .998 |
| Classworking | -0.02 | [-0.13, 0.08] | -0.44 | 3668 | .657 |
Linear model 2A
Let’s just see if class is predictive of minimum wage support at all.
Here, we’ll treat class a numeric variable (1 = lower; 2 = working; 3 =
middle; 4 = upper).
Outcome: Minimum wage support;
Predictor: class
m1 <- lm(minwage_R ~ class_num,data = df_anes)
apa_lm <- apa_print(m1)
kbl(apa_lm$table) %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| term | estimate | conf.int | statistic | df | p.value |
|---|---|---|---|---|---|
| Intercept | 3.53 | [3.47, 3.59] | 116.03 | 4779 | < .001 |
| Class num | -0.03 | [-0.05, -0.02] | -3.94 | 4779 | < .001 |
Linear model 2B
Outcome: Minimum wage support;
Predictor: class;
Controls: Education and income
m1 <- lm(minwage_R ~ class_num + edu + income,data = df_anes)
apa_lm <- apa_print(m1)
kbl(apa_lm$table) %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| term | estimate | conf.int | statistic | df | p.value |
|---|---|---|---|---|---|
| Intercept | 3.49 | [3.40, 3.57] | 79.43 | 4376 | < .001 |
| Class num | -0.04 | [-0.06, -0.02] | -3.57 | 4376 | < .001 |
| Edu | 0.04 | [0.02, 0.07] | 3.20 | 4376 | .001 |
| Income | 0.00 | [-0.01, 0.00] | -2.46 | 4376 | .014 |
Linear model 2C
Outcome: Minimum wage support;
Predictor: class;
Controls: Education, income, race, gender, age
m1 <- lm(minwage_R ~ class_num + edu + income + race + gender + age,data = df_anes)
apa_lm <- apa_print(m1)
kbl(apa_lm$table) %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| term | estimate | conf.int | statistic | df | p.value |
|---|---|---|---|---|---|
| Intercept | 3.30 | [3.11, 3.48] | 34.81 | 4239 | < .001 |
| Class num | -0.03 | [-0.05, 0.00] | -2.41 | 4239 | .016 |
| Edu | 0.04 | [0.01, 0.06] | 2.95 | 4239 | .003 |
| Income | 0.00 | [0.00, 0.00] | -0.57 | 4239 | .566 |
| Raceblack,non-hisp | 0.26 | [0.10, 0.42] | 3.20 | 4239 | .001 |
| Racehisp | 0.15 | [-0.01, 0.31] | 1.86 | 4239 | .063 |
| Racemultiracial,non-hisp | 0.07 | [-0.12, 0.26] | 0.74 | 4239 | .457 |
| Racenative,non-hisp | 0.15 | [-0.21, 0.50] | 0.81 | 4239 | .418 |
| Racewhite,non-hisp | -0.02 | [-0.16, 0.12] | -0.31 | 4239 | .760 |
| Gendernonbinary | 0.43 | [0.17, 0.69] | 3.27 | 4239 | .001 |
| Genderother | 0.26 | [-0.18, 0.70] | 1.16 | 4239 | .247 |
| Genderwoman | 0.23 | [0.18, 0.28] | 8.88 | 4239 | < .001 |
| Age | 0.00 | [0.00, 0.00] | -1.08 | 4239 | .282 |
Linear model 2D
Outcome: Minimum wage support;
Predictor: class;
Controls: Education, income, race, gender, age, party ID, and political
ideology
m1 <- lm(minwage_R ~ class_num + edu + income + race + gender + age + party_id + ideo,data = df_anes)
apa_lm <- apa_print(m1)
kbl(apa_lm$table) %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| term | estimate | conf.int | statistic | df | p.value |
