+ All Categories
Home > Documents > OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote....

OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote....

Date post: 02-Aug-2020
Category:
Upload: others
View: 0 times
Download: 0 times
Share this document with a friend
19
Revolutionizing Opinion Polling in India with path breaking new systems and presenting the entire process in the public domain, including methodology and raw data OSOP: Open Source Opinion Poll of Karnataka Dr Praveen Patil
Transcript
Page 1: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

Revolutionizing Opinion Polling in India with path breaking new

systems and presenting the entire process in the public

domain, including methodology and raw data

OSOP: Open Source

Opinion Poll of

Karnataka

Dr Praveen Patil

Page 2: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

5Fort3 – Zealously Neutral Election Analysis Page 1

Index

1) Index ……………………………………………………………………………………………………... Page 1 2) Methodology ………………………………………………………………………………………….. Page 2 3) Target Sample Size …………………………………………………………………………………. Page 2 4) Work Plan ………………………………………………………………………………………………. Page 3 5) Selection Process ……………………………………………………………………………………. Page 3 6) VWISM …………………………………………………………………………………………………… Page 4 7) Constituency Selection Criteria (SSI & RSSI) …………………………………………….. Page 5 8) Mixed Mode Methodology ……………………………………………………………………….. Page 8 9) Questionnaire System ……………………………………………………………………………… Page 9 10) Error Management Cycle ………………………………………………………………………… Page 9 11) Raw Data ……………………………………………………………………………………………… Page 10 12) Raw Data (2014 Vote, Tele Survey) ……………………………………………………….. Page 11 13) Raw Data (PM preference, Tele Survey) ………………………………………………… Page 14 14) GPL ………………………………………………………………………………………………………. Page 17 15) Notes ……………………………………………………………………………………………………. Page 18

Page 3: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

5Fort3 – Zealously Neutral Election Analysis Page 2

For the first time in India, if not the whole world, we are attempting an open source model of opinion polling. Poll surveys in India are almost an occult craft as most of them employ black-box techniques which do not withstand any public scrutiny and are simply accepted on the basis of the “reputation” of the pollster. This ad-hoc approach to election analysis has created many armchair psephologists who give out seat projections at the drop of the hat, especially in the era of social media when even vote-shares are projected by Tweeple every day with terrifying confidence. In a market oversaturated with dubious pollsters, 5Forty3 has decided to restore the “Bharosa” in polling by creating an open source model wherein not only methodologies and techniques are put into the public domain, but also raw data is shared openly for the first time!

Methodology

This Karnataka poll is a mixed method, mixed level survey – using both modified CATI & Face-to-face methodology and gives us two independent results (individual MP seat projections for all 28 MPs and overall projection based on total sample size).

Survey Targets Target districts 26 Target MP constituencies 25 Target Assembly segments 75 Target Polling Booths (not less than) 300 Target polling booths/assembly segment 3-5 Target respondents/assembly segment 160 Target sample size 12000+

This classification is based on the following criteria;

1. The 26 districts and 25 MP constituencies have been chosen keeping in mind the maximum geographic spread, caste-ethnic diversity, political affiliations and past electoral data. For instance, Mysore and Chamrajnagar are 2 districts and MP constituencies but have very similar data patterns, so only Mysore is part of our survey and the same data can then be extrapolated to Chamrajanagar.

2. Assembly segments have been chosen based on primarily the swing factor of the last 4 LS polls and last 4 assembly elections (with our own independent black-box mathematical model, explained later)

3. Polling Booths have been chosen primarily based on caste-religion matrix as per our database of Karnataka and secondarily by geography

4. Targeted sample size per polling booth is a broad derivative of our mathematical model calculated on the basis of religion-caste-swing-vote matrix by giving adequate weightage to each criteria and multiplying by 4 for swing vote criteria so as to remove whipsaws and then adding 1.

