Criminal Politicians and the constituency
2020-06-29
Document History
Original Publish Date: 29 June, 2020
Updated on: 01:10 PM – 21 October, 2021
In this document we will try to uncover how having criminal running in the elections affect the perception of voters and over all law and order in the constituency.
In the first part we use the 2014 NES data to find the relation between the voters mobilisation and participation, voters perception on increase in malpractices and fair election to the criminal candidacy in a constituency. We denote criminality in vast array of variables. They are as follows:
criminal_candidate -There is at least one criminal candidate in a constituency
serious_criminal_w - A criminal candidate has won that election
serious_1_2 - Either the winner or runner-up are serious criminal
serious_crime_cand - There is atleast one serious criminal candidate in the constituency
serious_crime_w_09 - A serious criminal won the 2009 election
NES
The following excerpt is from the csds website on the nes 14 sample-
For this survey, we selected samples from 26 States (the survey was not conducted in Goa, Nagaland and Sikkim): first we chose 306 of the 543 Lok Sabha constituencies. Within the parliamentary constituencies, 347 Assembly segments were selected, and then a further 1388 individual polling station areas were selected for conducting interviews. From each rural polling station 25 persons were selected from the electoral rolls and from each urban polling station 30 persons were drawn. Around 37,000 voters randomly selected from the most updated electoral rolls were approached for the interview, of which 22,295 voters could be successfully interviewed
The table below summarises the key info on criminal candidates in the 2014 election.
Summary table
| year | non_serious_criminal | serious_criminal | non_serious_winners | non_serious_winners_1plus | serious_winners | serious_winner_1plus |
|---|---|---|---|---|---|---|
| 2009 | 0.12 | 0.10 | 0.25 | 0.23 | 0.16 | 0.12 |
| 2014 | 0.14 | 0.11 | 0.30 | 0.26 | 0.22 | 0.14 |
| 2019 | 0.16 | 0.14 | 0.38 | 0.32 | 0.29 | 0.21 |
DIstribution
rdd checks
## # A tibble: 2 × 2
## serious_criminal n
## <dbl> <int>
## 1 0 13
## 2 1 9Models
participation
The index of participation a variable that combines answers to all sub questions of Q16 that measures the participation of voters in the campaigning process.
| Q16. During the elections people participate in various activities. In the recent elections, out of the activities stated below, in which did you participate? |
+================================================================================================================================================================+ +—————————————————————————————————————————————————————-+
Activities :
a. Attended election meetings/rallies?
b. Participated in processions/nukkad natak etc.?
c. Participated in door to door canvassing?
d. Contributed or collected money?
e. Distributed election leaflets or put up posters?
all
| Dependent variable: | |
| participation | |
| criminal_winner09_in14 | -0.018 |
| (0.044) | |
| serious_1_2 | -0.048 |
| (0.031) | |
| enop | -0.029 |
| (0.029) | |
| reservation | -0.001 |
| (0.025) | |
| State fixed effects | Yes |
| Caste control | Yes |
| Age control | Yes |
| Education control | Yes |
| Asset control | Yes |
| Clustered error PC | Yes |
| Observations | 20,899 |
| R2 | 0.086 |
| Adjusted R2 | 0.085 |
| Residual Std. Error | 0.440 (df = 20865) |
| Note: | p<0.1; p<0.05; p<0.01 |
new models
mobilisation
The index of mobilisation asses the efforts of party and the party workers in mobilising the electorate. It is an index that combines all the answers to the following question.
+—————————————————————————————————————————————————————————————+ +—————————————————————————————————————————————————————————————+
a. Candidate/Party worker/canvasser came to your house to ask your vote .
b. Party/Candidate contacted you or a family member through a phone call or recorded voice or SMS.
c. Party/Candidate offering food/honorarium etc to voters in your locality.
d. Party/Candidate offering to drive voters in your locality to the polling stations .
all
| Dependent variable: | |
| participation | |
| criminal_winner09_in14 | -0.018 |
| (0.044) | |
| serious_1_2 | -0.048 |
| (0.031) | |
| enop | -0.029 |
| (0.029) | |
| reservation | -0.001 |
| (0.025) | |
| female | -0.200*** |
| (0.011) | |
| log(total_assets + 1) | 0.002 |
| (0.006) | |
| State fixed effects | Yes |
| Caste control | Yes |
| Age control | Yes |
| Education control | Yes |
| Asset control | Yes |
| Clustered error PC | Yes |
| Observations | 20,899 |
| R2 | 0.086 |
| Adjusted R2 | 0.085 |
| Residual Std. Error | 0.440 (df = 20865) |
| Note: | p<0.1; p<0.05; p<0.01 |
fair election
Q20: Thinking of the way elections are conducted in India, what do you feel - are elections fair, somewhat fair or unfair?
