##Introduction and Background: Data Action Lab for Emerging Societies undertook a nationwide survey of 36,000 registered voters across 20 states via face-to-face interactions. This comprehensive effort aims to create a detailed portrait of the Indian electorate, exploring diverse areas such as electoral preferences, access to welfare, and political leadership. Specifically, the survey includes sections on electoral preferences, access to welfare schemes/benefits, political leadership, perceptions of voters, citizenship, media usage and tailored questions for the states. It aims to probe deep into themes of nationalism, secularism, federalism, and the impact of misinformation, all contextualized through the voters’ social positions.
##Methodology & Sampling Design: Samples were drawn from 200 randomly chosen parliamentary constituencies (PCs) across 20 key states in India. This entails 1,600 polling booths and a total sample size of 38,400 respondents. The selected states account for 513 out of the 543 parliamentary constituencies in the country. Among the 200 PCs, 33 were reserved for Scheduled Castes (SCs) and 16 for Scheduled Tribes (STs).
A systematic random sampling with replacement technique was used to choose 4 assembly constituencies (ACs) from each PC ,i.e., 800 ACs across the country. In each AC, 2 polling booths (PBs) were selected at random, and 24 respondents were randomly chosen from each booth. For the smooth of the survey, a face page was provided to enumerators, containing details about the parliamentary constituency, assembly constituency, polling booth, and the selected and replacement respondents from each booth (with 3 replacements for each respondent). This ensured that the designated sampling design was strictly followed and selection bias was avoided. The use of electoral rolls for each PB was necessary to reduce post-estimation errors and maintain accurate tracking of respondents.
A dictionary of state-level and national welfare schemes was created to account for differences in welfare delivery across various Indian states. The field teams were also encouraged to provide feedback and suggest questions specific to the state’s political landscape. The final questionnaire was translated into selected regional languages, and was later scripted on a survey platform.
al_party <- alliances %>%
dplyr::select(`Party Name`, `Revised Alliance Categories`, `Adjustments Made (Yes/No)`) %>%
filter(`Adjustments Made (Yes/No)` == "No")
ies_study_ls24_dta <- merge(ies_study, al_party, by.x = "ep2", by.y = "Party Name") %>%
rename("Alliance" = `Revised Alliance Categories`)
###Voter Category I
ies_study_ls24_dta <- ies_study_ls24_dta %>%
mutate (voter_cat1 = ifelse(ep2 == "Indian National Congress", "Voted for the Congress",
ifelse(ep2 != "Indian National Congress" & Alliance == "INDIA", "Voted for INDIA but not Congress",
ifelse(ep2 == "Bharatiya Janata Party", "Voted for the BJP",
ifelse(ep2 != "Bharatiya Janata Party" & Alliance == "NDA", "Voted for the NDA but not BJP",
ifelse(ep2 == "Independent Candidate" | ep2 == "Others" | ep2 == "NOTA", "Voted for Independent Candidate/Others/NOTA",
ifelse(ep2 == "Don't want to specify", "Don't want to specify", "Voted for the Third Front")))))))
ies_study_ls24_dta <- ies_study_ls24_dta %>%
mutate (voter_cat2 = ifelse(voter_cat1 == "Voted for the Congress" & pl20 == "Yes", "Voted for the Congress and likes RG",
ifelse(voter_cat1 == "Voted for the Congress" & pl20 == "No", "Voted for the Congress and does not like RG",
ifelse(voter_cat1 == "Voted for INDIA but not Congress" & pl20 == "Yes", "Voted for INDIA and like RG",
ifelse(voter_cat1 == "Voted for INDIA but not Congress" & pl20 == "No", "Voted for INDIA and does not like RG",
ifelse(voter_cat1 == "Voted for the BJP" & pl20 == "Yes", "Voted for the BJP and like RG",
ifelse(voter_cat1 == "Voted for the BJP" & pl20 == "No", "Voted for the BJP and does not like RG",
ifelse(voter_cat1 == "Voted for the NDA but not BJP" & pl20 == "Yes", "Voted for the NDA and like RG",
ifelse(voter_cat1 == "Voted for the NDA but not BJP" & pl20 == "No", "Voted for the NDA and does not like RG",
ifelse( pl20 == "Yes" &
(voter_cat1 == "Voted for the Third Front" | voter_cat1 == "Voted for Independent Candidate/Others/NOTA" | voter_cat1 == "Don't want to specify"),
"Voted for the Third Front/Others/didn't specify and like RG", "Voted for the Third Front/Others/didn't specify and does not like RG"))))))))))
##Descriptive Analysis of Vote Share and Preferences for RG
ies_study_ls24_dta$voter_cat1 <- factor(ies_study_ls24_dta$voter_cat1,
levels = c("Voted for the Congress", "Voted for INDIA but not Congress", "Voted for the BJP",
"Voted for the NDA but not BJP", "Voted for the Third Front", "Voted for Independent Candidate/Others/NOTA"
, "Don't want to specify"))
prop_variables <- c("pl20", "pl33", "pl34", "pl35", "pl36", "pov1", "pov2", "pov3", "pov4", "pov5", "pov6", "pov7", "pov8", "c1", "c5", "c6", "c7", "c8", "c10", "c11", "c12", "mu3")
var_labels <- c("Is Rahul Gandhi a good leader?", "Biggest achievement of BJP government?", "Biggest failure of BJP government?",
"Which system do you prefer?", "Is it okay for a party to give tickets to defectors?", "Do women in your household decide their vote?",
"Should women have reservations in Parliament?", "Should women have reservations in jobs?", "Should scheme money be in women’s accounts?",
"Should there be a caste census?", "Is the EVM electoral process fair?", "Why do you think the constitution is important?",
"Should CM and PM be from the same party?", "Has inequality risen in India?", "Is it a right to demand welfare?",
"How should the government view citizens?", "What is India’s religious character?", "Should temples replace mosques in Mathura, Kashi?",
"Have Hindus been discriminated against historically?", "Will BJP change the constitution if re-elected?",
"Does Congress play appeasement politics?", "What is the most important news source?")