|---|---|---|---|---|---|
| Intercept | 4.27 | [4.07, 4.47] | 41.38 | 3678 | < .001 |
| Class num | -0.02 | [-0.04, 0.00] | -2.18 | 3678 | .030 |
| Edu | -0.02 | [-0.05, 0.01] | -1.53 | 3678 | .127 |
| Income | 0.00 | [-0.01, 0.00] | -0.94 | 3678 | .348 |
| Raceblack,non-hisp | 0.07 | [-0.09, 0.24] | 0.84 | 3678 | .399 |
| Racehisp | 0.10 | [-0.05, 0.26] | 1.29 | 3678 | .196 |
| Racemultiracial,non-hisp | 0.07 | [-0.11, 0.26] | 0.78 | 3678 | .436 |
| Racenative,non-hisp | 0.15 | [-0.24, 0.54] | 0.76 | 3678 | .445 |
| Racewhite,non-hisp | 0.01 | [-0.13, 0.14] | 0.08 | 3678 | .935 |
| Gendernonbinary | 0.12 | [-0.14, 0.38] | 0.91 | 3678 | .361 |
| Genderother | 0.07 | [-0.46, 0.61] | 0.27 | 3678 | .783 |
| Genderwoman | 0.14 | [0.09, 0.19] | 5.41 | 3678 | < .001 |
| Age | 0.00 | [0.00, 0.00] | 0.29 | 3678 | .775 |
| Party idind | -0.12 | [-0.19, -0.05] | -3.48 | 3678 | < .001 |
| Party idother | -0.47 | [-0.65, -0.30] | -5.37 | 3678 | < .001 |
| Party idrep | -0.32 | [-0.40, -0.23] | -7.34 | 3678 | < .001 |
| Ideo | -0.15 | [-0.17, -0.13] | -14.48 | 3678 | < .001 |
Linear model 4A
Outcome: Minimum wage support;
Predictor: interaction of subjective and objective class
m1 <- lm(minwage_R ~ wrkclass_obj*class_subjbin,data = df_anes)
apa_lm <- apa_print(m1)
kbl(apa_lm$table) %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| term | estimate | conf.int | statistic | df | p.value |
|---|---|---|---|---|---|
| Intercept | 3.42 | [3.38, 3.46] | 165.75 | 4801 | < .001 |
| Wrkclass obj | -0.11 | [-0.18, -0.05] | -3.36 | 4801 | < .001 |
| Class subjbin | 0.04 | [-0.05, 0.12] | 0.87 | 4801 | .386 |
| Wrkclass obj \(\times\) Class subjbin | 0.15 | [0.05, 0.26] | 2.81 | 4801 | .005 |
interact_plot(m1,
pred = "wrkclass_obj",
modx = "class_subjbin")
Oh. This is counter-intuitive. Let’s see if I’m reading this interaction right. For those who are not objectively working class (those with a bachelor’s degree or higher), their subjective social class doesn’t really matter: they’ll be relatively supporting of increasing the minimum wage regardless. For those are objective working class (those without a bachelor’s degree or higher), if they identify as higher social class than they actually are, they’ll oppose increasing the minimum wage; those who do identify as working or lower class show the same amount of support for minimum wage increase those who are not objectively part of the working class. The story here may be that believe in mobility, or that they separate themselves from their class, and therefore less likely to support policy that’s in their class interest. Ok - that actually makes more sense now.
Linear model 4B
Let’s add some controls, see if the interaction holds.