Social status of targeted sample size

Social Status Target Percentage Urban 33% Rural 67% Men 60% Women 40% Lingayat (Veerashaiva) 21% (Overall, but with sub-regional differences)

Page 4: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

5Fort3 – Zealously Neutral Election Analysis Page 3

Vokkaliga 16% (Overall, but with sub-regional differences) SC (right-wing) 12% (Overall, but with sub-regional differences) SC (Left-wing) 9% (Overall, but with sub-regional differences) Kuruba 8% (Overall, but with sub-regional differences) Muslim 14% (Overall, but with sub-regional differences) Random/others 20% (Overall, but with sub-regional differences)

Based on our own system of identifying ethnic groups through past field studies of the state

Men are given more weightage because male voters tend to make decision in Karnataka unlike in heartland where women have more independent choices

Urban-rural divide is finely classified based on non-agrarian activity in small towns. For instance, even a district centre like Yadgir is classified as “rural” because of a largely agrarian activity, whereas a small town like Hospet is classified as urban because of its urban milieu of tourism industry and mining industry.

Different caste-ethnic groups representation targets in the sample are chosen not just based on overall census data but by our own methodology (this method; explained later; helps in avoiding adding weightage post-survey, substantially)

Sub-regional weightages are different for different regions – For instance; Lingayats are given only 9% weightage in Old-Mysore-Mandya region, but 39% in Bombay-Karnataka region.

Since targeted vote-base is culled from electoral roles, economic criteria of the voters has not been part of our target demographic, but has been analyzed post-survey, based on responses

Work Plan

Questionnaires for data collection have been prepared based on three criteria – preferences, opinions and factual information. All questionnaires are in 6 languages – Kannada, Hindi, Urdu, Telugu, Tamil and English. Our target was to achieve at least 60% face-to-face responses and 40% telephone responses on our overall sample size (telephone database has been obtained independently by a third party for our targeted sample based on electoral rolls).

37 individual data collection officers spread out across different districts from the 11th of January, covering different targets of sample sizes under the aegis of three different organizations. We completed the exercise by the 22nd of January and finished data collection and number crunching by the 26th of January.

Budgetary allocation

Our rough target was to get face-to-face responses at 55Rs/respondent and telephonic responses at 10Rs/respondent; actually achieved – 65Rs/respondent (face to face) and 18Rs/respondent (telephonic)

Median pay per node per day = 400Rs (exclusive of TA/DA) Median TA/DA per node per day = 600Rs Telephone database = 27000Rs (with culled data as per electoral roles) In all, 3 volunteer telecallers and 4 paid telecallers worked on this project for 6-9 hours every day

from 10th Jan to 21st Jan 2014 for the modified CATI Sponsorship funds allotted = 4 Lakhs; actual expenditure = 432000Rs

Selection process

Page 5: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

5Fort3 – Zealously Neutral Election Analysis Page 4

The fundamental idea of our mixed-mode sampling was to target rural voters by direct face to face interviews and urban voters through telephonic interviews, although there has been some overlapping. The second most important reason was to reduce costs, as face to face interviews are very expensive to carry out. For face to face interviews, our field workers travelled to 102 villages, 59 small towns and 21 cities to cover 204 polling booths of all the 75 targeted assembly segments, while our telephonic survey covered a total of 132 polling booths.

One of the most important factors that decides the robustness of a poll survey is the selection process of the targeted population and this was the area that we concentrated on for more than a fortnight before the actual data collection process began. After considerable alignment of census data with our own database we arrived at percentage weightage to be given to each of the ethnic groups.

This is a new sampling methodology developed by Dr. Praveen Patil after years of field studies in electoral zones. Most pollsters simply use census as the sampling frame and use a percentage derivative for their sample sizes of different caste-ethnic groups, but we are using a revolutionary new technique

known as VWISM – Voter Weightage Index Sampling Methodology – which is a far more dynamic measure of different caste-ethnic groups because it not only takes actual population percentages into consideration, but also of weightage to their vote in terms of turnout ability, influence in the community etc. A simple example is that of the Muslim Vote, which is roughly 11% in Karnataka in terms of population, but because of consistently higher turnouts in the past, it gets a higher weightage in the larger vote-pattern scheme.

A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects) in the state, out of a possible 4Cr 36 lakh total voters. This gives the Lingayat’s roughly 15% weightage in the vote-matrix, but in reality the weightage is higher than that because of 3 following factors;

Lingayats being the dominant caste (in north & central Karnataka mostly) also carry along other small castes like Jangams, Valikars, a section of Kurubas etc. in villages and towns in an electoral domino effect

Lingayats mostly vote en-block, which gives them advantage over other smaller groups like say Kamhars or Kolis etc who are much more fragmented along localized conditions

The Lingayat-Brahmin combo, famously known as LIBRA (propagated and nurtured by former CM, Ramkrishna Hegde) works in tandem in most of Mumbai-Karnataka and Hyderabad-Karnataka districts, which gives this vote an added edge of roughly 6-9 lakh more votes from the pie of 15-18 lakh Brahmin vote.