a <- codebook_iesmain %>%
dplyr::select(`Variable Label (English)`, `Variable Name`) %>%
filter(`Variable Name`%in% prop_variables)
# Create an empty list to store plots
plot_list <- list()
# Loop over each variable in prop_variables
for (i in seq_along(prop_variables)) {
var <- prop_variables[i] # Get the current variable
var_label <- var_labels[i] # Get the corresponding label
# Compute proportions for the specific variable
proportion_data <- ies_study_ls24_dta %>%
group_by(voter_cat1, !!sym(var)) %>% # Group by voter_cat1 and the current variable
summarise(count = n()) %>%
mutate(proportion = count / sum(count)) %>%
ungroup()
# Generate the plot for the current variable
p <- ggplot(proportion_data, aes(x = voter_cat1, y = proportion, fill = !!sym(var))) +
geom_bar(stat = "identity", position = "fill") + # Stacked bars with proportions summing to 1
geom_text(aes(label = scales::percent(proportion, accuracy = 0.1)),
position = position_stack(vjust = 0.5), # Center the labels inside the bars
size = 3, color = "white") +
labs(subtitle = paste("Proportion Chart:", var_label),
x = "Voter Category",
y = "Proportion",
fill = var) +
newtheme + # Assuming `newtheme` is a predefined theme you're using
scale_x_discrete(labels = function(x) str_wrap(x, width = 10)) + # Wrap x-axis labels
theme(legend.position = "right") + # Keep the legend on the right
scale_fill_discrete(labels = function(x) str_wrap(x, width = 20))
# Store the plot in the list
plot_list[[var]] <- p
}
for (var in names(plot_list)) {
print(plot_list[[var]])
}
##Descriptive Analysis of Vote Share, Preferences for RG and selected survey questions
ies_study_ls24_dta$voter_cat2 <- as.factor(ies_study_ls24_dta$voter_cat2)
ies_study_ls24_dta$voter_cat2 <- factor(ies_study_ls24_dta$voter_cat2,
levels = c("Voted for the Congress and likes RG", "Voted for the Congress and does not like RG",
"Voted for INDIA and like RG", "Voted for INDIA and does not like RG",
"Voted for the BJP and like RG", "Voted for the BJP and does not like RG",
"Voted for the NDA and like RG", "Voted for the NDA and does not like RG",
"Voted for the Third Front/Others/didn't specify and like RG", "Voted for the Third Front/Others/didn't specify and does not like RG"))
prop_variables <- c("pl33", "pl34", "pl35", "pl36", "pov1", "pov2", "pov3", "pov4", "pov5", "pov6", "pov7", "pov8", "c1", "c5", "c6", "c7", "c8", "c10", "c11", "c12", "mu3")
var_labels <- c("Biggest achievement of BJP government?", "Biggest failure of BJP government?",
"Which system do you prefer?", "Is it okay for a party to give tickets to defectors?", "Do women in your household decide their vote?",
"Should women have reservations in Parliament?", "Should women have reservations in jobs?", "Should scheme money be in women’s accounts?",
"Should there be a caste census?", "Is the EVM electoral process fair?", "Why do you think the constitution is important?",
"Should CM and PM be from the same party?", "Has inequality risen in India?", "Is it a right to demand welfare?",
"How should the government view citizens?", "What is India’s religious character?", "Should temples replace mosques in Mathura, Kashi?",
"Have Hindus been discriminated against historically?", "Will BJP change the constitution if re-elected?",
"Does Congress play appeasement politics?", "What is the most important news source?")
a <- codebook_iesmain %>%
dplyr::select(`Variable Label (English)`, `Variable Name`) %>%
filter(`Variable Name`%in% prop_variables)
# Create an empty list to store plots
plot_list <- list()
# Loop over each variable in prop_variables
for (i in seq_along(prop_variables)) {
var <- prop_variables[i] # Get the current variable
var_label <- var_labels[i] # Get the corresponding label
# Compute proportions for the specific variable
proportion_data <- ies_study_ls24_dta %>%
group_by(voter_cat2, !!sym(var)) %>% # Group by voter_cat2 and the current variable
summarise(count = n()) %>%
mutate(proportion = count / sum(count)) %>%
ungroup()
# Generate the plot for the current variable
p <- ggplot(proportion_data, aes(x = voter_cat2, y = proportion, fill = !!sym(var))) +
geom_bar(stat = "identity", position = "fill") + # Stacked bars with proportions summing to 1
geom_text(aes(label = scales::percent(proportion, accuracy = 0.1)),
position = position_stack(vjust = 0.5), # Center the labels inside the bars
size = 3, color = "white") +
labs(subtitle = paste("Proportion Chart:", var_label),
x = "Voter Category",
y = "Proportion",
fill = var) +
newtheme + # Assuming `newtheme` is a predefined theme you're using
scale_x_discrete(labels = function(x) str_wrap(x, width = 10)) + # Wrap x-axis labels
theme(legend.position = "right") + # Keep the legend on the right
scale_fill_discrete(labels = function(x) str_wrap(x, width = 20))
# Store the plot in the list
plot_list[[var]] <- p
}
for (var in names(plot_list)) {
print(plot_list[[var]])
}
##Model1: Modelling preferences for RG among all voters ###Defining dependent and independent variables
###Definining Predictors & Outcome Variables
all_voters <- ies_study_ls24_dta %>%
filter(bi15 < 100)
y0 <- ifelse(all_voters$pl20 == "Yes", 1, 0)
##Predictors
#Demography
