{"id":111574,"date":"2019-09-17T14:07:00","date_gmt":"2019-09-17T19:07:00","guid":{"rendered":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/decoded\/\/\/exploring-the-link-between-european-political-parties-ideology-and-their-popularity-on-twitter\/"},"modified":"2024-04-14T04:10:38","modified_gmt":"2024-04-14T09:10:38","slug":"exploring-the-link-between-european-political-parties-ideology-and-their-popularity-on-twitter","status":"publish","type":"decoded","link":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/decoded\/2019\/09\/17\/exploring-the-link-between-european-political-parties-ideology-and-their-popularity-on-twitter\/","title":{"rendered":"Exploring the link between European political parties\u2019 ideology and their popularity on Twitter"},"content":{"rendered":"\n
\"\"<\/a><\/figure>\n\n\n\n

Politicians and political groups regularly use social media to disseminate their views to constituents, supporters and others. But some political figures have more reach than others. In the United States, for example, the most liberal and most conservative members of the 115th Congress had more Facebook followers<\/a>than their moderate counterparts, as Pew Research Center has found. The same pattern holds when looking at congressional lawmakers\u2019 Twitter accounts<\/a>.<\/p>\n\n\n\n

Does this same phenomenon apply outside the U.S.? This is a difficult question to answer. But an initial analysis shows that, across 12 European countries, parties on the far left or right of the ideology scale aren\u2019t more popular on Twitter than their more moderate counterparts. In this post, we\u2019ll explain how we arrived at this preliminary finding and why expanding this kind of analysis beyond the U.S. is trickier than it sounds.<\/p>\n\n\n\n

Different measures of ideology<\/h4>\n\n\n\n
\"\"<\/a><\/figure>\n\n\n\n

Our previous work in the U.S. used an ideology score known as DW-NOMINATE<\/a>, which measures the ideology of members of Congress based on their roll call voting records. DW-NOMINATE is widely used in the research community, but it is only available for U.S. legislators. Moreover, this sort of measure doesn\u2019t easily translate to other countries, especially in the roughly 40% of nations<\/a> under some form of autocratic or single-party government rule.<\/p>\n\n\n\n

In an effort to see if we could identify an alternative source of objective ideology scores that might work in an international context, we turned to the University of North Carolina\u2019s 2017 Chapel Hill Experts\u2019 Survey<\/a> (CHES). We also decided to focus specifically on 12 European Union countries.<\/p>\n\n\n\n

The survey data<\/h4>\n\n\n\n

The 2017 Chapel Hill Experts\u2019 Survey asked a group of 228 political scientists around the world to score 132 European political parties on a variety of measures, from their views of multiculturalism to their preferred economic policies. Most important for the purposes of this analysis, the survey also included a right-left ideology scale.<\/p>\n\n\n\n

Multiple experts assign values to each party across all indicators, leaving the final point estimate allotted to each party for each variable \u2014 including ideology \u2014 as the mean of all experts\u2019 scores. This sort of collaborative scoring scheme allows for a fairly consistent evaluation of the same standards across multiple countries. Academics and previous Pew Research Center projects<\/a> have used the 2017 CHES data when classifying parties based on anti-elite sentiment and ideology.<\/p>\n\n\n\n

The social media data<\/h4>\n\n\n\n

The analysis involved building a new dataset of Twitter handles for European political parties alongside the number of people following each party\u2019s account. We compiled a list of political parties from across Europe and matched it to those included in the 2017 CHES survey. We then searched for each party on Twitter, using both their English name and the native language translation, to compile the list of handles. We could have also recorded the number of followers for each party at this stage in our research, but the counts on the Twitter website are less precise than those obtained via Twitter\u2019s public API.<\/p>\n\n\n\n

