How The Hindu Data Team mapped Chennai’s street-by-street voting patterns

(This article forms a part of the Data Point newsletter curated by The Hindu’s Data team. To get the newsletter in your inbox, subscribe here.)

Hello, data enthusiasts!

On July 21, 2024, The Hindu’s Tamil Nadu edition carried a Sunday special that we, the Data Team, are particularly proud of. We dove deep into the voting patterns of Chennai, providing a street-by-street analysis that revealed some fascinating insights into the political landscape of the city.

The online versions of the story can be read here:

1. 2024 polls: How people in high and low income areas voted in Chennai’s Mylapore, T Nagar and other areas

2. How voting patterns differed between Chennai’s areas with “huts” and “apartments” in 2024

3. How Chennai’s areas and streets voted in the 2024 Lok Sabha polls: A searchable list

Our data-driven story highlighted how the DMK continues to resonate strongly with the urban poor, while the BJP sees more support among higher-income groups. Despite this general trend, the DMK managed to garner votes even in affluent streets, whereas the BJP struggled in lower-income areas. But how did we get this detailed, street-wise voting data?

To paint this comprehensive picture, we triangulated data from three primary sources:

First step

Polling station data: We started with a list of all polling stations and the specific areas they served during the 2024 Lok Sabha (LS) polls.

The data is available here.

When you select an assembly constituency, it shows the “polling station number,” “polling station location,” and “polling area.” It’s important to note that multiple polling stations can be situated at the same location, with each serving specific streets or a group of streets within that area.

For example, polling station number 1 in the Thousand Lights Assembly Segment covers households in Rajeevgandhi Nagar and Canal Bank Streets. This detailed information was collected for all assembly segments across the three Parliamentary constituencies of Chennai.

This information was then converted into Excel using Adobe PDF to Excel converter or other similar software tools. The data was then cleaned and reorganised, ensuring that any mismatched columns were corrected. All the data from each assembly segment was consolidated into a single sheet, and subsequently, the data for each parliamentary constituency was merged into one comprehensive sheet. This resulted in a single file — for Chennai as a whole — containing all the polling station numbers, locations, and areas that voted in them.

Second step

Next we tapped into the Form-20 Data from the Election Commission.

The data is available here.

This dataset details the party-wise votes polled at each station. For example, in the previously mentioned station, 260 valid votes were cast, with the DMK receiving 148 votes, the DMDK 47 votes, the BJP 33 votes, and so on. We compiled this information for every polling station to analyse voting patterns.

This data was collected for all polling stations across all assembly constituencies and later merged into a single comprehensive file containing polling station-wise voting information for the entire city of Chennai.

Third step

The two files were merged using the VLOOKUP function in Excel. This function requires a key column, meaning a column that is common to both files being combined. We created this key column by merging the polling station number with the name of the assembly segment. For instance, in the aforementioned example, the key value would be “Thousand Lights 1,” which is a combination of the assembly segment name and the polling station number.

Fourth Step

To analyse the voting trends based on wealth and income, we used the guideline value of streets/areas, which is the minimum value at which properties can be registered with the Tamil Nadu Registration Department. This guideline value serves as a good indicator of the wealth/income levels of voters. The values ranged from ₹3,800/sq ft in lower-income neighbourhoods to over ₹10,000/sq ft in higher-income areas. The lower the value, the poorer the households residing in that street and vice-versa.

The data can be found here. 

Problems faced in the final push

We reviewed the records for each area and determined the guideline values of all the streets in Chennai. This allowed us to categorise the streets as low-income or high-income. With this third database, we now had all the necessary data for our analysis: we could identify which polling stations served which areas, how those areas voted, and whether they were generally richer or poorer neighbourhoods.

Here is a screenshot of Pallaku Ma Nagar, a low-income neighbourhood in Mylapore, Chennai.

This step was the trickiest of all. The street names in the third database often didn’t match directly with the street names in the Election Commission’s polling booth data. Additionally, the guideline data might be more recent, while the ECI’s street data could be outdated. There were also many instances of streets with the same name within an area.

We encountered a unique problem in Velachery. A place near the 100 Feet Road had a higher guideline value, while a place with the same name, located further in the interiors of Velachery, had a lower guideline value. Proximity to main roads significantly increased the value.

Another challenge arose in Adyar and Thiruvanmiyur, where posh areas and poorer areas voted at the same polling station due to their proximity. The streets were too intermixed to discern a clear pattern, which had to be considered.

Furthermore, it was possible for the first five streets in, say, Area X to vote at one polling station, while the next five streets in the same area voted at a different station, thus splitting the votes. To address this, we provided a “range of votes” to establish a pattern for such areas.

In Anna Nagar, which is well-structured, we faced another challenge: all the roads are named as alphabetical blocks rather than regular street names. While this might not seem like an issue, it was confusing to navigate through rows and rows of streets labelled as AC Block, AL Block, AJ Block, and so on.

Arya Gowda Road, one of the longest roads in Chennai, extends into West Mambalam but is classified under the Kodambakkam area by the Tamil Nadu Registration Department. We had to take such geographical nuances into account while labelling areas. 

That’s it for now, folks! We hope you enjoyed this insider’s view into how we conducted the street-by-street vote analysis.

While you’re here, be sure to check out some of the other important stories we’ve published in recent weeks.

Nepal’s treacherous skies : With 741 plane crash deaths, country ranks 11 of 207 nations 

Sikar, Namakkal, Kota: Select “coaching hubs” are host to many high scoring NEET-UG-2024 candidates

Among party-hopping defectors, 89% fielded again, 70% re-elected, many become ministers: Data

No ministers represent Tamil Nadu in the new Union Ministry: Data

Personal loans disbursed via digital apps have the highest share of overdue accounts: Data

Here are some important News in Numbers from last week:

Time taken by Pan Zhanle to finish 100m free-style in Paris

46.4 in seconds

Pan Zhanle of China broke his own world record on the way to winning the men’s 100-meter freestyle Wednesday night, for his country’s first swimming gold medal of the Paris Olympics. That mark topped his 46.80 swum at the February world championships in Doha.

Number of complaints of child pornography, rape received on NCRP portal

1.94 in lakh

The National Cybercrime Reporting Portal (NCRP) has received 16.18 crore visitors till April 30, according to women and children development minister Annapurna Devi. The situation remains so despite more than 69.05 lakh cyber tipline reports being shared and the Emergency Response Support System being operational in all states and union territories.

Number of lives lost due to dog bites in India in 2023

286

Nearly 30.5 lakh cases of dog bites were reported during 2023, as per the Integrated Disease Surveillance Programme (IDSP). This is despite the number of anti-rabies shots for dog bite cases during the year 2023 being 46,54,398, as reported by National Centre for Disease Control.

Funds raised by the Kamala Harris campaign in the 2024 US presidential race

200 in $ million

This amount was raised in less than a week. Vice President Harris’ candidacy earned enough delegates to become the presumptive nominee 24 hours after announcing her campaign, following President Joe Biden dropping out of the race. She will be formally announced presidential nominee in next month’s Democratic National Convention (DNC) in Chicago.

Thank you for reading this week’s edition of the Data Point newsletter! To subscribe, click here. Please send your feedback to vignesh.r@thehindu.co.in