Professional Documents
Culture Documents
BEHAVIOR
Factors and biases
affecting the voting
behavior of people
Group 5
Gursartaj Singh Nijjar (1311091)
Lavanya (1311097)
Raj Kumar (1311111)
Shrikant Vijayrao Nikade (1311123)
02ndDecember 2013
Table of Contents
Introduction ................................................................................................................................ 3
Objective .................................................................................................................................... 3
Scope and Methodology ............................................................................................................ 3
Factors affecting Voting behaviour ........................................................................................... 4
Analysis...................................................................................................................................... 5
Demographics of respondents ................................................................................................ 5
Hypothesis.................................................................................................................................. 6
1. Analysis of factors influencing voter preference for Narendra Modi ............................. 6
2. Analysis of factors influencing Vote preference for Rahul Gandhi ................................ 7
3. Analysis of voters behaviour of sharing political views on social media ...................... 8
4. Analysis of voter behaviour depending on its profession and family income................. 9
5. Experiment checking offering none of the above option ............................................ 11
Limitations of study ................................................................................................................. 13
Conclusion ............................................................................................................................... 13
Appendix .................................................................................................................................. 14
Survey Voting behaviour - General Election ........................................................................... 16
Introduction
General election is the biggest political event in India. Being the largest democracy makes it
even more important. This makes voting behaviour an important area of research in field of
political science and psychology. Studying voting behaviour can explain why certain
decisions are made by voters and why political parties adopt certain ways to attract voters. To
carry out the study and draw inferences about behaviour concerned to voting decision, certain
factors such as gender, religion, region, culture etc need to be considered. Apart from these,
key influencers such as emotion, political socialization, media, and diverse political views
play major roles. The study focuses mainly on establishing a relation between different
factors discussed above and the voting pattern of individuals. We are also trying to find out
certain biases such as status quo through a separate experiment. The results would be very
helpful for general public as well as political parties. The study may help voters to take
informed decision by taking into account the biases and effect of different factors. Political
parties can design and define their strategies accordingly to have anticipated impact on the
public and hence get desired result.
Objective
1. To study what factors influence the voting behaviour of people
2. To analyse how social media influence political behaviour of people
3. To study the impact of offering none of the option to voters when the candidates are
not strong or deviate from party ideology
Candidate
1. Political Party- Some people have preferences for certain political party. The voting
decision for them depends mostly on the party to which the candidate belongs to and
nothing else.
2. Candidates past work/ experience- Some people value candidates past experience
more than anything else. They analyse the past work of the candidate before voting.
3. Personality of the candidate- The overall personality of the candidate is also an
important factor for voters. For them, it determines his eligibility and his decision
making powers.
4. General perception about the candidate and image on social media- The general
perception about the candidate does influence the voting decisions of certain people.
This perception is sometimes created by social media. Some leaders are always talked
about positively in social media and they maintain a very clean image through this.
Candidates also try to connect with the voters frequently on social media and project
their good work continuously. Some candidates who fail to do so or are projected in
bad light on various social mediums create a bad image for themselves. This can play
as a catalyst and influence vast majority of people very quickly.
Analysis
The objectives of this study have been achieved by using the primary data collected through
the survey conducted. The right statistical tools such as regression analysis, F-test and t-stat
test have been used to prove or disprove the hypothesis set so as to reach the objectives.
Demographics of respondents
Age
Gender
69%
18-25
21%
24%
31%
25-35
29%
76%
Male
Female
0-5
5-10
36%
10-20
> 20
Profession
4%
9%
6%
7%
74%
Student
Self-employed
Salaried Non-Govt
Govt
Other
Religion
5%
1%
7% 5%
82%
Hindu
Muslim
Sikh
Christian
Others
Hypothesis
1) Factors influencing voter preference for Narendra Modi
2) Factors influencing Vote preference for Rahul Gandhi
3) Factors affecting Voters behaviour of sharing political views on social media
4) Factors affecting voter behaviour depending on its profession and family income
5) Effect of None of the above option
Hypothesis
Party, past performance, personality, clean image, anti-incumbency affect the voting decision
of people while deciding their vote for Narendra Modi.