Thus for all practical purposes, Lingayat’s enjoy a weightage of 21% to 23% in the overall electoral scheme of things. Thus any sample which has less than 20 Lingayat respondents out of 100 would more often than not produce wrong results. Similar permutations using VWISM are done for different ethnic groups to arrive at our weightage formula for the sample survey in Karnataka.

We have also taken adequate care to maintain sub-regional ethnic differential factors using a method

known as RCW (Regional Caste Weightage) For instance, there are roughly 48-51 lakh Vokkaliga voters in the state, but most are concentrated in the southern part of the state and especially more so in the Old-Mysore region. Again, despite having an actual presence of only about 11-12% overall, the weightage given for Vokkaligas is between 24% to 27% in Old Mysore region because they have domino effect on sections of Gonda Kurubas, schedule castes and tribes etc. of southern Karnataka. Thus while collecting sample from say Mandya, Vokkaligas should get weightage of 30%, but while collecting samples from say Khanapur or Arabhavi, they don’t get any independent weightage at all.

Page 6: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

5Fort3 – Zealously Neutral Election Analysis Page 5

These are ideal targets percentages, but in real-life conditions we achieve slight variances due to various factors such as non-availability of respondents, unwilling women folk (orthodoxy), absent men folk (displaced due to work) and lack of time during data collection or even simple human errors etc. This is what we achieved in our controlled random sampling exercise all over the state;

Social Status Target percentage Percentage achieved in sampling

Urban 33% 42% Rural 67% 58% Men 60% 58% Women 40% 42% Lingayat (Veerashaiva) 21% 22% Vokkaliga 16% 14% SC (right-wing) 12% 13% SC (Left-wing) 9% 11%* Kuruba 8% 8% Muslim 14% 15% Random/Others 20% 17% (*Banjara votes numbering about 9-10 lakhs have been taken as part of SC-left-wing)

The weightages were then adjusted during number crunching, post-data collection phase, to arrive at our final overall projections.

Constituency selection criteria – A new revolution in opinion polling in India

This type of poll surveys can be done in a state like Karnataka with a sample size of about 3000 (for ex, the latest CSDS pan-India survey has a sample size of about 18k), but a much higher 12000+ target was chosen in order to not only get a pan-Karnataka picture but also clearly understand the picture of each of the parliamentary seats individually (a client requirement). This is possibly for the first time that such a deep level survey has been published in the public domain, for usually such surveys are conducted by political parties for internal consumption during elections mainly for ticket distribution purposes. Thus we have two different data points – 1) a pan-Karnataka vote-share and seat conversion directive and 2) Individual MP seat projections.

Of the 28 parliamentary seats, this survey was conducted in 25 seats and covered roughly 3 (2 being the lowest and 5 the highest) assembly segments of each parliamentary seat. Controlled random sampling method not only involves adequate ethnic profile weightage but also perfect polling booth selection criteria. This is where most of the opinion polls go wrong, for although their ethnic weightage formula is mostly right, the choice of targeted assembly segments and polling booths is faulty. Most of the pollsters, like CSDS, use the randomized PPS system – Probability Proportionate to Size system – for selecting assembly segments, whereas we at 5Forty3 have used a unique new mathematical model perfected after years of ground experience. Such a methodology, we believe, will completely alter our polling techniques and create near 100% accuracy in predicting outcomes.

Our revolutionary new methodology has been christened as SSI system – Social Swing Impact system. Due to proprietary reasons, the full methodology cannot yet be revealed at this point, but here we are presenting one of the results of this new system in the form of a real live example.

For instance, let us consider Mumbai-Karnataka as our geography, a region where BJP had made a clean sweep in 2009 LS polls by winning all the 6 seats and even now, after the decimation of the party in the

Page 7: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

5Fort3 – Zealously Neutral Election Analysis Page 6

assembly election, this is one regions where BJP still has strong presence. Our SSI system, when employed, projects that Bijapur LS seat is one of the most vulnerable of all of BJP’s seats in this region. The reason is quite simple – Congress party has won 7 out of 8 assembly segments in this LS seat in the May election. Now what assembly segments or polling booths should we target for an ideal sample survey in this LS seat? CSDS or other pollsters would simply use a randomized PPS system which may not be able to gauge the swing factor, but our system even with a similar sample size, adds another level of sophistication to the sampling methodology and targets the proper swing areas thereby giving far better results.