Labels: value label 1 1: Fair 2 2: Somewhat fair 3 3: Unfair 8 8: No opinion
In the continuous scale, dependent variable is marked as follows : +2 - Fair, +1 - Somewhat fair, 0 - No opinion, -1 - Unfair
fair election - dummy
| Dependent variable: | |
| fair_election_dummy | |
| criminal_winner09_in14 | -0.022 |
| (0.043) | |
| serious_1_2 | -0.022 |
| (0.033) | |
| enop | -0.054* |
| (0.032) | |
| reservation | -0.036 |
| (0.028) | |
| female | -0.019** |
| (0.008) | |
| log(total_assets + 1) | 0.005 |
| (0.007) | |
| State fixed effects | Yes |
| Caste control | Yes |
| Age control | Yes |
| Education control | Yes |
| Asset control | Yes |
| Clustered error PC | Yes |
| Observations | 20,899 |
| R2 | 0.060 |
| Adjusted R2 | 0.058 |
| Residual Std. Error | 0.471 (df = 20863) |
| Note: | p<0.1; p<0.05; p<0.01 |
fair election removing the dummy excluding dk
Fair election continuous variable
malpractice
Q14: Now compare the recently held election in your area with elections held in the past. Do you think in this election things like rigging, intimidation, fraud and other malpractices have increased, decreased or remained the same?
Labels: value label 1 1: Increased 2 2: Same as before 3 3: Decreased 4 4: Malpractices never take place 8 8: Can’t say/Don’t know
Malpractice - binary
| Dependent variable: | |
| malpractise_increase | |
| criminal_winner09_in14 | -0.053*** |
| (0.019) | |
| serious_1_2 | 0.034** |
| (0.015) | |
| enop | 0.009 |
| (0.017) | |
| reservation | 0.030* |
| (0.017) | |
| female | -0.007 |
| (0.005) | |
| log(total_assets + 1) | 0.004 |
| (0.004) | |
| urban | 0.039** |
| (0.017) | |
| State fixed effects | Yes |
| Caste control | Yes |
| Age control | Yes |
| Education control | Yes |
| Asset control | Yes |
| Clustered error PC | Yes |
| Observations | 20,899 |
| R2 | 0.040 |
| Adjusted R2 | 0.039 |
| Residual Std. Error | 0.326 (df = 20862) |
| Note: | p<0.1; p<0.05; p<0.01 |
removing the dk
malpractice contd
In the continuous variable, dependent variable is marked as follows: +2 - Increased, +1 - Same as before, 0 - Can’t say/ Don’t know, , -1 - Decreased, -2 - Malpractices never take place
turnout
All turnout - 2014 only
| Dependent variable: | ||||
| turnout_percentage | ||||
| (1) | (2) | (3) | (4) | |
| criminal_candidate | -0.174 | |||
| (0.469) | ||||
| serious_criminal_w | -0.013 | |||
| (0.350) | ||||
| serious_1_2 | -0.496 | |||
| (0.353) | ||||
| serious_criminal | -0.439 | |||
| (0.371) | ||||
| enop | -1.614*** | -1.616*** | -1.577*** | -1.581*** |
| (0.512) | (0.511) | (0.510) | (0.507) | |
| reservation | 0.667* | 0.677* | 0.640* | 0.640* |
| (0.340) | (0.331) | (0.338) | (0.339) | |
| incumbent_cand | -0.114 | -0.122 | -0.101 | -0.112 |
| (0.374) | (0.386) | (0.378) | (0.377) | |
| log(median_asset) | -0.051 | -0.045 | -0.074 | -0.070 |
| (0.179) | (0.178) | (0.176) | (0.176) | |
| lagged_turnout | 0.728*** | 0.728*** | 0.727*** | 0.727*** |
| (0.049) | (0.050) | (0.049) | (0.049) | |
| lagged_criminal_winner | -0.279 | -0.290 | -0.237 | -0.239 |