all_voters$bi1 <- as.factor(all_voters$bi1)
all_voters$bi1 <- relevel(all_voters$bi1, ref = "Gujarat")
s_ <- all_voters$bi1
level_of_education <- all_voters$revised_education
age <- all_voters$bi15
male <- ifelse(all_voters$bi13 == "Male", 1, 0)
female <- ifelse(all_voters$bi13 == "Female", 1, 0)
rural <- ifelse(all_voters$bi11 == "Rural", 1, 0)
urban <- ifelse(all_voters$bi11 == "Urban", 1, 0)
all_voters$caste_rel <- as.factor(all_voters$caste_rel)
all_voters$caste_rel <- relevel(all_voters$caste_rel, ref = "Others/Refused to Mention")
caste_religion <- all_voters$caste_rel
farmers <- ifelse(all_voters$dh3 == "Farmer", 1, 0)
casual_labour <- ifelse(all_voters$dh3 == "Daily wage labourer/Casual Labour", 1, 0)
unemployed <- ifelse(all_voters$dh3 == "Unemployed/Looking for jobs", 1, 0)
#Asset Ownership & Wealth
asset_cat <- all_voters$revised_assetcat
ac <- all_voters$Air_Conditioner
car <- all_voters$Car_Private
#Like or dislike modi
dislike_modi <- ifelse(all_voters$pl1 == "No", 1, 0)
modi_communal_caa <- ifelse(all_voters$pl34 == "Communalisation and Hurting Religious Harmony" | all_voters$pl34 == "Communalisation and Hurting Religious Harmony", 1,0)
modi_unemployment_corrupt <- ifelse(all_voters$pl34 == "Increased Unemployment" | all_voters$pl34 == "Increased Corruption", 1,0)
#Perceptions, Media Ownership and prevailing ideological notions (dummies)
caste_census <- ifelse(all_voters$pov5 == "Yes", 1, 0)
secular <- ifelse(all_voters$c7 == "India is a Secular country where all religions are considered equal", 1, 0)
strongman <- ifelse(all_voters$pl35 ==
"A democracy in which a strong leader can make decisions without interference from the parliament or the courts.", 1, 0)
inequality <- ifelse(all_voters$c1 == "Agree" | all_voters$c1 == "Strongly Agree", 1, 0)
bjp_constitution_change <- ifelse(all_voters$c11 == "Agree" | all_voters$c11 == "Strongly Agree" , 1, 0)
inc_appeasement <- ifelse(all_voters$c12 == "Agree" | all_voters$c12 == "Strongly Agree" , 1, 0)
###Summary Statistics: All Voters
SummaryStats_allvoters <- ies_study_ls24_dta %>%
dplyr::select(pl20, bi1, bi11, bi13, revised_religion, dh3, dh5, asset_count, voter_cat1, voter_cat2, pov5, c7, pl35, pl1, revised_assetcat, revised_education, marg_prom) %>%
rename("Pref for RG" = pl20,
"State" = bi1,
"Presence of Congress" = marg_prom,
"Urban/Rural" = bi11,
"Gender" = bi13,
"Primary Activity" = dh3,
"Official Caste" = dh5,
"Voting Pref (LS 24)" = voter_cat1,
"Voting Pref and like RG (LS24)" = voter_cat2,
"Caste Census" = pov5,
"Secular/Hindu" = c7,
"Leadership" = pl35,
"Pref for Modi" = pl1,
"Asset Cateogry (1-4)" = revised_assetcat,
"Education Category (0-8)" = revised_education ) %>%
mutate(Leadership = ifelse(Leadership == "A democracy in which a strong leader can make decisions without interference from the parliament or the courts.", "Strongman", "Consensus Building"))
SummaryStats_allvoters$asset_count <- as.factor(SummaryStats_allvoters$asset_count)
summary_output_allvoters <- dfSummary(SummaryStats_allvoters,
varnumbers = TRUE,
valid.col = FALSE,
title = "Summary Statistics: Dependent Variables (Among All Voters)")
# Render the summary output
print(summary_output_allvoters, file = "dfSummary_DepVariables.html")
includeHTML("dfSummary_DepVariables.html")
Data Frame Summary
Data Frame Summary
SummaryStats_allvoters
Dimensions: 33365 x 17
Duplicates: 2805
No
Variable
Stats / Values
Freqs (% of Valid)
Graph
Missing
1
Pref for RG
[character]
1. No
2. Yes
13276
(
39.8%
)
20089
(
60.2%
)
0
(0.0%)
2
State
[character]
1. Uttar Pradesh
2. Maharashtra
3. Bihar
4. West Bengal
5. Tamil Nadu
6. Andhra Pradesh
7. Madhya Pradesh
8. Rajasthan
9. Karnataka
10. Odisha
[ 10 others ]
5453
(
16.3%
)
3282
(
9.8%
)
2999
(
9.0%
)
2881
(
8.6%
)
2244
(
6.7%
)
2057
(
6.2%
)
1712
(
5.1%
)
1667
(
5.0%
)
1535
(
4.6%
)
1423
(
4.3%
)
8112
(
24.3%
)
0
(0.0%)
3
Urban/Rural
[character]
1. Rural
2. Urban
24630
(
73.8%
)
8735
(
26.2%
)
0
(0.0%)
4
Gender
[character]
1. Female
2. Male
3. Others
15212
(
45.6%
)
18149
(
54.4%
)
4
(
0.0%
)
0
(0.0%)
5
revised_religion
[character]
1. Hinduism
2. Islam
3. Other Minorities
4. Refused to Mention
26912
(
80.7%
)
4642
(
13.9%
)
1719
(
5.2%
)
92
(
0.3%
)
0
(0.0%)
6
Primary Activity
[character]
1. Aged/Retired
2. Daily wage labourer/Casua
3. Farmer
4. Government Job
5. Housewife
6. Others
7. Private job (Regular sala
8. Self-Employed/Business
9. Student
10. Unemployed/Looking for jo
11. Working
1812
(
5.4%
)
4395
(
13.2%
)
7098
(
21.3%
)
925
(
2.8%
)
8045
(
24.1%
)
551
(
1.7%
)
3567
(
10.7%
)
3203
(
9.6%
)
2006
(
6.0%
)
1762
(
5.3%
)
1
(
0.0%
)
0
(0.0%)
7
Official Caste
[character]
1. General
2. Other Backward Class (OBC
3. Scheduled Caste (SC)/Dali
4. Scheduled Tribe (ST)/Adiv
9682
(
29.0%
)
15228
(
45.6%
)
5835
(
17.5%
)
2620
(
7.9%
)
0
(0.0%)
8
asset_count
[factor]
1. 0
2. 1
3. 2
4. 3
5. 4
6. 5
7. 6
8. 7
9. 8
10. 9
[ 3 others ]
1869
(
5.6%
)
5735
(
17.2%
)
8178
(
24.5%
)
8033
(
24.1%
)
4384
(
13.1%
)
2322
(
7.0%
)
1301
(
3.9%
)
716
(
2.1%
)
430
(
1.3%
)
296
(
0.9%
)
101
(
0.3%
)
0
(0.0%)
9
Voting Pref (LS 24)
[factor]
1. Voted for the Congress
2. Voted for INDIA but not C
3. Voted for the BJP
4. Voted for the NDA but not