Page 8: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

5Fort3 – Zealously Neutral Election Analysis Page 7

Now consider this, our SSI system has been able to project Basavana Bagevadi as one of the swing assembly segments of Bijpaur LS seat for this exercise of ours. The reasons are not far to seek. In the last 4 assembly and parliamentary elections Basavan Bagewadi has been swinging between BJP and Congress each time. Thus a sample survey of B. Bagevadi is likely to give us a far better understanding of Bijapur LS seat than say a random survey of Sindgi assembly segment which may only throw-in a lot of whipsaws in favor of the BJP.

Once we have zeroed in on the assembly segment, our next quest is to target the right polling booths – once again, most pollsters just use a random methodology which is mostly flawed (but tracker polls do solve this problem to some extent by analyzing change and adding newer geographies with each subsequent sampling). Here again our SSI system has a new revolutionary approach to target specific polling booths based on our electoral roll data feed. Consider this; of the 5 target polling booths selected through our Randomized SSI (RSSI) system – a modified version of SSI – these are the two examples; polling booth number 67 (in B. Bagevadi town, dominated by Lingayats and middle class) and polling booth number 39 (of Krishnapur Tanda village and dominated by Banjaras). Now let us do a snap analysis of the second polling booth dominated by the Banjara community and correlate the same with our sample analysis.

Polling booth 39 Male Female Below 35 Above 35 Number of Voters 280 224 165 339

32.73% of the voters in polling booth number 39 are from the 18-35 age group, thus 5 of our respondents out of the total target of 15 in Krishnapur Tanda had to be from the younger age group and preferably 7 of the 15 respondents had to be female. In reality, although we achieved the 1st part, the second part was difficult to achieve and only 5 of the total respondents were women (but then, the Banjara community is overwhelmingly patriarchal in nature so the results of our survey can stand the test of time). Once these

Page 9: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

5Fort3 – Zealously Neutral Election Analysis Page 8

selection criteria are arrived upon, true randomization process kicks in. Using simple random sampling methodology for the 4 population sub-groups – male (280), female (224), above 35 (339) and below 35 (165) – we picked two lists of 15 names each to be targeted for our data collection. The first list is the “primary list” and the second list is the “mirror list”. In case any targeted respondents are unavailable during our visit for data collection, corresponding numbered respondents from the mirror list are then chosen for their responses. This way we achieve a relatively higher respondent hit rate. Now here is a stunning data point – all the 5 young voters (of the 18-35 age group of the Krishnapur Tanda), including a woman voter, preferred Narendra Modi as the next PM of India.

Thus our poll survey not only manages to get the social composition right, by giving adequate voice to marginalized communities like the Banjaras, but also takes into consideration the swing factor to arrive at the best possible result scenario. We believe that such a sophisticated methodology hasn’t been employed by any polling agency in India and most of them blindly follow western polling techniques like ESOMAR.

Mixed-Mode Methodology

We achieved roughly 80% of our targeted sample size of 12000, as we were able to get responses from 9468 respondents across the state. 4215 responses were achieved using the telephone survey methodology and 5253 respondents were interviewed by our field officers at their homes. Our field officers talked to the respondents clearly in a language that they understood, to get the responses from them and also carried a dummy ballot box for a mock voting drill with a proper ballot paper containing party names and symbols to be ticked in. Apart from Congress, BJP and JDS (in that order), our ballot papers also consisted of AAP, NOTA and a blank space for “others” (those who chose not to vote for any party, including “others”, were counted as part of “undecided/can’t say” category).

Telephone survey was also a similar exercise conducted by telecallers who interviewed voters on the phone rather than an IVR-based response system that many CATI systems employ. The same selection criteria, as mentioned in detail above, was used to select targeted responses for our telephone survey and the respective respondents (of specific geography and ethnicity) were contacted using a third party mobile number database. Thus there was hardly any difference between our face-to-face interview and telephone survey, except for the fact that there was no mock poll and preferences were ticked in manually (plus an overt urban bias in selection and a lower base to choose from).