Outcome: Minimum wage support;
Predictor: interaction of subjective and objective class;
Controls: race, gender, age political ideology
m1 <- lm(minwage_R ~ wrkclass_obj*class_subjbin + race + gender + age + ideo,data = df_anes)
apa_lm <- apa_print(m1)
kbl(apa_lm$table) %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| term | estimate | conf.int | statistic | df | p.value |
|---|---|---|---|---|---|
| Intercept | 4.07 | [3.90, 4.23] | 49.12 | 3815 | < .001 |
| Wrkclass obj | -0.01 | [-0.08, 0.06] | -0.21 | 3815 | .835 |
| Class subjbin | 0.01 | [-0.07, 0.09] | 0.33 | 3815 | .741 |
| Raceblack,non-hisp | 0.16 | [0.00, 0.32] | 1.92 | 3815 | .054 |
| Racehisp | 0.15 | [-0.01, 0.30] | 1.84 | 3815 | .066 |
| Racemultiracial,non-hisp | 0.11 | [-0.08, 0.30] | 1.14 | 3815 | .255 |
| Racenative,non-hisp | 0.20 | [-0.19, 0.59] | 1.00 | 3815 | .319 |
| Racewhite,non-hisp | 0.01 | [-0.13, 0.14] | 0.10 | 3815 | .923 |
| Gendernonbinary | 0.07 | [-0.18, 0.32] | 0.53 | 3815 | .597 |
| Genderother | 0.22 | [-0.29, 0.72] | 0.83 | 3815 | .405 |
| Genderwoman | 0.15 | [0.10, 0.20] | 5.81 | 3815 | < .001 |
| Age | 0.00 | [0.00, 0.00] | 1.22 | 3815 | .224 |
| Ideo | -0.21 | [-0.22, -0.19] | -26.66 | 3815 | < .001 |
| Wrkclass obj \(\times\) Class subjbin | 0.12 | [0.01, 0.23] | 2.19 | 3815 | .029 |
interact_plot(m1,
pred = "wrkclass_obj",
modx = "class_subjbin")
YES. If you take away ideology (which explains a ton in support for minimum wage increase, obviously), the interaction looks a lot more like what we’d expect. For those who do not have a bachelor’s degree or higher, their subjective identification doesn’t really matter. But for thow who do, their self-identification matters a lot.
Linear model 5A
Let’s see what happens when we class_zs to the model. I think we’re
looking for a three-way interaction (yikes).
Outcome: Minimum wage support;
Predictor: interaction of subjective class, objective class, and class
zero-sum beliefs
m1 <- lm(minwage_R ~ wrkclass_obj*class_subjbin*class_zs,data = df_anes)
apa_lm <- apa_print(m1)
kbl(apa_lm$table) %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| term | estimate | conf.int | statistic | df | p.value |
|---|---|---|---|---|---|
| Intercept | 2.73 | [2.62, 2.85] | 47.36 | 4778 | < .001 |
| Wrkclass obj | 0.07 | [-0.11, 0.25] | 0.78 | 4778 | .437 |
| Class subjbin | -0.04 | [-0.28, 0.20] | -0.31 | 4778 | .757 |
| Class zs | 0.22 | [0.19, 0.26] | 12.66 | 4778 | < .001 |
| Wrkclass obj \(\times\) Class subjbin | 0.19 | [-0.12, 0.49] | 1.21 | 4778 | .227 |
| Wrkclass obj \(\times\) Class zs | -0.06 | [-0.11, -0.01] | -2.15 | 4778 | .032 |
| Class subjbin \(\times\) Class zs | 0.00 | [-0.07, 0.06] | -0.11 | 4778 | .913 |
| Wrkclass obj \(\times\) Class subjbin \(\times\) Class zs | 0.00 | [-0.08, 0.09] | 0.06 | 4778 | .949 |
Ok, wait, let’s take a step back.
Linear model 5B
Let’s look at each interaction with class_zs separately.
Outcome: Minimum wage support;
Predictor: interaction of subjective class and class zero-sum
beliefs
m1 <- lm(minwage_R ~ class_subjbin*class_zs,data = df_anes)
apa_lm <- apa_print(m1)
kbl(apa_lm$table) %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| term | estimate | conf.int | statistic | df | p.value |
|---|---|---|---|---|---|
| Intercept | 2.77 | [2.68, 2.86] | 62.50 | 4848 | < .001 |
| Class subjbin | 0.11 | [-0.03, 0.25] | 1.60 | 4848 | .110 |
| Class zs | 0.20 | [0.17, 0.23] | 14.71 | 4848 | < .001 |
| Class subjbin \(\times\) Class zs | -0.02 | [-0.06, 0.02] | -1.02 | 4848 | .307 |
Linear model 5C
Let’s look at each interaction with class_zs separately.