Model (Only includes significant variables)
Modi_vote = 0.3559 + 0.3544 * Imp_Party + 0.5646 * Imp_Pastperf
Dependent variable
Modi_vote: votes in favour of Narendra Modi
Independent variable
From the regression analysis (included in appendix) it can be deducted that out of all the
factors, only party and past performance of Narendra Modi significantly drive the voters in
his favour.
From the regression analysis, we find that only the clean image and the antiincumbancy
factor against congress are the only significant factors which influences voters decision to
vote for Rahul Gandhi. The coefficient for antiincumbancy factor is negative, which shows
that though the factor is significant, as the factor becomes more negative, the voters
inclination to vote for Rahul Gandhi decreases.
Hypothesis
Congress supporters share articles, which portray positive image of Rahul Gandhi.
To test this we performed a regression keeping the 5 articles as dependent variables and votes
to Rahul Gandhi as the independent variable.
From the regression analysis, the p value is significant only in case of article 1 only. Article 1
was against Narendra Modi, which is seen in the positive value of its coefficient. But, articles
2,3,4 and 5 are not significant.
speeches.However, recent media reports speculate that NaMo is losing steam building up to
the elections as the crowds at his recent rallies are dwindling1.
Traditionally the upper income groups show variation in voting patterns when compared to
weaker sections of the society, especially in times of growing inflation and slowdown of
economy. The urban-high income brackets may support industrialization and the Gujarat
model. The identification of this trend can help political party address issues relevant to the
target group & garner support.
Hypothesis
a) Ho: Narendra Modi is equally popular among the youth/students and professionals.
Result:
p1= The proportion of respondents among students in favour of Modi.
p2= The proportion of respondents among professionals in favour of Modi.
n1= The number of students.
n2= The number of professionals.
x1= Number ofrespondents among students in favour of Modi.
x2= Number of respondents among professionals in favour of Modi.
Survey 1
Survey 2
) (
X
36
7
N
58
20
P
0.62
0.35
P cap=0.55
Z=2.0988
We conduct a two-tailed test for the above values. This value corresponds to a significance
level of 2% and shows that we can conclude with 98% confidence that being a professional
impacts the choice of candidate.
b) Narendra Modi is more popular among the high family income group (>5 lakh) than
in low income group (<5 lakh)
Result:
1
http://indiatoday.intoday.in/story/bjp-pm-candidate-narendra-modi-agra-rally-mulayam-singhyadav/1/325877.html
10
p1= The proportion of respondents among high income group in favour of Modi.
p2= The proportion of respondents among low income group in favour of Modi.
n1= The number of people in high income group.
n2= The number of people in low income group.
x1= Number of respondents in high income groupin favour of Modi.
x2= Number of respondents in low income group in favour of Modi.
Survey 1
Survey 2
) (
X
25
20
n
54
34
p
0.568
0.588
P cap=0.5769
Z= 0.177
For a two-tailed test this value suggests very low confidence interval ~12% for rejecting the
null hypothesis.
Thus we conclude there is no difference between high and low income group in terms of
popularity of Modi.
http://articles.timesofindia.indiatimes.com/2004-04-17/mumbai/28346695_1_candidates-face-criminalcharges-lok-sabha-candidates-democratic-reforms-adr
11
have an option to reject all candidates by pressing NOTA on EVM machine with NOTA
option provided after the last candidates name in the ballot3.
The study is to assess the impact of providing explicit information and publicity about the
option to voters prior to voting and providing a None of the below option instead. A strong
influence will make it mandatory for parties to contest elections with candidates of strong
credentials.
Hypothesis
The voters irrespective of party affiliation will exercise the none of the option in the same
proportion irrespective of whether the option is provided at the first or at the last to the voter.
Result
p1= The proportion of respondents of survey 1 who voted for None of the above option.