Roughly 70% of the respondents in the BKT (Bangalore-Kolar-Tumkur) region were targeted using telephone survey, whereas only 40% of the respondents in Mumbai-Karnataka/Hyderabad-Karnataka were targeted using the same. Urban polling booths targeted by mostly phone survey had larger sample sizes due to bigger voting populace. None of our targeted polling booths had a sample size of more than 41 and less than 12.

Three different agencies independently conducted face-to-face survey in Karnataka, divided into three parts;

1) North & central Karnataka (consisting of Belgaum, Bijapur, Dharwad, Haveri, Bagalkot, Gulbarga, Bidar, Raichur, Yadgir, Koppal, Bellary and Davangere districts)

2) South Karnataka (consisting of Bangalore urban, Bangalore rural, Ramnagar, Kolar, Tumkur, Chikballapur, Mysore and Mandya districts)

3) Coastal Karnataka (consisting of Shimoga, Udupi, Chikmagalur, Uttara Kannada and Dakshina Kannada districts).

(None of the three agencies have any political affiliations to any of the political parties or organizations)

Page 10: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

5Fort3 – Zealously Neutral Election Analysis Page 9

Telephone survey was conducted jointly from Bangalore and Gulbarga by 3 volunteers and 4 paid telecallers simultaneously by dividing the state into 7 different regions for convenience. As is the industry norm, mobile number database is mapped using pin-codes, but we also independently mapped the requisite mobile number database to electoral rolls which reduced the base size of the target population (of the electoral roll) by about 25-30%; a tradeoff that was unavoidable.

Questionnaire System

Apart from getting polling numbers for different political parties, any smart survey also gets some very important inputs from the voters that gives us vital information about electoral trends. Unfortunately, most of this information is not “sexy” enough to enthuse viewers/readers unlike those big numbers of vote-share and seat-share, so pollsters just gloss over this information. Since 5Forty3 was involved with this poll survey from the scratch, we devised a robust questionnaire system which would give us maximum information within the limited time. This Questionnaire system is divided into three parts;

A. Factual Information: a. Education level – illiterate, primary schooling, matriculation, intermediate, graduate or

postgraduate b. Income level – BPL, lower middle class, Middle class, upper middle class, high class etc. c. Marital status – married, unmarried, divorced, widowed or none/no answer d. Political affiliation – 2009 LS poll vote, 2008 & 2013 assembly election vote

B. Opinions a. State government performance, comparison with previous government/s b. CM performance and comparison c. Central government performance and PM rating d. BSY’s return to BJP e. Issues on which voters decide on voting preference and who is best suited to tackle those

issues f. Performance of Local MP g. Heard of AAP? If yes, would you vote for AAP?

C. Preferences a. Choice for PM b. Vote 2014

Factual information and preferences are simple multiple choice questions to avoid ambiguity, whereas most opinions are complex answers from the voters. Choice for PM had 5 options – Modi, Rahul, Deve Gowda, Others (specify) and “can’t say”. Vote 2014 had Congress, BJP, JDS, AAP, Nota and Others were the options (not voting for any option was also a choice of the voter).

Error management cycle

All poll surveys are prone to error, but some are more prone and others are less prone to errors. Our OSOP survey falls in the third category of a relatively zero error system (although no survey is absolutely correct because of the inherent nature of poll surveys which depend on statistical selection of a subset of the population rather than surveying the whole population). Based on our long experience we can say that there are three main error zones for political poll surveys in India;

1) Selection bias: Our methodology has possibly a near perfect selection probability (as explained earlier), therefore this poll survey has produced one of the most robust results.

Page 11: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

5Fort3 – Zealously Neutral Election Analysis Page 10

2) Random Sampling Error: Since we have used a controlled random sampling methodology this problem has been almost completely negated; for instance, most pollsters simply use a randomized sample across the geography to arrive at findings, but our system has extensively used electoral rolls in conjecture with our own ethnic (caste) database as the sampling frame.