5. Voted for the Third Front
6. Voted for Independent Can
7. Don't want to specify
6639
(
19.9%
)
6433
(
19.3%
)
11239
(
33.7%
)
2401
(
7.2%
)
3164
(
9.5%
)
473
(
1.4%
)
3016
(
9.0%
)
0
(0.0%)
10
Voting Pref and like RG (LS24)
[factor]
1. Voted for the Congress an
2. Voted for the Congress an
3. Voted for INDIA and like
4. Voted for INDIA and does
5. Voted for the BJP and lik
6. Voted for the BJP and doe
7. Voted for the NDA and lik
8. Voted for the NDA and doe
9. Voted for the Third Front
10. Voted for the Third Front
5786
(
17.3%
)
853
(
2.6%
)
4685
(
14.0%
)
1748
(
5.2%
)
4254
(
12.7%
)
6985
(
20.9%
)
1239
(
3.7%
)
1162
(
3.5%
)
4125
(
12.4%
)
2528
(
7.6%
)
0
(0.0%)
11
Caste Census
[character]
1. Can't Say
2. No
3. Yes
6179
(
18.5%
)
5993
(
18.0%
)
21193
(
63.5%
)
0
(0.0%)
12
Secular/Hindu
[character]
1. India is a Hindu country
2. India is a Secular countr
6713
(
20.1%
)
26652
(
79.9%
)
0
(0.0%)
13
Leadership
[character]
1. Consensus Building
2. Strongman
25582
(
76.7%
)
7783
(
23.3%
)
0
(0.0%)
14
Pref for Modi
[character]
1. No
2. Yes
11023
(
33.0%
)
22342
(
67.0%
)
0
(0.0%)
15
Asset Cateogry (1-4)
[numeric]
Mean (sd) : 1.7 (0.8)
min ≤ med ≤ max:
1 ≤ 2 ≤ 4
IQR (CV) : 1 (0.5)
1
:
15782
(
47.3%
)
2
:
12417
(
37.2%
)
3
:
4339
(
13.0%
)
4
:
827
(
2.5%
)
0
(0.0%)
16
Education Category (0-8)
[numeric]
Mean (sd) : 3.5 (2.4)
min ≤ med ≤ max:
0 ≤ 4 ≤ 8
IQR (CV) : 3 (0.7)
0
:
5556
(
16.7%
)
1
:
2147
(
6.4%
)
2
:
4156
(
12.5%
)
3
:
4015
(
12.0%
)
4
:
4849
(
14.5%
)
5
:
6105
(
18.3%
)
6
:
1049
(
3.1%
)
7
:
4752
(
14.2%
)
8
:
736
(
2.2%
)
0
(0.0%)
17
Presence of Congress
[character]
1. Marginal
2. Prominent
18549
(
55.6%
)
14816
(
44.4%
)
0
(0.0%)
Generated by summarytools 1.0.0 (R version 4.0.3) 2024-10-15
###Cross-tabs: Preference for RG and dependent Variables (Among All Voters)
## Caste Categories
## General Other Backward Class (OBC) Scheduled Caste (SC)/Dalit
## No 44.42 37.35 37.58
## Yes 55.58 62.65 62.42
## Caste Categories
## Scheduled Tribe (ST)/Adivasi
## No 42.10
## Yes 57.90
print(religion_pl20)
## Religion
## Hinduism Islam Other Minorities Refused to Mention
## No 43.95 16.74 37.26 41.30
## Yes 56.05 83.26 62.74 58.70
print(nu_pl20)
## Nature of Locality
## Rural Urban
## No 40.49 37.91
## Yes 59.51 62.09
print(gender_pl20)
## Gender
## Female Male Others
## No 39.63 39.97 25.00
## Yes 60.37 60.03 75.00
## Primary Activity
## Aged/Retired Daily wage labourer/Casual Labour Farmer Government Job
## No 37.52 35.05 41.10 43.89
## Yes 62.48 64.95 58.90 56.11
## Primary Activity
## Housewife Others Private job (Regular salaried) Self-Employed/Business
## No 40.87 42.29 40.72 39.55
## Yes 59.13 57.71 59.28 60.45
## Primary Activity
## Student Unemployed/Looking for jobs Working
## No 41.17 38.23 0.00
## Yes 58.83 61.77 100.00
print(secular_pl20)
## Secular/Hindu Country
## India is a Hindu country
## No 53.53
## Yes 46.47
## Secular/Hindu Country
## India is a Secular country where all religions are considered equal
## No 36.36
## Yes 63.64
print(castecensus_pl20)
## Opinion on Caste Census
## Can't Say No Yes
## No 48.16 48.11 35.03
## Yes 51.84 51.89 64.97
print(leadership_pl20)
## Nature of Political Leadership
## A democracy in which a strong leader can make decisions without interference from the parliament or the courts.
## No 46.70
## Yes 53.30
## Nature of Political Leadership
## A democracy in which elected representatives decide what becomes law.
## No 37.71
## Yes 62.29
###Model 1a: Preference for RG and Age, Gender (Dummy) and Nature of Locality (Dummy); Among All Voters
logit.fit.1a <- glm(y0 ~ s_ + age + female + rural + level_of_education, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.1a)
###Model 1b: Preference for RG ~ Religion (Ref == Hinduism), Caste (Ref == General); Among All Voters
logit.fit.1b <- glm(y0 ~ caste_religion, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.1b)
##
## Call:
## glm(formula = y0 ~ caste_religion, family = binomial(link = "logit"))
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.8906 -1.3140 0.6054 1.0467 1.2045
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.51230 0.04860 10.541 < 2e-16 ***
## caste_religionHindu and OBC -0.19675 0.05193 -3.789 0.000151 ***
## caste_religionHindu and SC -0.01746 0.05663 -0.308 0.757800
## caste_religionHindu and ST -0.22993 0.06364 -3.613 0.000303 ***
## caste_religionHindu and UC -0.57579 0.05405 -10.652 < 2e-16 ***
## caste_religionMuslim 1.09171 0.06253 17.459 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 44783 on 33308 degrees of freedom
## Residual deviance: 43169 on 33303 degrees of freedom
## AIC: 43181
##
## Number of Fisher Scoring iterations: 4
###Model 1c: Preference for RG ~ Primary Acitivty Dummies: Farmers, Casual Labourers, Unemployed (Among All Voters)
logit.fit.1c <- glm(y0 ~ farmers + unemployed + casual_labour, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.1c)
##
## Call:
## glm(formula = y0 ~ farmers + unemployed + casual_labour, family = binomial(link = "logit"))
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.4481 -1.3438 0.9815 1.0196 1.0289
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.38310 0.01438 26.649 < 2e-16 ***