| (0.520) | (0.494) | (0.510) | (0.516) | |
| State fixed effects | Yes | Yes | Yes | Yes |
| Clustered error - State | Yes | Yes | Yes | Yes |
| PC level asset control | Yes | Yes | Yes | Yes |
| Observations | 506 | 506 | 506 | 506 |
| R2 | 0.890 | 0.890 | 0.890 | 0.890 |
| Adjusted R2 | 0.884 | 0.884 | 0.884 | 0.884 |
| Residual Std. Error (df = 478) | 3.452 | 3.453 | 3.446 | 3.448 |
| Note: | p<0.1; p<0.05; p<0.01 | |||
female turnout
| Dependent variable: | |
| female_turnout | |
| serious_1_2 | -0.985 |
| (0.847) | |
| enop | -1.186 |
| (0.816) | |
| reservation | |
| incumbent_cand | 0.879 |
| (0.756) | |
| log(median_asset) | 0.178 |
| (0.440) | |
| lagged_turnout | 0.738*** |
| (0.078) | |
| lagged_criminal_winner | 1.038 |
| (1.083) | |
| State fixed effects | Yes |
| Clustered error - State | Yes |
| PC level asset control | Yes |
| Observations | 1,029 |
| R2 | 0.982 |
| Adjusted R2 | 0.956 |
| Residual Std. Error | 7.023 (df = 429) |
| Note: | p<0.1; p<0.05; p<0.01 |
male turnout
| Dependent variable: | |
| male_turnout | |
| serious_1_2 | -0.951** |
| (0.389) | |
| enop | -2.115*** |
| (0.370) | |
| reservation | -0.100 |
| (0.385) | |
| incumbent_cand | 0.571* |
| (0.329) | |
| log(median_asset) | -0.175 |
| (0.181) | |
| lagged_turnout | 0.475*** |
| (0.024) | |
| lagged_criminal_winner | -0.117 |
| (0.446) | |
| State fixed effects | Yes |
| Clustered error - State | Yes |
| PC level asset control | Yes |
| Observations | 1,029 |
| R2 | 0.981 |
| Adjusted R2 | 0.980 |
| Residual Std. Error | 4.892 (df = 971) |
| Note: | p<0.1; p<0.05; p<0.01 |
Spatial trends
Contestants
winners
state level
Margin
Asset
Does the median asset of the candidates increase if there is a serious criminal in the constituency
CSES
- 13963 observations
participation
| Serious crime winner 2014/19 | count_14 | count_19 | Avg. participation - NES | Avg. participation - CSES |
|---|---|---|---|---|
| 0 | 243 | 169 | 0.3 | 0.72 |
| 1 | 43 | 70 | 0.3 | 0.64 |
mobilisation
| Serious crime winner 2014/19 | count_14 | count_19 | Avg. mobilisation - NES | Avg. mobilisation - CSES |
|---|---|---|---|---|
| 0 | 243 | 169 | 0.66 | 0.81 |
| 1 | 43 | 70 | 0.63 | 0.81 |
ACLED
In this second part we use the ACLED data on violence to asses violence during the run upto the 2019 election. For India the data is available from 2015 and we will test our hypothesis on 2019 elections. We are looking at the violence from Jan’19 to May’19 and the criminality measures are the same the above one.
ACLED provides data on 4 kind of incidents - Battles , Explosions/Remote violence Protests, Riots, Strategic developments, Violence against civilians. For our analysis we are using Violence against civilians, Riots, Protests and all others in category called violence others.
Before focusing on the months run upto 2019 election it is essential that we look into the rest of the data to get an idea on the how the distribution looks like and how it is spread out throughout the year.