To collect information on each party\u2019s Twitter following, we used the \u201crtweet<\/a>\u201d package in R, the open-source statistical program. This library allows users to access information via the Twitter API<\/a> for all public Twitter accounts, including the number of followers they have, the language of their tweets and the text of their Twitter bios. To use this function, however, we first had to apply for a Twitter developer account<\/a>, which enables you to register your app\/project with the company and access their data.<\/p>\n\n\n\n

Since we were only interested in the number of followers for each political party\u2019s Twitter account, we used the \u201clookup_users\u201d function within the \u201crtweet\u201d package. After manually compiling the list of political parties\u2019 Twitter handles, we executed the following lines of code to extract Twitter data for each user.<\/p>\n\n\n\n

\"\"<\/a><\/figure>\n\n\n\n

Running that code returns a tibble data frame with one row for each party and 90 variables, including the user\u2019s listed location, language of the last tweet, user-provided biography and the number of followers. Since the information is collected almost simultaneously, we have more confidence that the followers for each party are comparable.<\/p>\n\n\n\n

After collecting follower counts, we removed columns containing data we didn\u2019t need and merged the social media data with the CHES dataset. Not all parties in the CHES dataset have active Twitter accounts, but we were able to collect this data for 112 of the 132 parties in the dataset.<\/p>\n\n\n\n

Exploratory analysis<\/h4>\n\n\n\n

Ultimately, we decided to drop parties whose Twitter accounts have fewer than 10,000 followers. We excluded these parties because they were not very active on Twitter, did not have accounts that were clearly official in nature, or existed to promote more narrow causes that didn\u2019t necessarily tell us much about political ideology in general. This left us with 83 parties across 12 countries, with a median Twitter following of 85,000.<\/p>\n\n\n\n

We next divided the sample of parties into three categories based on the CHES ideology scale, which we re-centered from 0 to 10 to -5 to 5. The categories include left parties (-5 to -2 on the scale), center parties (-2 to 2) and right parties (2 to 5). When graphing party ideology versus the log number of Twitter followers, we observed that more ideological parties tended to have modestly fewer<\/em> Twitter followers:<\/p>\n\n\n\n

\"\"<\/a>
Far left\/right parties don’t tend to have more Twitter followers<\/figcaption><\/figure>\n\n\n\n

Estimating a random-effects model to explore the pattern<\/h4>\n\n\n\n

The scatterplot does not<\/em> take into account the fact that each political party\u2019s number of followers could be affected by factors such as a country\u2019s population, rates of Twitter use or education levels. As a result, we decided to estimate a multi-level model with separate intercepts for each country to further examine ideology\u2019s effect on Twitter following:<\/p>\n\n\n\n

\"\"<\/a><\/figure>\n\n\n\n

This allows us to predict the number of followers a party has on Twitter based on the absolute value of their ideology score and gauge how \u201cextreme\u201d they are in either direction. The random effects<\/a> help account for unobserved attributes of countries that might influence the number of people who follow parties (and thus make cross-country comparisons more difficult). In this case, we assumed that different countries\u2019 Twitter usage rates are not correlated with parties\u2019 ideological extremity, for example.<\/p>\n\n\n\n

\"\"<\/a>
Predicted number of followers by extremism of CHES ideology ratings<\/figcaption><\/figure>\n\n\n\n

We found that absolute value of ideology has a negative but statistically insignificant relationship with popularity. In other words, we don\u2019t find evidence that far left and right parties have systematically more Twitter followers. However, a fully specified model that also captures attributes like a party\u2019s overall popularity or other attributes of a party might find different results.<\/p>\n\n\n\n

Challenges and next steps<\/h4>\n\n\n\n

As noted earlier, this analysis suggests that European political parties on the far left or right of the ideology scale aren\u2019t any more popular on Twitter than their moderate counterparts. However, the comparison to our earlier U.S. findings is imperfect for several reasons. For one, legislators are not the same as parties. And while the U.S. ideology measure is based on roll call voting records, the European scale is a composite of subjective experts\u2019 scores. Without cross-country legislator-level ideology estimates that are consistent across countries, we are limited in our ability to directly replicate the U.S. analysis. Still, this initial finding provides an interesting basis for further analysis of cross-country patterns in European politics.<\/p>\n","protected":false},"excerpt":{"rendered":"