## farmers -0.02330 0.02810 -0.829 0.4070
## unemployed 0.09690 0.05114 1.895 0.0581 .
## casual_labour 0.23379 0.03474 6.729 1.71e-11 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 44783 on 33308 degrees of freedom
## Residual deviance: 44730 on 33305 degrees of freedom
## AIC: 44738
##
## Number of Fisher Scoring iterations: 4
###Model 1d: Preference for RG ~ Assent Rank (Numeric Range; c(1,5)), Ownership of AC (dummy), Ownership of Car (dummy); Among All Voters
logit.fit.1d <- glm(y0 ~ asset_cat, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.1d)
##
## Call:
## glm(formula = y0 ~ asset_cat, family = binomial(link = "logit"))
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.365 -1.354 1.001 1.011 1.031
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.45665 0.02679 17.046 <2e-16 ***
## asset_cat -0.02538 0.01424 -1.783 0.0746 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 44783 on 33308 degrees of freedom
## Residual deviance: 44780 on 33307 degrees of freedom
## AIC: 44784
##
## Number of Fisher Scoring iterations: 4
###Model 1e: Preference for RG ~ Political Positions: Secular/Hindu Country (dummy), Leadership (dummy), Caste Census (dummy), Likes Modi (dummy); Among All Voters
logit.fit.1e <- glm(y0 ~ secular + strongman + caste_census + bjp_constitution_change + inequality + inc_appeasement, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.1e)
##
## Call:
## glm(formula = y0 ~ secular + strongman + caste_census + bjp_constitution_change +
## inequality + inc_appeasement, family = binomial(link = "logit"))
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.8234 -1.1832 0.7845 0.9781 1.8768
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.73923 0.03584 -20.62 <2e-16 ***
## secular 0.62304 0.02876 21.66 <2e-16 ***
## strongman -0.23743 0.02757 -8.61 <2e-16 ***
## caste_census 0.41109 0.02466 16.67 <2e-16 ***
## bjp_constitution_change 0.43129 0.02560 16.85 <2e-16 ***
## inequality 0.72591 0.02697 26.92 <2e-16 ***
## inc_appeasement -0.59609 0.02540 -23.46 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 44783 on 33308 degrees of freedom
## Residual deviance: 42085 on 33302 degrees of freedom
## AIC: 42099
##
## Number of Fisher Scoring iterations: 4
###Model 1f: Preference for RG ~ Biggest Failure of BJP: Communalisation, Unemoployment (dummies); Among All Voters
logit.fit.1f <- glm(y0 ~ modi_communal_caa + modi_unemployment_corrupt, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.1f)
##
## Call:
## glm(formula = y0 ~ modi_communal_caa + modi_unemployment_corrupt,
## family = binomial(link = "logit"))
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.3916 -1.3916 0.9774 0.9774 1.1498
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.06526 0.02831 2.305 0.0211 *
## modi_communal_caa 0.31587 0.04437 7.119 1.08e-12 ***
## modi_unemployment_corrupt 0.42539 0.03119 13.639 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 44783 on 33308 degrees of freedom
## Residual deviance: 44598 on 33306 degrees of freedom
## AIC: 44604
##
## Number of Fisher Scoring iterations: 4
###Model 1g: Preference for RG ~ All dependent variables; Among All Voters
## Caste Categories
## General Other Backward Class (OBC) Scheduled Caste (SC)/Dalit
## No 51.23 43.44 44.58
## Yes 48.77 56.56 55.42
## Caste Categories
## Scheduled Tribe (ST)/Adivasi
## No 50.93
## Yes 49.07
print(religion_pl20)
## Religion
## Hinduism Islam Other Minorities Refused to Mention
## No 49.70 22.24 45.70 48.10
## Yes 50.30 77.76 54.30 51.90
print(nu_pl20)
## Nature of Locality
## Rural Urban
## No 46.82 45.59
## Yes 53.18 54.41
print(gender_pl20)
## Gender
## Female Male Others
## No 45.99 46.95 33.33
## Yes 54.01 53.05 66.67
## Primary Activity
## Aged/Retired Daily wage labourer/Casual Labour Farmer Government Job
## No 44.16 41.55 47.96 50.07
## Yes 55.84 58.45 52.04 49.93
## Primary Activity
## Housewife Others Private job (Regular salaried) Self-Employed/Business
## No 47.54 52.31 46.91 46.03
## Yes 52.46 47.69 53.09 53.97
## Primary Activity
## Student Unemployed/Looking for jobs
## No 48.09 45.00
## Yes 51.91 55.00
print(secular_pl20)
## Secular/Hindu Country
## India is a Hindu country
## No 58.36
## Yes 41.64
## Secular/Hindu Country
## India is a Secular country where all religions are considered equal
## No 43.37
## Yes 56.63
print(castecensus_pl20)
## Opinion on Caste Census
## Can't Say No Yes
## No 55.28 52.95 42.00
## Yes 44.72 47.05 58.00
print(leadership_pl20)
## Nature of Political Leadership
## A democracy in which a strong leader can make decisions without interference from the parliament or the courts.
## No 51.88
## Yes 48.12
## Nature of Political Leadership
## A democracy in which elected representatives decide what becomes law.
## No 44.81
## Yes 55.19
###Model 2a: Preference for RG and Age, Gender (Dummy) and Nature of Locality (Dummy); Among Non-Congress Voters
logit.fit.2a <- glm(y0 ~ s_ + age + female + rural + level_of_education, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.2a)
logit.fit.2c <- glm(y0 ~ farmers + unemployed + casual_labour, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.2c)
##
## Call:
## glm(formula = y0 ~ farmers + unemployed + casual_labour, family = binomial(link = "logit"))
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.325 -1.226 1.036 1.130 1.143
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.11257 0.01573 7.157 8.24e-13 ***
## farmers -0.03105 0.03099 -1.002 0.316
## unemployed 0.08810 0.05598 1.574 0.116
## casual_labour 0.22888 0.03799 6.024 1.70e-09 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 36860 on 26682 degrees of freedom
## Residual deviance: 36817 on 26679 degrees of freedom
## AIC: 36825
##
## Number of Fisher Scoring iterations: 3
###Model 2d: Preference for RG ~ Assent Rank (Numeric Range; c(1,5)), Ownership of AC (dummy), Ownership of Car (dummy); Among Non-Congress Voters
logit.fit.2d <- glm(y0 ~ asset_cat, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.2d)
##
## Call:
## glm(formula = y0 ~ asset_cat, family = binomial(link = "logit"))
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.239 -1.237 1.117 1.119 1.123
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 0.147643 0.029441 5.015 5.31e-07 ***