2019 winner graphs
2014 winner graphs
summary
The following table shows the break-up of those events for the 5 months period of 2019 elections.
| event_type | 2019 |
|---|---|
| Battles | 416 |
| Explosions/Remote violence | 115 |
| Protests | 6330 |
| Riots | 2457 |
| Strategic developments | 135 |
| Violence against civilians | 587 |
| violence | no | yes |
|---|---|---|
| Other violence | 0.46 | 0.59 |
| Protests | 8.24 | 6.63 |
| Riots | 2.87 | 3.34 |
| Violence against civilians | 0.80 | 0.84 |
| event type | no | yes |
|---|---|---|
| Other violence | 0.53 | 0.33 |
| Protests | 8.30 | 5.14 |
| Riots | 3.20 | 2.14 |
| Violence against civilians | 0.85 | 0.65 |
Protests
| Dependent variable: | ||||
| Number of protests in between Jan and May 2019 | ||||
| (1) | (2) | (3) | (4) | |
| serious_crime_w | 0.130 | |||
| (0.140) | ||||
| criminal_candidate | 0.111 | |||
| (0.198) | ||||
| serious_1_2 | 0.043 | |||
| (0.163) | ||||
| serious_criminalyes | 0.130 | |||
| (0.140) | ||||
| enop | 0.473** | 0.462** | 0.463** | 0.473** |
| (0.221) | (0.221) | (0.222) | (0.221) | |
| incumbent_cand | -0.231* | -0.236* | -0.234* | -0.231* |
| (0.135) | (0.135) | (0.135) | (0.135) | |
| log(median_asset) | 0.035 | 0.035 | 0.035 | 0.035 |
| (0.072) | (0.072) | (0.072) | (0.072) | |
| lagged_criminal_winner | 0.024 | 0.035 | 0.040 | 0.024 |
| (0.164) | (0.164) | (0.163) | (0.164) | |
| n_events_month_lag | 1.027*** | 1.026*** | 1.027*** | 1.027*** |
| (0.025) | (0.025) | (0.025) | (0.025) | |
| Year fixed effects | Yes | Yes | Yes | Yes |
| Observations | 480 | 480 | 480 | 480 |
| R2 | 0.879 | 0.878 | 0.878 | 0.879 |
| Adjusted R2 | 0.866 | 0.866 | 0.865 | 0.866 |
| Residual Std. Error (df = 433) | 1.288 | 1.289 | 1.289 | 1.288 |
| Note: | p<0.1; p<0.05; p<0.01 | |||
Violence against civilians
| Dependent variable: | ||||
| Number of Violence against civilians happened between Jan and May 2019 | ||||
| (1) | (2) | (3) | (4) | |
| serious_crime_w | 0.023 | |||
| (0.025) | ||||
| criminal_candidate | 0.048 | |||
| (0.035) | ||||
| serious_1_2 | 0.014 | |||
| (0.029) | ||||
| serious_criminalyes | 0.023 | |||
| (0.025) | ||||
| enop | -0.010 | -0.013 | -0.013 | -0.010 |
| (0.040) | (0.040) | (0.040) | (0.040) | |
| incumbent_cand | -0.015 | -0.017 | -0.016 | -0.015 |
| (0.024) | (0.024) | (0.024) | (0.024) | |
| log(median_asset) | -0.010 | -0.010 | -0.010 | -0.010 |
| (0.013) | (0.013) | (0.013) | (0.013) | |
| lagged_criminal_winner | -0.016 | -0.015 | -0.013 | -0.016 |
| (0.029) | (0.029) | (0.029) | (0.029) | |
| n_events_month_lag | 0.403*** | 0.399*** | 0.404*** | 0.403*** |
| (0.044) | (0.044) | (0.044) | (0.044) | |
| Year fixed effects | Yes | Yes | Yes | Yes |
| Observations | 480 | 480 | 480 | 480 |
| R2 | 0.452 | 0.454 | 0.452 | 0.452 |
| Adjusted R2 | 0.394 | 0.396 | 0.393 | 0.394 |
| Residual Std. Error (df = 433) | 0.231 | 0.230 | 0.231 | 0.231 |
| Note: | p<0.1; p<0.05; p<0.01 | |||
Riots
| Dependent variable: | ||||
| Average number of riots happened in every month during Jan and May 2019 | ||||
| (1) | (2) | (3) | (4) | |
| serious_crime_w | 0.071 | |||
| (0.087) | ||||
| criminal_candidate | 0.091 | |||
| (0.124) | ||||