3) Non-Response evaluation error: Most poll surveys completely ignore “no responses” and simply concentrate on the positive responses, which artificially inflates findings. Our system actually treats “no responses” as positive findings and assigns value to them, which adds a whole new dimension to understanding people’s choices

Number analysis and derivatives

Processing raw data clinically after scanning and digitizing the ballot papers and questionnaires is itself an exercise that can lead to various errors. It is of paramount importance that the derivatives of vote-shares be adjusted accurately based on various parameters, for raw data in itself only represents roughly 60% of the reality. Over the years, Dr. Praveen Patil, the primary analyst, has developed independent methodologies for different states of India to derive the final numbers. Apart from standard operating procedures like adjusting data for over representation for larger (national) parties and under representation for independents and “others”, there are also minor tweaks to the analysis system – for instance, past experience has shown us that poll surveys tend to show higher degree of pro-incumbency in and around Bangalore rather than interiors of northern Karnataka, so data is tweaked accordingly. Here are the factors that have been incorporated before deriving final percentages;

1) Balancing caste-representation based on vote weightage rather than merely census data (unique system unlike any other pollsters in India)

2) Making gross adjustments to higher representation of ruling parties (standard operating procedure)

3) Making gross adjustments to higher representation to well entrenched bigger (national) parties and lower representation to smaller parties and independents (standard operating procedure)

4) Tweaking system to balance sub-regional anomalies – polls tend to over represent dominant parties in specific geographies which is tweaked proportionally (unique system unlike any other pollsters in India)

5) Adjusting weightage of marginalized castes and minorities who are classically expected to falsify their choices mainly due to the fear factor (standard operating procedure)

Raw Data

For the first time in India, 5Forty3 is putting up partial unmapped raw data in the public domain. This raw data is oriented along pin-codes as this is the result of our telephone survey. We have put up both raw vote-data as well as raw PM preference data. The much more substantial and robust raw data of the Face-to-Face survey has not been published as of now due to non-permission from our primary sponsor and client, but going forward 5Forty3 would like to put-up everything into the public domain under the Open Source Model so that everyone can validate polling data independently. Our open source model is a big step towards liberating election analysis from the black magic territory to a much more robust scientific validation and analysis.

Important Note: This is simply raw data as collected by our Tele-callers, which is then remapped geographically to polling booths and to ethnicity of our original model based on sequential numbered system.

