Analyzing Political Engagement on Twitter: A Statistical Comparison Between Democrats and Republican

Jiabei Zhang
Apr 12, 2023
6 min read

Introduction

Social media platforms have become a significant part of our daily lives, and Twitter, in particular, plays an essential role in political discourse. A recent report from Pew Research Center shows that one-quarter of American adults use Twitter, with one-third of tweets from U.S. adults being political. Twitter has become a vital platform for political parties to spread their political values and expand their influence. This study examines the difference in political engagement on Twitter between Democratic and Republican users and aims to determine which group is more politically active on the platform.

Research Question

Do Democrats have higher political engagements than Republicans on Twitter?

Hypothesis

H1: Democrats are more likely to have higher political engagement on Twitter than Republicans.

H2: Democrats are more likely to have higher political engagement on Twitter than Republicans across age groups.

H3: On average, Democrats have higher political engagement on Twitter than Republicans.

Identification of Data

This study uses a dataset (ATP W90) from Pew Research Center, where 2,548 U.S. adult Twitter users were surveyed in May 2021. The sample is representative of Twitter users aged 18 and older, living in the U.S., including Alaska and Hawaii. The dataset is pre-existing, and a secondary analysis was conducted.

Measures

Political engagement on Twitter is defined as an index measuring the extent to which users participate in politics on the platform. It includes various political activities, such as showing support for candidates or campaigns, tweeting or retweeting about political issues or events, using politically related hashtags, replying to political tweets, and changing profiles or posting for a cause.

To measure the Political engagement on twitter, I constructed an index called "TWPolitic_Engage" based on the following survey questions on ATP W90:

Here’s a list of activities some people do on social media, and others do not. Please indicate if you have done each of the following activities on TWITTER over the past year. [RANDOMIZE ITEMS]

a. Expressed your support for a political campaign or candidate

b. Replied to a tweet from a politician, political campaign or candidate

c. Posted a picture or changed your profile picture to show your support for a cause

d. Used a hashtag related to political or social issues

e. Tweeted or retweeted about a political or social issue

f. Tweeted or retweeted about a protest, boycott, or similar call to action

g. Tweeted or retweeted memes or humorous content related to political or social issues

RESPONSE OPTIONS:

1 Yes, I have done this in the past year

2 No, I have not done this in the past year

The responses to survey questions were recoded as follows: "Yes" as 1, "No" as 0, and missing values were assigned (99). The reliability test resulted in a Cronbach's Alpha of .845, indicating strong reliability.

The index, TWPolitic_Engage, is ordinal and ranges from 0 (no engagement) to 7 (very high engagement). The median value is 2.0, corresponding to 'low engagement'.

Methods

For H1 and H2, Crosstab was used for statistical analysis, with Age as a control variable for H2. Chi-Square test assessed the statistical significance, and Kendall's Tauc measured the association between variables. F_PARTYSUM_FINAL is a dichotomy, while TWPolitic_Engage and F_AGECAT are ordinal variables.

For H3, an independent samples T-test compared the means of Democrats and Republicans on TWPolitic_Engage. Although it is ordinal, the evenly spaced scale allows for treatment as interval-level data.

Analysis

Hypothesis 1

H1: Democrats are more likely to have higher Political engagement on Twitter than Republicans.

H2: Democrats are less likely to have higher Political engagement on Twitter than Republicans

The crosstab results display a noticeable trend, where a higher proportion of Democrats tend to be more politically engaged on Twitter across different levels of engagement, except for the 'very low engagement' category. This suggests that Democrats are generally more active in discussing politics, sharing opinions, and engaging with political content on the platform.

This pattern could be attributed to various factors, such as differences in demographics, social and political contexts, or the nature of the Twitter platform itself. For example, Democrats might be more inclined to use Twitter as a medium for political discourse and engagement, while Republicans might prefer other platforms or communication methods. Additionally, the age distribution and other demographic factors could also influence the level of engagement, as different age groups might have different preferences when it comes to using social media for political purposes.