| serious_1_2 | 0.157 | |||
| (0.102) | ||||
| serious_criminalyes | 0.071 | |||
| (0.087) | ||||
| enop | -0.216 | -0.223 | -0.233* | -0.216 |
| (0.139) | (0.139) | (0.139) | (0.139) | |
| incumbent_cand | -0.074 | -0.078 | -0.082 | -0.074 |
| (0.084) | (0.085) | (0.084) | (0.084) | |
| log(median_asset) | 0.075* | 0.075* | 0.075* | 0.075* |
| (0.045) | (0.045) | (0.045) | (0.045) | |
| lagged_criminal_winner | 0.010 | 0.014 | 0.012 | 0.010 |
| (0.103) | (0.102) | (0.102) | (0.103) | |
| n_events_month_lag | 0.740*** | 0.737*** | 0.738*** | 0.740*** |
| (0.065) | (0.065) | (0.065) | (0.065) | |
| Year fixed effects | Yes | Yes | Yes | Yes |
| Observations | 480 | 480 | 480 | 480 |
| R2 | 0.530 | 0.530 | 0.532 | 0.530 |
| Adjusted R2 | 0.481 | 0.480 | 0.483 | 0.481 |
| Residual Std. Error (df = 433) | 0.805 | 0.805 | 0.804 | 0.805 |
| Note: | p<0.1; p<0.05; p<0.01 | |||
all violence
##
## <table style="text-align:center"><tr><td colspan="5" style="border-bottom: 1px solid black"></td></tr><tr><td style="text-align:left"></td><td colspan="4"><em>Dependent variable:</em></td></tr>
## <tr><td></td><td colspan="4" style="border-bottom: 1px solid black"></td></tr>
## <tr><td style="text-align:left"></td><td colspan="4">Average number of violence happened in every month during Jan and May 2019</td></tr>
## <tr><td style="text-align:left"></td><td>(1)</td><td>(2)</td><td>(3)</td><td>(4)</td></tr>
## <tr><td colspan="5" style="border-bottom: 1px solid black"></td></tr><tr><td style="text-align:left">serious_crime_w</td><td>0.058</td><td></td><td></td><td></td></tr>
## <tr><td style="text-align:left"></td><td>(0.050)</td><td></td><td></td><td></td></tr>
## <tr><td style="text-align:left"></td><td></td><td></td><td></td><td></td></tr>
## <tr><td style="text-align:left">criminal_candidate</td><td></td><td>-0.012</td><td></td><td></td></tr>
## <tr><td style="text-align:left"></td><td></td><td>(0.070)</td><td></td><td></td></tr>
## <tr><td style="text-align:left"></td><td></td><td></td><td></td><td></td></tr>
## <tr><td style="text-align:left">serious_1_2</td><td></td><td></td><td>0.050</td><td></td></tr>
## <tr><td style="text-align:left"></td><td></td><td></td><td>(0.058)</td><td></td></tr>
## <tr><td style="text-align:left"></td><td></td><td></td><td></td><td></td></tr>
## <tr><td style="text-align:left">serious_criminalyes</td><td></td><td></td><td></td><td>0.058</td></tr>
## <tr><td style="text-align:left"></td><td></td><td></td><td></td><td>(0.050)</td></tr>
## <tr><td style="text-align:left"></td><td></td><td></td><td></td><td></td></tr>
## <tr><td style="text-align:left">enop</td><td>-0.095</td><td>-0.097</td><td>-0.102</td><td>-0.095</td></tr>
## <tr><td style="text-align:left"></td><td>(0.078)</td><td>(0.078)</td><td>(0.078)</td><td>(0.078)</td></tr>
## <tr><td style="text-align:left"></td><td></td><td></td><td></td><td></td></tr>
## <tr><td style="text-align:left">incumbent_cand</td><td>0.030</td><td>0.030</td><td>0.027</td><td>0.030</td></tr>
## <tr><td style="text-align:left"></td><td>(0.048)</td><td>(0.048)</td><td>(0.048)</td><td>(0.048)</td></tr>
## <tr><td style="text-align:left"></td><td></td><td></td><td></td><td></td></tr>