Page 12: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

5Fort3 – Zealously Neutral Election Analysis Page 11

The 2014 Vote: Telephone Survey

Pin code Congress BJP JDS AAP NOTA Others/Can't say 582101 8 6 0 0 0 4

582101 7 2 0 0 0 5

582120 7 7 1 0 0 4

581110 4 8 2 0 0 4

581110 3 7 1 0 1 3

580001 9 18 3 3 0 4

580001 13 15 4 2 0 3

580003 7 9 1 1 0 5

580011 6 7 2 4 1 5

580006 5 7 2 3 0 4

580020 12 15 5 2 1 4

580020 8 5 1 1 0 6

582208 6 8 0 0 0 2

591201 5 5 1 0 0 3

591223 4 9 0 0 0 3

586101 6 8 2 1 0 3

586101 7 7 1 0 0 4

586103 6 9 1 0 0 3

586108 5 7 2 1 0 4

586115 7 5 2 0 0 9

590001 15 9 4 2 0 7

590001 11 17 2 1 1 6

590001 9 14 6 0 0 7

590016 9 15 4 1 0 8

591307 6 6 1 0 0 4

587101 6 14 1 0 0 7

587101 8 12 0 1 0 7

587301 7 5 0 1 1 4

587118 5 3 1 0 0 5

583231 14 12 5 0 0 5

583228 14 9 5 0 0 6

583227 15 7 6 1 0 5

584101 13 9 6 0 0 9

584101 11 8 4 0 0 11

584102 18 6 7 0 0 6

585201 15 15 4 0 0 5

585223 12 16 7 0 0 4

585101 14 12 3 1 0 6

585102 19 8 3 1 0 5

585222 12 6 2 0 0 4

585301 14 3 3 0 0 4

585401 14 7 5 0 0 6

Page 13: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

5Fort3 – Zealously Neutral Election Analysis Page 12

585403 13 5 5 0 0 5

585327 9 6 11 0 0 5

585226 8 12 6 0 0 4

585311 9 8 3 0 0 3

583101 14 9 3 0 0 10

583103 12 9 4 2 0 10

583201 12 14 2 0 0 8

577001 17 8 3 1 0 6

577002 16 10 2 0 0 7

577601 9 5 5 0 0 6

562101 12 5 6 0 0 4

562123 14 8 7 0 0 5

562114 12 8 9 0 0 7

560064 12 12 4 0 0 7

563101 14 8 6 0 0 8

563101 15 9 5 0 0 7

563131 11 4 6 0 0 4

563119 7 3 2 0 0 7

562106 15 7 7 0 0 6

562106 14 11 6 0 0 5

562120 11 4 15 0 0 6

560098 13 11 8 0 0 6

560098 14 10 9 0 0 5

560078 13 13 6 3 0 4

560036 13 16 2 1 0 6

560036 13 14 3 2 0 7

560024 11 15 2 0 0 8

560023 11 10 11 0 0 6

560100 8 16 4 5 2 6

560005 21 8 2 0 0 7

560005 19 10 4 0 0 6

560086 8 17 3 3 1 6

560086 11 15 2 4 2 5

560051 17 12 0 0 0 9

560093 9 21 2 4 0 5

560092 11 15 4 0 0 7

560038 9 16 4 6 2 4

560032 16 11 1 2 0 7

560058 14 10 9 2 0 6

560058 13 14 5 0 0 5

560018 13 9 8 0 1 5

560018 14 11 6 0 0 8

560068 7 16 2 6 0 7

560068 12 13 4 2 0 6

560076 14 14 2 4 1 5

Page 14: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

5Fort3 – Zealously Neutral Election Analysis Page 13

560076 11 16 1 5 0 6

560040 17 12 2 4 0 4

560041 14 13 3 5 1 5

560040 15 13 2 3 1 6

560048 11 13 6 5 0 5

560041 17 13 2 1 0 6

560004 13 15 2 4 0 5

560053 16 11 2 2 1 7

560050 12 14 4 0 3 6

572101 6 14 11 0 0 5

572101 7 12 13 0 0 6

572201 5 9 6 0 0 5

572132 6 6 11 0 0 4

571401 12 4 11 0 0 5

571401 14 2 7 0 0 7

571401 11 0 14 0 0 5

571438 9 2 7 0 0 6

571431 5 1 6 0 0 9

570001 14 6 12 1 0 6

570005 16 5 9 0 0 7

570008 17 3 8 0 1 6

571105 10 2 8 0 1 5

570019 11 6 6 1 0 5

575001 14 14 2 0 0 7

575001 15 14 2 0 0 7

575001 14 18 1 0 0 4

575001 15 18 4 0 0 4

574227 12 10 3 0 0 6

574211 11 11 2 0 0 4

577201 9 12 2 0 0 5

577201 8 13 0 1 0 4

577202 10 11 2 0 0 6

577427 6 27 1 0 0 3

577429 11 12 1 0 0 7

577301 8 14 10 0 0 4

577302 9 13 9 0 0 6

576101 9 9 5 0 0 5

576101 11 8 4 0 0 6

577139 8 14 6 0 0 3

577101 9 15 4 0 0 3

581301 8 12 3 0 0 7

581301 7 12 2 0 0 8

581401 11 13 3 0 0 6

581402 12 13 2 0 0 5

581359 13 8 4 0 0 5

Page 15: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

5Fort3 – Zealously Neutral Election Analysis Page 14

PM Preference: Telephone Survey

Pin Code Modi Rahul D.Gowda Others Can't Say 580011 12 9 0 2 2

580006 11 8 0 2 0

580020 24 10 4 0 1

580020 11 10 0 0 0

582208 10 6 0 0 0

591201 8 6 0 0 0

591223 11 4 0 0 1

586101 12 7 1 0 0

586101 14 3 0 2 0

582101 10 4 0 1 3

582101 9 5 0 0 0

582120 9 7 0 1 2

581110 12 3 0 0 3

581110 11 2 1 1 0

580001 23 9 2 0 3

580001 18 13 1 3 2

580003 13 6 0 3 1

586103 14 4 0 0 1