## asset_cat -0.004494 0.015464 -0.291 0.771
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 36860 on 26682 degrees of freedom
## Residual deviance: 36860 on 26681 degrees of freedom
## AIC: 36864
##
## Number of Fisher Scoring iterations: 3
###Model 2e: Preference for RG ~ Political Positions: Secular/Hindu Country (dummy), Leadership (dummy), Caste Census (dummy), Likes Modi (dummy); Among Non-Congress Voters
logit.fit.2e <- glm(y0 ~ secular + strongman + caste_census + bjp_constitution_change + inequality + inc_appeasement, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.2e)
##
## Call:
## glm(formula = y0 ~ secular + strongman + caste_census + bjp_constitution_change +
## inequality + inc_appeasement, family = binomial(link = "logit"))
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.6621 -1.1552 0.7607 1.0446 1.9078
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.97315 0.03967 -24.529 < 2e-16 ***
## secular 0.55266 0.03161 17.486 < 2e-16 ***
## strongman -0.16901 0.03012 -5.611 2.01e-08 ***
## caste_census 0.34832 0.02707 12.866 < 2e-16 ***
## bjp_constitution_change 0.42292 0.02799 15.111 < 2e-16 ***
## inequality 0.74125 0.02962 25.029 < 2e-16 ***
## inc_appeasement -0.50086 0.02780 -18.020 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 36860 on 26682 degrees of freedom
## Residual deviance: 34931 on 26676 degrees of freedom
## AIC: 34945
##
## Number of Fisher Scoring iterations: 4
###Model 2f: Preference for RG ~ Biggest Failure of BJP: Communalisation, Unemoployment (dummies); Among Non-Congress Voters
logit.fit.2f <- glm(y0 ~ modi_communal_caa + modi_unemployment_corrupt, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.2f)
##
## Call:
## glm(formula = y0 ~ modi_communal_caa + modi_unemployment_corrupt,
## family = binomial(link = "logit"))
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.257 -1.256 1.100 1.100 1.218
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.09449 0.03092 -3.056 0.00224 **
## modi_communal_caa 0.27927 0.04804 5.814 6.10e-09 ***
## modi_unemployment_corrupt 0.27815 0.03410 8.157 3.43e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 36860 on 26682 degrees of freedom
## Residual deviance: 36792 on 26680 degrees of freedom
## AIC: 36798
##
## Number of Fisher Scoring iterations: 3
###Model 2g: Preference for RG ~ All dependent variables; Among Non-Congress Voters
## Caste Categories
## General Other Backward Class (OBC) Scheduled Caste (SC)/Dalit
## No 56.68 51.49 48.70
## Yes 43.32 48.51 51.30
## Caste Categories
## Scheduled Tribe (ST)/Adivasi
## No 52.16
## Yes 47.84
print(religion_pl20)
## Religion
## Hinduism Islam Other Minorities Refused to Mention
## No 54.45 28.57 46.59 59.57
## Yes 45.55 71.43 53.41 40.43
print(nu_pl20)
## Nature of Locality
## Rural Urban
## No 52.23 53.85
## Yes 47.77 46.15
print(gender_pl20)
## Gender
## Female Male Others
## No 51.45 53.62 33.33
## Yes 48.55 46.38 66.67
## Primary Activity
## Aged/Retired Daily wage labourer/Casual Labour Farmer Government Job
## No 48.44 48.96 53.40 53.52
## Yes 51.56 51.04 46.60 46.48
## Primary Activity
## Housewife Others Private job (Regular salaried) Self-Employed/Business
## No 54.24 58.68 52.00 54.28
## Yes 45.76 41.32 48.00 45.72
## Primary Activity
## Student Unemployed/Looking for jobs
## No 53.43 50.04
## Yes 46.57 49.96
print(secular_pl20)
## Secular/Hindu Country
## India is a Hindu country
## No 64.05
## Yes 35.95
## Secular/Hindu Country
## India is a Secular country where all religions are considered equal
## No 49.37
## Yes 50.63
print(castecensus_pl20)
## Opinion on Caste Census
## Can't Say No Yes
## No 59.20 58.20 48.98
## Yes 40.80 41.80 51.02
print(leadership_pl20)
## Nature of Political Leadership
## A democracy in which a strong leader can make decisions without interference from the parliament or the courts.
## No 57.85
## Yes 42.15
## Nature of Political Leadership
## A democracy in which elected representatives decide what becomes law.
## No 51.09
## Yes 48.91
###Model 3a: Preference for RG and Age, Gender (Dummy) and Nature of Locality (Dummy); Among Non-India Voters
logit.fit.3a <- glm(y0 ~ s_ + age + female + rural + level_of_education, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.3a)
logit.fit.3c <- glm(y0 ~ farmers + unemployed + casual_labour, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.3c)
##
## Call:
## glm(formula = y0 ~ farmers + unemployed + casual_labour, family = binomial(link = "logit"))
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.195 -1.121 -1.120 1.234 1.236
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.13294 0.01811 -7.340 2.13e-13 ***
## farmers -0.00313 0.03514 -0.089 0.9290
## unemployed 0.13116 0.06237 2.103 0.0355 *
## casual_labour 0.17446 0.04423 3.944 8.01e-05 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 28030 on 20259 degrees of freedom
## Residual deviance: 28011 on 20256 degrees of freedom
## AIC: 28019
##
## Number of Fisher Scoring iterations: 3
###Model 3d: Preference for RG ~ Assent Rank (Numeric Range; c(1,5)), Ownership of AC (dummy), Ownership of Car (dummy); Among Non-India Voters
logit.fit.3d <- glm(y0 ~ asset_cat, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.3d)
##
## Call:
## glm(formula = y0 ~ asset_cat, family = binomial(link = "logit"))
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.148 -1.135 -1.128 1.221 1.227
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.13254 0.03355 -3.951 7.79e-05 ***
## asset_cat 0.01566 0.01740 0.900 0.368
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 28030 on 20259 degrees of freedom
## Residual deviance: 28030 on 20258 degrees of freedom
## AIC: 28034
##
## Number of Fisher Scoring iterations: 3
###Model 3e: Preference for RG ~ Political Positions: Secular/Hindu Country (dummy), Leadership (dummy), Caste Census (dummy), Likes Modi (dummy); Among Non-India Voters
logit.fit.3e <- glm(y0 ~ secular + strongman + caste_census + bjp_constitution_change + inequality + inc_appeasement, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.3e)
##
## Call:
## glm(formula = y0 ~ secular + strongman + caste_census + bjp_constitution_change +
## inequality + inc_appeasement, family = binomial(link = "logit"))
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.5302 -1.1131 -0.7021 1.1076 1.9883