## <tr><td style="text-align:left">log(median_asset)</td><td>0.045<sup>*</sup></td><td>0.045<sup>*</sup></td><td>0.045<sup>*</sup></td><td>0.045<sup>*</sup></td></tr>
## <tr><td style="text-align:left"></td><td>(0.025)</td><td>(0.026)</td><td>(0.026)</td><td>(0.025)</td></tr>
## <tr><td style="text-align:left"></td><td></td><td></td><td></td><td></td></tr>
## <tr><td style="text-align:left">lagged_criminal_winner</td><td>0.055</td><td>0.064</td><td>0.061</td><td>0.055</td></tr>
## <tr><td style="text-align:left"></td><td>(0.058)</td><td>(0.058)</td><td>(0.058)</td><td>(0.058)</td></tr>
## <tr><td style="text-align:left"></td><td></td><td></td><td></td><td></td></tr>
## <tr><td style="text-align:left">n_events_month_lag</td><td>0.552<sup>***</sup></td><td>0.555<sup>***</sup></td><td>0.553<sup>***</sup></td><td>0.552<sup>***</sup></td></tr>
## <tr><td style="text-align:left"></td><td>(0.032)</td><td>(0.032)</td><td>(0.032)</td><td>(0.032)</td></tr>
## <tr><td style="text-align:left"></td><td></td><td></td><td></td><td></td></tr>
## <tr><td colspan="5" style="border-bottom: 1px solid black"></td></tr><tr><td style="text-align:left">Year fixed effects</td><td>Yes</td><td>Yes</td><td>Yes</td><td>Yes</td></tr>
## <tr><td style="text-align:left">Observations</td><td>480</td><td>480</td><td>480</td><td>480</td></tr>
## <tr><td style="text-align:left">R<sup>2</sup></td><td>0.572</td><td>0.571</td><td>0.572</td><td>0.572</td></tr>
## <tr><td style="text-align:left">Adjusted R<sup>2</sup></td><td>0.527</td><td>0.525</td><td>0.526</td><td>0.527</td></tr>
## <tr><td style="text-align:left">Residual Std. Error (df = 433)</td><td>0.456</td><td>0.457</td><td>0.457</td><td>0.456</td></tr>
## <tr><td colspan="5" style="border-bottom: 1px solid black"></td></tr><tr><td style="text-align:left"><em>Note:</em></td><td colspan="4" style="text-align:right"><sup>*</sup>p<0.1; <sup>**</sup>p<0.05; <sup>***</sup>p<0.01</td></tr>
## </table>ADR - Daksh
note: ADR Daksh data : CSES data
why did you vote for a criminal candidate
Data Frame Summary
daksh_crime
Dimensions: 238692 x 6
Duplicates: 238012
| No | Variable | Stats / Values | Freqs (% of Valid) | Valid | Missing |
|---|---|---|---|---|---|
| 1 | candidate is of similar caste or religion [character] |
1. (Empty string) 2. No 3. Yes |
9741 ( 4.1%) 140785 (59.0%) 88166 (36.9%) |
238692 (100.0%) |
0 (0.0%) |
| 2 | candidate is powerful [character] |
1. (Empty string) 2. No 3. Yes |
9072 ( 3.8%) 138307 (57.9%) 91313 (38.3%) |
238692 (100.0%) |
0 (0.0%) |
| 3 | candidate otherwise does good work [character] |
1. (Empty string) 2. No 3. Yes |
9487 ( 4.0%) 74529 (31.2%) 154676 (64.8%) |
238692 (100.0%) |
0 (0.0%) |
| 4 | cases against him are not serious [character] |
1. (Empty string) 2. No 3. Yes |
11094 ( 4.6%) 124018 (52.0%) 103580 (43.4%) |
238692 (100.0%) |
0 (0.0%) |
| 5 | candidate has spent generously in elections [character] |
1. (Empty string) 2. No 3. Yes |
8939 ( 3.7%) 139421 (58.4%) 90332 (37.8%) |
238692 (100.0%) |
0 (0.0%) |
| 6 | voters don’t know about the criminal record [character] |
1. (Empty string) 2. No 3. Yes |
9773 ( 4.1%) 128873 (54.0%) 100046 (41.9%) |
238692 (100.0%) |
0 (0.0%) |