Adarsha Rural Municipality

Kaski

Bhaktapur

Dharan

Social Media Analysis
Nepal by-election 2019

Nepal By-elections 2019 Electoral Social Media Analysis is a platform where citizens’ participation on elections is mapped, analyzed and visualized.

Local Interventions Group, with the support of the European Union and UNDP launched an innovative mechanism with geo-mapping, crowd-sourced citizen reports and visualization – to understand how digital technology and social media shape and impact democratic engagement, civic freedoms, and public discourse on elections. 4 electoral constituencies were picked for the test run of this tech innovation: Parliamentary by-election in Kaski, Provincial Assembly by-election in Bhaktapur, Municipality by-election in Dharan and Ward by-election in Doti. 

See how we are conclusively creating a digital technology blueprint for future elections and forever change the way the integrity of elections is protected. 

The test pole project ran from November 23, a week before the by election of November 30 to December 03, roughly 48 hours after the election results were announced.

Note: The data, findings and analysis do not reflect the views and official position of the EU and UNDP.

Social Media Enagegment of Citizens
Key numbers in aggregate
.

Measuring election discourse on social media. 

By-election of Dharan municipality showed strong social media presence, with 143 verified posts. 

Parliamentary by-election in Kaski followed closely behind, with 130 verified social media posts. 

Bhaktapur’s provincial by-election was third most active on social media post, with 92 verified social media posts. 

Doti’s Ward level by-election, a remote constituency, was least active with 37 verified social media posts. 

Total Verified
Reports Analysed

399

Analysis of Social Media Platforms…

Which social media platforms were prominent? 

Facebook was the most widely and comprehensively used social media platform during the by-election. 

85% of the verified reports were on Facebook. 

Twitter’s presence on the citizens’ election discourse was significantly less at 14% of the total social media reports. 

Youtube’s presence was insignificant, at less than 1%.

Total Engagement

399

Breakdown of social media
posts by categories.
Why was social media used?

Social media usage by citizens were grouped into 8 different categories to better understand why and how citizens used social media during the election. 

Most social media posts (41.07%) were related to campaigning – meaning citizens used social media to promote their own political beliefs and affiliations. 

14.79% of posts were related to post-election reflections; people tend to be active on social media after they have participated in voting or when the initial results have started to come in. 

Almost 10% of the social media posts were used to attack opposition parties or candidates and 5.10% of the posts fell into the category of disinformation – where users have actively fabricated truths to favor their own political beliefs and inclinations. 

Positive reinforcement, for example encouraging people to use their right to vote, sharing voter education materials and calling for free and fair elections stood at 8.16% – an encouraging aspect to be noted – that citizens are spreading positive messages around election.

Posts relating to disputes were just a few, totaling just 0.25% – another positive aspect for Election Commission that conducted the by-elections in 52 different constituencies across the country. 

Breakdown of social media
posts by categories and constituencies .

Why was social media used?

Across the selected electoral constituencies, social media posts relating to campaigns scored the highest, with posts in Adarsha Doti at 45.9% of the posts relating to campagns, followed by 43.5% in Bhaktapur, 42.9% in Dharan and 33.8% in Kaski. 

Social media posts that did not fall into pre-determined categories were catalogued as ‘others’ – with 16.9% of posts falling into this category in Kaski, 14.3% in Dharan, 13.6% in Adarsha, Doti and 10.8% in Bhaktapur. 

Social media posts that were clearly opposition attack was highest in Kaski – with 13.8%, followed by Adarsha, Doti at 10.8%, Dharan at 8.6% and Bhaktapur at 3.3%. 

13% of the posts in Bhaktapur fell under ‘voters education’ in Bhaktapur, which was the highest of the 4 target constituencies, followed by Dharan at 9.3%, Adarsha, Doti at 5.4% and Kaski at 4.6%. 

Social media posts that sought to spread disinformation/misinformation stood generally low, with 7.7% in Kaski, 5.4% in Adarsha Doti, 3.6% in Dharan, and 3.3% in Bhaktapur. 

Only 2 social media posts, resulting in 0.7% of the total fell under ‘dispute resolution’ in Dharan, with 0% for all other 3 constituencies.

Sentiment, Factual and
Partisan Analysis.

Measuring the bias
and prejudice

All social media posts were further categorized and analysed. 64.9% of the posts fell under positive – which is encouraging – 22.3% were of negative sentiment, and 12.8% did not fall under either positive or negative.

62.4% of the posts were partisan, and 37.6% were non-partisan – meaning no inclinations towards any political party or belief were shown.

51.6% of the posts were factual, with substantial 45% not known/unclear, and 3.3% of the posts were definitely non-factual.

Considerable human input was mobilized to analyze these data, as natural language processing is required in any analytical tools that were not used due to time and resource constraints.
Gender Spread in Social
Media Engagement.
How inclusive is social media presence?

Social media presence of women is at 10% of the total engagement of men in general. Women in Kaski topped the engagement level with 10.8% compared to men. 

The lowest participation of women on social media around election was in Adarsha, Doti, with 5.4% against men’s participation of 94.6%.

Total Engagement

399

Social Media engagement
relating to political parties
.

Mapping the visibility/mentions of political parties

Nepal Communist Party featured strongly in Parliamentary by-election in Kaski and Ward level by-election in Doti.

Social Media posts relating to Nepali Congress were at the highest in provincial by-election in Bhaktapur – the constituency where Nepali Congress eventually won.

Total Engagement

399

Social Media engagement
relating to political parties
.

Mapping the visibility/mentions of political parties

Nepal Communist Party featured strongly in Parliamentary by-election in Kaski and Ward level by-election in Doti.


Social Media posts relating to Nepali Congress were at the highest in provincial by-election in Bhaktapur – the constituency where Nepali Congress eventually won.

Total Engagement

399

When were social media platforms most active?

When were the citizens more active during the election? 

Social media was mostly used during the campaigning phase, 38% of the posts were during the campaigns.


35% of the posts from members of the public were after the results were announced.


25% of the social media users posted on their social media accounts on the election day.

Posts During Silence Period

Silent Period:
NOT SO SILENT

Social media use during the silent period of the by-election (48 hours before the voting on November 30, 2019) was widespread. 141 social
media post out of 399 (36% of the total) analyzed were during the silent period. Although identification of political cadres was not done during this project, most of the posts during the silent period came from the members of the public with deep
affiliations to political parties or had string political beliefs along party lines. 60% of the posts during silent period were for campaigning, disinformation or opposition attack

What People Said.

WORD CLOUD:
visual representation of text data.

What did the people talk about the most during the by-election?

We have analyzed text from all social media reports and retrieved the frequency of each word within the text as Word Cloud – a visual representation of text data whose importance is visualized by way of their size and color.

Within ‘Opposition Attack’ category – picnic,
allowance, don’t vote featured strong – indicating
public messaging around how political parties are
spending on picnic (feasts) during the by-election, and using the word allowance (elders’ allowance) to underscore the Nepal Communist
Party’s trademark policy initiative.

Within ‘Disinformation’ category, Nepali Congress
featured strongly, which illustrates the word Nepali Congress was used frequently during social media posts that fell under disinformation/-
fabrication of truths category.

Total Engagement

399

Social Media Analysis
Nepal by-election 2019

Doti. Kaski. Bhaktapur. Dharan.