It's important to note that the relationship between party affiliation and political engagement on Twitter is not very strong, as indicated by the Tauc value. This means that while there is a statistically significant relationship between the variables, there are likely other factors that contribute to political engagement on Twitter. Further research could explore these factors and investigate how they interact with party affiliation to influence political engagement on the platform.

In conclusion, the analysis supports the hypothesis that Democrats are more likely to have higher political engagement on Twitter than Republicans. However, it is crucial to consider the potential influence of other factors and the relatively weak relationship between the variables when interpreting these results.

Hypothesis 2

H2: Democrats are more likely to have higher Political engagement on Twitter than Republicans across age groups.

H0: Democrats are less likely to have higher Political engagement on Twitter than Republicans across age groups.

The TWPolitic_Engage * Party summary * Age category Crosstab reveals interesting insights about how people of different ages engage with politics on the platform. Understanding these differences can help political parties and campaigners better target their messages and connect with their audience.

Younger age group (18-29): The strongest relationship between party affiliation and political engagement is found in this age group. This suggests that younger individuals are more likely to express their political views on Twitter, with Democrats being more engaged than Republicans. Given the growing importance of social media in political campaigns, it's crucial for both parties to engage with younger voters on the platform.

Middle age groups (30-49 and 50-64): The relationship between party affiliation and political engagement is still evident in these age groups, but with varying strength. This could indicate that middle-aged individuals have diverse ways of engaging with politics on Twitter, and their engagement may be influenced by other factors such as education, occupation, or personal interests.

Older age group (65+): This group exhibits the weakest relationship between party affiliation and political engagement on Twitter. It may be because older individuals are less likely to use Twitter or social media in general for political discussions, preferring more traditional means of communication. However, as internet access becomes more widespread, political parties should not overlook this age group's potential for engagement on social media platforms.

Overall, the TWPolitic_Engage * Party summary * Age category Crosstab suggests that age should be considered as a crucial factor when examining political engagement on Twitter. Political parties and campaigners should tailor their strategies to better engage with different age groups, leveraging the unique characteristics and strengths of each demographic to maximize their outreach and impact.

Hypothesis 3

H3: On average, Democrats have higher political engagement on Twitter than Republicans.

H0: On average, Democrats have lower political engagement on Twitter than Republicans.

The analysis of the Group Stats table and the Independent T-test table reveals a significant difference in the average political engagement on Twitter between Republicans and Democrats. With a mean of 2.57 for Democrats and 1.64 for Republicans, the difference of 0.93 at p < .01 level indicates that this difference is statistically significant and not due to chance. Consequently, we reject the hypothesis that Democrats have lower political engagement on Twitter than Republicans on average.

The 0.93 mean difference suggests that Democrats' political engagement level on Twitter is almost one scale higher than Republicans'. This finding supports the argument that Democrats are more likely to have higher political engagement on the platform.

Additionally, the effect sizes table shows that the point estimate of Cohen's d is -0.41, indicating a moderate effect of party affiliation on political engagement on Twitter (TWPolitic_Engage). This further strengthens the argument that there is a meaningful difference in political engagement on Twitter between Democrats and Republicans.

Conclusion

After testing all the hypotheses, it is evident that Democrats exhibit higher political engagement than Republicans on Twitter. Democrats are more politically engaged across almost all age groups on Twitter, while certain age groups that show a different trend in political engagement levels warrant further investigation.

One limitation of this study is that the index constructed may not fully capture the political engagement level of American adults on Twitter. Besides the activities mentioned in the survey questions, Twitter users can participate in politics through other methods, such as following or blocking politically-related accounts, fundraising for political campaigns, or attending political webinars on the platform. Additionally, the dataset itself has limitations, as the unbalanced party distribution among respondents may introduce bias in the statistical testing and affect the results.

Further research could explore additional ways people engage with politics on Twitter, as well as other factors that might impact political engagement on social media platforms. Moreover, future studies could examine how political engagement on Twitter translates to real-world political action or voter turnout. By refining the index and addressing the limitations of the current study, a more comprehensive understanding of political engagement on Twitter can be achieved, which could inform strategies for political parties to better engage with their constituents on social media.