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.16003 0.04551 -25.492 < 2e-16 ***
## secular 0.55413 0.03612 15.342 < 2e-16 ***
## strongman -0.18024 0.03535 -5.099 3.42e-07 ***
## caste_census 0.26030 0.03113 8.361 < 2e-16 ***
## bjp_constitution_change 0.37275 0.03218 11.585 < 2e-16 ***
## inequality 0.77229 0.03427 22.534 < 2e-16 ***
## inc_appeasement -0.48738 0.03192 -15.267 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 28030 on 20259 degrees of freedom
## Residual deviance: 26674 on 20253 degrees of freedom
## AIC: 26688
##
## Number of Fisher Scoring iterations: 4
###Model 3f: Preference for RG ~ Biggest Failure of BJP: Communalisation, Unemoployment (dummies); Among Non-India Voters
logit.fit.3f <- glm(y0 ~ modi_communal_caa + modi_unemployment_corrupt, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.3f)
##
## Call:
## glm(formula = y0 ~ modi_communal_caa + modi_unemployment_corrupt,
## family = binomial(link = "logit"))
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.171 -1.142 -1.070 1.214 1.288
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.25730 0.03483 -7.388 1.49e-13 ***
## modi_communal_caa 0.24309 0.05467 4.446 8.74e-06 ***
## modi_unemployment_corrupt 0.17277 0.03855 4.482 7.41e-06 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 28030 on 20259 degrees of freedom
## Residual deviance: 28005 on 20257 degrees of freedom
## AIC: 28011
##
## Number of Fisher Scoring iterations: 3
###Model 3g: Preference for RG ~ All dependent variables; Among Non-India Voters
## Caste Categories
## General Other Backward Class (OBC) Scheduled Caste (SC)/Dalit
## No 62.87 58.28 58.68
## Yes 37.13 41.72 41.32
## Caste Categories
## Scheduled Tribe (ST)/Adivasi
## No 57.75
## Yes 42.25
print(religion_pl20)
## Religion
## Hinduism Islam Other Minorities Refused to Mention
## No 60.05 48.11 59.71 84.62
## Yes 39.95 51.89 40.29 15.38
print(nu_pl20)
## Nature of Locality
## Rural Urban
## No 59.25 61.33
## Yes 40.75 38.67
print(gender_pl20)
## Gender
## Female Male Others
## No 58.77 60.50 50.00
## Yes 41.23 39.50 50.00
## Primary Activity
## Aged/Retired Daily wage labourer/Casual Labour Farmer Government Job
## No 58.08 58.18 58.08 58.60
## Yes 41.92 41.82 41.92 41.40
## Primary Activity
## Housewife Others Private job (Regular salaried) Self-Employed/Business
## No 62.04 70.49 58.25 62.19
## Yes 37.96 29.51 41.75 37.81
## Primary Activity
## Student Unemployed/Looking for jobs
## No 61.10 56.55
## Yes 38.90 43.45
print(secular_pl20)
## Secular/Hindu Country
## India is a Hindu country
## No 66.94
## Yes 33.06
## Secular/Hindu Country
## India is a Secular country where all religions are considered equal
## No 57.20
## Yes 42.80
print(castecensus_pl20)
## Opinion on Caste Census
## Can't Say No Yes
## No 69.69 65.54 55.74
## Yes 30.31 34.46 44.26
print(leadership_pl20)
## Nature of Political Leadership
## A democracy in which a strong leader can make decisions without interference from the parliament or the courts.
## No 63.46
## Yes 36.54
## Nature of Political Leadership
## A democracy in which elected representatives decide what becomes law.
## No 58.68
## Yes 41.32
###Model 4a: Preference for RG and Age, Gender (Dummy) and Nature of Locality (Dummy); Among NDA Voters
logit.fit.4a <- glm(y0 ~ s_ + age + female + rural + level_of_education, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.4a)
logit.fit.4c <- glm(y0 ~ farmers + unemployed + casual_labour, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.4c)
##
## Call:
## glm(formula = y0 ~ farmers + unemployed + casual_labour, family = binomial(link = "logit"))
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.0677 -0.9947 -0.9947 1.3205 1.3718
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.44626 0.02275 -19.620 < 2e-16 ***
## farmers 0.12023 0.04221 2.848 0.00439 **
## unemployed 0.18268 0.07830 2.333 0.01965 *
## casual_labour 0.11607 0.05663 2.050 0.04041 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 18368 on 13624 degrees of freedom
## Residual deviance: 18355 on 13621 degrees of freedom
## AIC: 18363
##
## Number of Fisher Scoring iterations: 4
###Model 4d: Preference for RG ~ Assent Rank (Numeric Range; c(1,5)), Ownership of AC (dummy), Ownership of Car (dummy); Among NDA Voters
logit.fit.4d <- glm(y0 ~ asset_cat, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.4d)
##
## Call:
## glm(formula = y0 ~ asset_cat, family = binomial(link = "logit"))
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.035 -1.017 -1.008 1.347 1.357
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.43509 0.04215 -10.32 <2e-16 ***
## asset_cat 0.02290 0.02161 1.06 0.289
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 18368 on 13624 degrees of freedom
## Residual deviance: 18367 on 13623 degrees of freedom
## AIC: 18371
##
## Number of Fisher Scoring iterations: 4
##Model 4e: Preference for RG ~ Political Positions: Secular/Hindu Country (dummy), Leadership (dummy), Caste Census (dummy), Likes Modi (dummy); Among NDA Voters
logit.fit.4e <- glm(y0 ~ secular + strongman + caste_census + bjp_constitution_change + inequality + inc_appeasement, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.4e)
##
## Call:
## glm(formula = y0 ~ secular + strongman + caste_census + bjp_constitution_change +
## inequality + inc_appeasement, family = binomial(link = "logit"))
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.3588 -1.0500 -0.7585 1.1667 2.0221
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -1.47924 0.05664 -26.117 <2e-16 ***
## secular 0.38061 0.04245 8.965 <2e-16 ***
## strongman -0.09923 0.04455 -2.227 0.0259 *
## caste_census 0.34351 0.03973 8.646 <2e-16 ***
## bjp_constitution_change 0.39171 0.03928 9.973 <2e-16 ***
## inequality 0.78029 0.04293 18.175 <2e-16 ***
## inc_appeasement -0.32745 0.03932 -8.329 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 18368 on 13624 degrees of freedom
## Residual deviance: 17556 on 13618 degrees of freedom
## AIC: 17570
##
## Number of Fisher Scoring iterations: 4
###Model 4f: Preference for RG ~ Biggest Failure of BJP: Communalisation, Unemoployment (dummies); Among NDA Voters