P2= The proportion of respondents of survey 2 who voted for None of the below option.
n1= The proportion of respondents of survey 1.
n2= The proportion of respondents of survey 2.
x1= Number of respondents of survey 1 who voted for None of the above option.
x2= Number of respondents of survey 2 who voted for None of the below option.
) (
)
X
Survey 1
18
44
0.409
Survey 2
31
54
0.574
http://www.ndtv.com/elections/article/assembly-polls/none-of-the-above-option-in-pink-in-assembly-pollswhite-in-parliamentary-440634
12
P=0.5
Z=1.584
This value corresponds to significance level of 12% for a two-tailed test for a normal
distribution curve. This implies at 88% confidence interval the null hypothesis (Ho) can be
rejected.
With 88% confidence we can say that the voters are more averse to give vote to a candidate
with dubious records when they are offered a None of the below option and well-informed
about the option in advance.
Limitations of study
Limited number of survey respondents: we have taken sample size of 78 people on the
basis of survey responses received.
Reliability of self-reported data: in the study the results are totally relied upon the self
reported data provided by respondents.
In the study, we have limited our focus area to Narendra Modi (BJP candidate) and
Rahul Gandhi (Congress candidate)
Conclusion
We set out to find out whether any trends exist in the voting behaviour exhibited by people.
There are several hypotheses that one can come up with intuitively but we also wanted to test
our intuition mathematically. Based on our hypotheses, we tested the data and came up with
some interesting findings. As expected perception matters. It was Narendra Modis party and
his past performance which influenced people to vote for him, whereas it was Rahul Gandhis
clean image which made him a favourite.
We also conducted an experiment which seemed to show that more people thought they had
to make a choice between two candidates, when none of the above was also an option.
However, they chose to exercise their veto power when none of the below was listed first.
13
Appendix
Hypothesis
1) Factors influencing voter preference for Narendra Modi:
Regression Statistics
0.42003
0.17643
0.11923
1.36669
78
R
R Square
Adjusted R Square
Standard Error
Total Number Of Cases
ANOVA
d.f.
SS
5.
72.
77.
Regression
Residual
Total
Coefficients
-0.087
0.18013
-0.18309
0.18598
0.23165
0.18446
Imp_Pastperf (1-5)
Imp_Pers (1-5)
2)
Standard
Error
1.54042
0.135
0.18273
Imp_Party (1-5)
Antiincumbency
28.80971
134.48516
163.29487
0.35592
0.35437
0.56459
Intercept
Imp_CleanImg (1-5)
MS
5.76194
1.86785
LCL
2.21088
0.12942
0.2601
0.38716
0.56908
-0.1214
3.0848
p-level
0.01404
UCL
t Stat
2.92272
0.57933
0.86907
0.23105
2.62491
3.08972
0.48299
0.79036
1.00819
0.21315
0.20291
0.49335
Regression Statistics
R
R Square
Adjusted R Square
Standard Error
Total Number Of
Cases
0.37758
0.14257
0.08302
1.09693
78
ANOVA
d.f.
Regression
Residual
5.
72.
SS
14.40457
86.6339
MS
F
p-level
2.88091 2.39428 0.04574
1.20325
14
p-level
0.81793
0.01058
0.00285
0.63057
0.43192
0.31674
Total
Intercept
Imp_Party (1-5)
Imp_Pastperf
(1-5)
Imp_Pers (1-5)
Imp_CleanImg
(1-5)
77.
101.03846
Coefficients
2.67655
0.00572
Standard
Error
1.23637
0.10836
0.11097
0.14666
-0.14613
0.14458
0.37455
0.18593
-0.41001
0.14805
Antiincumbency
T (10%)
1.66629
LCL - Lower value of a reliable interval (LCL)
UCL - Upper value of a reliable interval (UCL)
LCL
UCL
t Stat p-level
0.6164 4.73669 2.16485 0.03371
-0.17484 0.18627 0.05278 0.95806
-0.13341 0.35535 0.75665 0.45173
-0.38704 0.09477 1.01077 0.31551
0.06474 0.68435 2.01449 0.04769
-0.65671 0.16331 2.76932 0.00714
0.37036
0.13717
0.07725
1.39889
78
ANOVA
d.f.