586108 12 2 1 1 3

586115 11 9 1 0 2

590001 14 12 3 4 4

590001 33 3 1 0 1

590001 27 8 0 1 0

590016 24 9 3 1 0

591307 8 7 2 0 0

587101 24 4 0 0 0

587101 21 2 2 3 0

587301 8 8 1 0 1

587118 9 3 2 0 0

576101 11 8 3 1 4

576101 10 11 3 5 0

577139 20 5 3 3 0

577101 23 5 2 1 0

581301 21 6 1 2 0

581301 20 8 0 1 0

581401 15 10 0 3 5

581402 16 10 0 2 5

581359 10 11 3 4 2

575001 27 10 0 0 0

575001 25 11 1 1 0

575001 30 6 0 0 1

575001 33 6 0 2 0

Page 16: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

5Fort3 – Zealously Neutral Election Analysis Page 15

574227 20 10 1 0 0

574211 16 9 0 0 3

577201 20 6 0 0 2

577201 20 4 0 2 0

577202 15 10 1 1 2

577427 37 0 0 0 0

577429 15 9 0 0 7

577301 20 3 10 1 2

577302 18 7 8 0 4

562101 10 12 4 1 0

562123 11 13 8 2 0

562114 16 13 4 1 2

560064 19 8 2 3 3

563101 12 12 4 4 2

563101 10 14 3 3 6

563131 9 13 3 0 0

563119 7 7 0 0 5

562106 10 13 5 2 5

562106 16 13 4 0 3

562120 10 10 10 3 3

560098 22 9 2 3 2

560098 24 8 3 3 0

560078 27 6 2 4 0

560036 30 3 0 5 0

560036 27 3 0 6 3

560024 24 4 1 7 0

560024 20 9 1 2 5

560005 11 16 2 2 7

560005 13 15 2 2 7

560086 33 0 1 3 1

560086 31 2 0 6 0

560051 18 15 0 3 2

560093 30 2 1 6 2

560092 21 8 3 1 4

560032 15 13 1 2 6

560018 15 13 6 2 0

560018 19 11 4 2 3

560068 27 3 0 8 2

560068 20 9 2 2 4

560076 19 11 0 6 4

560076 25 5 0 8 1

560040 17 14 0 8 0

560041 18 7 2 9 5

560040 19 9 1 5 6

Page 17: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

5Fort3 – Zealously Neutral Election Analysis Page 16

560041 17 14 0 3 5

560004 22 6 1 7 3

560053 16 12 2 5 5

560050 24 7 2 4 2

572101 30 0 6 0 0

572101 31 0 5 2 0

572201 15 3 5 2 0

572132 12 1 8 2 4

583231 17 13 5 1 0

583228 12 15 1 1 5

583227 10 14 2 2 6

584101 13 11 3 3 4

584101 14 14 2 3 1

584102 9 18 6 4 0

585201 21 11 3 4 0

585223 20 11 4 2 2

585101 15 15 6 0 0

585102 12 20 3 0 1

585222 14 10 0 0 0

585301 10 14 0 0 0

585401 22 8 2 0 0

585403 16 10 2 0 0

585327 22 3 6 0 0

585226 23 2 1 4 0

585311 13 7 1 2 0

583101 11 18 1 4 2

583103 16 11 2 5 3

583201 20 14 1 1 0

577001 17 11 2 3 2

577002 17 10 1 4 3

577601 10 13 2 0 0

571401 10 8 13 1 0

571401 7 12 9 2 0

571401 6 4 16 4 0

571438 5 8 9 2 0

571431 7 5 6 1 2

570001 10 10 12 4 3

570005 11 10 9 3 4

570008 10 12 9 1 3

571105 5 6 12 2 1

570019 12 8 6 3 0

Page 18: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

5Fort3 – Zealously Neutral Election Analysis Page 17

Open Source Licensing

The same GNU General Public License is applicable to this OSOP. The GNU General Public License is a free, copyleft license for an open source model of opinion polling, methodology and data release. All the contents of this OSOP can be shared, used and analyzed by anybody freely, unless specified in parts. All rights granted under this License are granted for the term of copyright on the entire operation, and are irrevocable provided the stated conditions are met. This License explicitly affirms your unlimited permission to run the unmodified OSOP as long as you credit the source. Each time you convey a covered work, the recipient automatically receives a license from the original licensors, to run, modify and propagate that work, subject to this License. You are not responsible for enforcing compliance by third parties with this License. Any profiteering or selling/reselling of the OSOP or the data and information contained within is prohibited under the Open Source global model of GPL.

Page 19: OSOP: Open Source Opinion Poll of Karnataka · A more complex example is that of the Lingayat Vote. For instance, there are roughly 65 to 69 lakh Lingayat voters (of all the subsects)

5Fort3 – Zealously Neutral Election Analysis Page 18

Notes


Recommended