logit.fit.4f <- glm(y0 ~ modi_communal_caa + modi_unemployment_corrupt, family=binomial(link = "logit"))
# Summary of the model
summary(logit.fit.4f)
##
## Call:
## glm(formula = y0 ~ modi_communal_caa + modi_unemployment_corrupt,
## family = binomial(link = "logit"))
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.0272 -1.0272 -0.9827 1.3355 1.3987
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -0.47697 0.04487 -10.630 <2e-16 ***
## modi_communal_caa -0.02962 0.07349 -0.403 0.6869
## modi_unemployment_corrupt 0.11274 0.04915 2.294 0.0218 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 18368 on 13624 degrees of freedom
## Residual deviance: 18359 on 13622 degrees of freedom
## AIC: 18365
##
## Number of Fisher Scoring iterations: 4
###Model 4g: Preference for RG ~ All dependent variables; Among NDA Voters
##Summary Model V: Where Congress is a Prominent Player (All Voter, Non-Congress Voters, Non-India Voters and NDA Voters) ###State Reference Value: Gujarat
se_list <- list(rep(NA, length(coef(model_1i))),
rep(NA, length(coef(model_2i))),
rep(NA, length(coef(model_3i))),
rep(NA, length(coef(model_4i))))
predictor_labels <- c("Assam", "Chhattisgarh","Haryana", "Himachal Pradesh", "Karnataka", "Kerala", "Madhya Pradesh", "Maharashtra", "Punjab", "Rajasthan", "Telangana", "Age","Female","Rural", "Level of Education", "Hindu OBC", "Hindu SC", "Hindu ST", "Hindu UC", "Muslims", "Farmers", "Unemployed", "Casual Labour", "Asset Category", "Secular", "Strongman Politics", "Support for Caste Census", "BJP Constitution Change", "High Inequality", "Congress Appeasement Politics", "Modi and Communal Politics", "Modi and Unemployment", "Constant")
# Use stargazer and pass the cleaned names as the custom coefficient labels
summary_model5 <- stargazer(model_1i, model_2i, model_3i, model_4i,
type = "text",
title = "Characterising Potetial Congress Voters", # Table title
dep.var.caption = "Preference for Rahul Gandhi in States where Congress is a Prominent Player",
p.auto = FALSE, # Automatically print p-values
p.style = "asterisk", # Style of p-value stars (* for coefficients only)
column.labels = c("1i All Voters", "2i Non-Congress", "3i Non-India", "4i NDA"),
dep.var.labels = c("Logistic Regression Estimates/Coefficiants"),
single.row = TRUE, # Display statistics in a single row
digits = 2, # Number of digits to display
se = se_list,
covariate.labels = predictor_labels
)
##Summary Model VI: Where Congress is a Marginal Player (All Voter, Non-Congress Voters, Non-India Voters and NDA Voters) ###State Reference Value: Delhi
se_list <- list(rep(NA, length(coef(model_1h))),
rep(NA, length(coef(model_2h))),
rep(NA, length(coef(model_3h))),
rep(NA, length(coef(model_4h))))
predictor_labels <- c("Andhra Pradesh", "Bihar", "Jharkhand", "Odisha", "Tamil Nadu", "Uttar Pradesh", "West Bengal", "Age","Female","Rural", "Level of Education", "Hindu OBC", "Hindu SC", "Hindu ST", "Hindu UC", "Muslims", "Farmers", "Unemployed", "Casual Labour", "Asset Category", "Secular", "Strongman Politics", "Support for Caste Census", "BJP Constitution Change", "High Inequality", "Congress Appeasement Politics", "Modi and Communal Politics", "Modi and Unemployment", "Constant")
# Use stargazer and pass the cleaned names as the custom coefficient labels
summary_model6 <- stargazer(model_1h, model_2h, model_3h, model_4h,
type = "text",
title = "Characterising Potetial Congress Voters", # Table title
dep.var.caption = "Preference for Rahul Gandhi in States where Congress is a Marginal Player",
p.auto = FALSE, # Automatically print p-values
p.style = "asterisk", # Style of p-value stars (* for coefficients only)
column.labels = c("1h All Voters", "2h Non-Congress", "3h Non-India", "4h NDA"),
dep.var.labels = c("Logistic Regression Estimates/Coefficiants"),
single.row = TRUE, # Display statistics in a single row
digits = 2, # Number of digits to display
se = se_list,
covariate.labels = predictor_labels
)
##
## Characterising Potetial Congress Voters
## ==========================================================================================================
## Preference for Rahul Gandhi in States where Congress is a Marginal Player
## ----------------------------------------------------------------------------
## Logistic Regression Estimates/Coefficiants
## 1h All Voters 2h Non-Congress 3h Non-India 4h NDA
## (1) (2) (3) (4)
## ----------------------------------------------------------------------------------------------------------
## Andhra Pradesh 0.86*** 1.15*** 1.68*** 2.58***
## Bihar -0.10 0.05 -0.02 0.65**
## Jharkhand -0.42*** -0.63*** -0.37* 0.29
## Odisha 0.47*** 0.58*** 1.12*** 1.80***
## Tamil Nadu 0.70*** 0.95*** 0.82*** 0.98***
## Uttar Pradesh 0.002 0.17 0.34* 0.92***
## West Bengal -0.74*** -0.50*** -0.19 0.42
## Age 0.0002 -0.001 -0.0002 0.002
## Female 0.06* 0.07* 0.09** 0.12**
## Rural -0.11** -0.11** -0.06 -0.17**
## Level of Education -0.03*** -0.03*** -0.03*** -0.03**
## Hindu OBC -0.15 -0.06 -0.06 -0.37*
## Hindu SC -0.02 0.07 0.15 -0.34
## Hindu ST -0.31** -0.20 -0.07 -0.46**
## Hindu UC -0.39*** -0.28** -0.12 -0.36*
## Muslims 1.41*** 1.38*** 0.91*** 0.02
## Farmers 0.08 0.11** 0.14** 0.18***
## Unemployed 0.10 0.06 0.10 0.17
## Casual Labour 0.10* 0.09 0.15** 0.13
## Asset Category 0.03 0.02 0.04 0.02
## Secular 0.37*** 0.34*** 0.27*** 0.20***
## Strongman Politics -0.13*** -0.12*** -0.09* 0.03
## Support for Caste Census 0.49*** 0.48*** 0.39*** 0.39***
## BJP Constitution Change 0.29*** 0.29*** 0.23*** 0.23***
## High Inequality 0.64*** 0.66*** 0.67*** 0.76***
## Congress Appeasement Politics -0.48*** -0.43*** -0.44*** -0.40***
## Modi and Communal Politics 0.26*** 0.26*** 0.29*** 0.10
## Modi and Unemployment 0.47*** 0.42*** 0.38*** 0.34***
## Constant -0.93*** -1.24*** -1.79*** -2.23***
## ----------------------------------------------------------------------------------------------------------
## Observations 18,521 16,630 11,376 7,663
## Log Likelihood -10,931.53 -9,989.53 -6,916.21 -4,514.02
## Akaike Inf. Crit. 21,921.07 20,037.06 13,890.43 9,086.03
## ==========================================================================================================
## Note: *p<0.1; **p<0.05; ***p<0.01
##
## Characterising Potetial Congress Voters
## ========
## asterisk
## --------