Outcome: Minimum wage support;
Predictor: interaction of objective class and class zero-sum beliefs
m1 <- lm(minwage_R ~ wrkclass_obj*class_zs,data = df_anes)
apa_lm <- apa_print(m1)
kbl(apa_lm$table) %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| term | estimate | conf.int | statistic | df | p.value |
|---|---|---|---|---|---|
| Intercept | 2.73 | [2.63, 2.83] | 53.82 | 4786 | < .001 |
| Wrkclass obj | 0.14 | [0.01, 0.28] | 2.07 | 4786 | .039 |
| Class zs | 0.22 | [0.19, 0.25] | 14.56 | 4786 | < .001 |
| Wrkclass obj \(\times\) Class zs | -0.05 | [-0.09, -0.01] | -2.48 | 4786 | .013 |
interact_plot(m1,
pred = "class_zs",
modx = "wrkclass_obj")
Ok. Class zero-sum matters a lot. And it matters more for those who are not objectively in the working class. That makes sense. I’m guessing this interaction would flow through subjective social class, but oh man, I don’t want to do that mediation model.
Linear model 5D
Let’s add ideology as a control (and others) - I’m sure that’ll
explain away a lot of the variance.
Outcome: Minimum wage support;
Predictor: interaction of objective class and class zero-sum beliefs
Controls: Gender, race, age, political ideology
m1 <- lm(minwage_R ~ wrkclass_obj*class_zs + gender + race + age + ideo,data = df_anes)
apa_lm <- apa_print(m1)
kbl(apa_lm$table) %>%
kable_styling(bootstrap_options = "hover",
full_width = F,
position = "left")| term | estimate | conf.int | statistic | df | p.value |
|---|---|---|---|---|---|
| Intercept | 3.60 | [3.39, 3.80] | 34.68 | 3812 | < .001 |
| Wrkclass obj | 0.04 | [-0.10, 0.18] | 0.61 | 3812 | .543 |
| Class zs | 0.11 | [0.07, 0.14] | 6.65 | 3812 | < .001 |
| Gendernonbinary | 0.08 | [-0.17, 0.33] | 0.60 | 3812 | .550 |
| Genderother | 0.22 | [-0.29, 0.72] | 0.84 | 3812 | .400 |
| Genderwoman | 0.14 | [0.09, 0.19] | 5.74 | 3812 | < .001 |
| Raceblack,non-hisp | 0.15 | [-0.01, 0.31] | 1.85 | 3812 | .065 |
| Racehisp | 0.14 | [-0.01, 0.30] | 1.80 | 3812 | .072 |
| Racemultiracial,non-hisp | 0.10 | [-0.08, 0.29] | 1.10 | 3812 | .272 |
| Racenative,non-hisp | 0.17 | [-0.22, 0.57] | 0.85 | 3812 | .394 |
| Racewhite,non-hisp | 0.00 | [-0.14, 0.13] | -0.01 | 3812 | .991 |
| Age | 0.00 | [0.00, 0.00] | 2.23 | 3812 | .026 |
| Ideo | -0.18 | [-0.20, -0.16] | -21.66 | 3812 | < .001 |
| Wrkclass obj \(\times\) Class zs | 0.00 | [-0.04, 0.04] | 0.04 | 3812 | .969 |
Ok. The interaction is gone. All that remains is class zsb’s.
Mediation
Mediation model 1
Predictor: Class Zero-Sum
Mediator: Class
Outcome: Minimum wage support
m1 <- psych::mediate(minwage_R ~ class_zs + (class_num),data = df_anes,std = T,n.iter = 10000)
beta coefficients are standardized
a = -0.15 (p = 0)
b = -0.02 (p = 0.16)
c = 0.26 (p = 0)
c’= 0.26 (p = 0)
Mediation model 2
Predictor: Class
Mediator: Class Zero-Sum
Outcome: Minimum wage support
m1 <- psych::mediate(minwage_R ~ class_num + (class_zs),data = df_anes,std = T,n.iter = 10000)
a = -0.15 (p = 0)
b = 0.26 (p = 0)
c = -0.06 (p = 0)
c’= -0.02 (p = 0.16)