Regression
5.
SS
22.39916
Standard
Coefficients
Error
LCL
2.582
0.49606
1.75542
Intercept
-0.29147
0.1592
-0.55675
Share1
0.20447
0.13267
-0.01661
Share2
0.45172
0.1963
0.12463
Share3
0.11549
0.17338
-0.17341
Share4
-0.13643
0.16227
-0.40683
Share5
T (10%)
1.66629
LCL - Lower value of a reliable interval (LCL)
UCL - Upper value of a reliable interval (UCL)
Residual
72.
140.89571
Total
77.
163.29487
MS
4.47983
UCL
3.40859
-0.0262
0.42554
0.77881
0.4044
0.13396
F
2.28927
t Stat
5.20499
-1.83086
1.54111
2.30119
0.66612
-0.84077
1.95688
Congress supporters share articles which portray positive image of Rahul Gandhi
15
p-level
0.
0.07126
0.12767
0.02428
0.50746
0.40326
Regression Statistics
R
R Square
Adjusted R Square
Standard Error
Total Number Of Cases
0.30229
0.09138
0.02828
1.1292
78
ANOVA
d.f.
Regression
Residual
Total
5.
72.
77.
Coefficients
SS
9.23251
91.80595
101.03846
MS
1.8465
1.27508
Standard
Error
LCL
2.88415
0.34642
Intercept
0.30159
0.13821
Share1
-0.05781
0.1153
Share2
-0.16289
0.14955
Share3
-0.11554
0.1604
Share4
-0.07864
0.13875
Share5
T (10%)
1.66629
LCL - Lower value of a reliable interval (LCL)
UCL - Upper value of a reliable interval (UCL)
2.30691
0.0713
-0.24993
-0.41209
-0.38281
-0.30983
Survey Form
18-25
25-35
35-50
> 50
Gender *
Male
Female
16
F
1.44814
p-level
0.21762
UCL
t Stat
3.46139
0.53189
0.13431
0.08631
0.15173
0.15255
8.32553
2.18217
-0.5014
-1.08915
-0.72032
-0.56679
p-level
3.78275E12
0.03236
0.61762
0.27972
0.47366
0.57262
Religion *
o
Hindu
Muslim
Sikh
Christian
Other
Profession *
Self-employed
Salaried Non-Government
Government
Student
Other
Family Income (in LPA) *
0-5
5-10
10-20
> 20
Which state were you brought up in? *
Are you influenced with the opinion of your friends in making decisions? *
Strongly disagree
Strongly agree
Yes
No
How likely are you to vote in the coming elections? *
1
Very unlikely
Very likely
17
Rate the following based on its importance while voting for a Member of Parliament nominee in
your constituency? *
1 - Least
important
Very unlikely
Very likely
1
Vey unlikely
Very likely
BJP
Congress
o
o
Others
Do you feel Gujarat Model is right for the country AND should be replicated? *
Yes
No
Do you feel a strong third alternative is required in the 2014 General Elections? *
18
5 - Most
important
Yes
No
Can't say
Which party you think should lead the Central Government? *
BJP
Congress
Other
According to you, which candidate do you feel is ideal to be the Prime Minister? *
Rahul Gandhi
Narendra Modi
Other
How many hours on an average do you spend on internet daily? *
0-2
2-5
5-10
o
o
> 10
Are you follower of any political party/ leader on social media? *
Yes
No
How likely are you to share the twitter feed, facebook post or article about your favorite
leader? *
Very unlikely
Very likely
Strongly disagree
Strongly agree
Do you feel social media plays an important role in influencing the voting behaviour of people? *
19
Strongly disagree
Strongly agree
Strongly disagree
Strongly agree
How willing are you to share the following posts on social media? *
Very
likely
Likely
20
Undecided
Unlikely
Very
unlikely