It can be hard to determine whether parties near the ends of the ideology scale are more popular on the platform than moderate parties.<\/p>\n","protected":false},"author":655,"featured_media":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"sub_headline":"","sub_title":"","_prc_public_revisions":[],"_ppp_expiration_hours":0,"_ppp_enabled":false,"ai_generated_summary":"","relatedPosts":[],"datacite_doi":"","datacite_doi_citation":"","_prc_seo_qr_attachment_id":0,"spoken_article_player_enabled":true,"displayBylines":true,"footnotes":"","prc_watchers":[],"_prc_fork_parent":0,"_prc_fork_status":"","_prc_active_fork":0},"categories":[353],"bylines":[936,853],"collection":[],"_post_visibility":[],"decoded-category":[530],"formats":[],"_fund_pool":[],"languages":[],"regions-countries":[],"research-teams":[524],"workflow-status":[],"class_list":["post-111574","decoded","type-decoded","status-publish","hentry","category-data-science","bylines-adam-hughes","bylines-kat-devlin","decoded-category-data-science","research-teams-decoded"],"label":"Decoded","post_parent":0,"word_count":1242,"canonical_url":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/decoded\/2019\/09\/17\/exploring-the-link-between-european-political-parties-ideology-and-their-popularity-on-twitter\/","art_direction":{"A1":{"id":125959,"rawUrl":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-content\/uploads\/sites\/20\/2022\/08\/09.17.2019_feature.png","url":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-content\/uploads\/sites\/20\/2022\/08\/09.17.2019_feature.png?w=564&h=317&crop=1","width":564,"height":317,"caption":"","chartArt":false},"A2":{"id":125959,"rawUrl":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-content\/uploads\/sites\/20\/2022\/08\/09.17.2019_feature.png","url":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-content\/uploads\/sites\/20\/2022\/08\/09.17.2019_feature.png?w=268&h=151&crop=1","width":268,"height":151,"caption":"","chartArt":false},"A3":{"id":125959,"rawUrl":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-content\/uploads\/sites\/20\/2022\/08\/09.17.2019_feature.png","url":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-content\/uploads\/sites\/20\/2022\/08\/09.17.2019_feature.png?w=194&h=110&crop=1","width":194,"height":110,"caption":"","chartArt":false},"A4":{"id":125959,"rawUrl":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-content\/uploads\/sites\/20\/2022\/08\/09.17.2019_feature.png","url":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-content\/uploads\/sites\/20\/2022\/08\/09.17.2019_feature.png?w=268&h=151&crop=1","width":268,"height":151,"caption":"","chartArt":false},"XL":{"id":125959,"rawUrl":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-content\/uploads\/sites\/20\/2022\/08\/09.17.2019_feature.png","url":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-content\/uploads\/sites\/20\/2022\/08\/09.17.2019_feature.png?w=720&h=405&crop=1","width":720,"height":405,"caption":"","chartArt":false},"social":{"id":125959,"rawUrl":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-content\/uploads\/sites\/20\/2022\/08\/09.17.2019_feature.png","url":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-content\/uploads\/sites\/20\/2022\/08\/09.17.2019_feature.png?w=1200&h=628&crop=1","width":1200,"height":628,"caption":"","chartArt":false}},"_embeds":[],"watchers":[],"table_of_contents":[],"datacite_doi":"","prc_seo_data":{"title":"Exploring the link between European political parties\u2019 ideology and their popularity on Twitter","description":"It can be hard to determine whether parties near the ends of the ideology scale are more popular on the platform than moderate parties.","og_title":"Exploring the link between European political parties\u2019 ideology and their popularity on Twitter","og_description":"It can be hard to determine whether parties near the ends of the ideology scale are more popular on the platform than moderate parties.","schema_type":"Article","noindex":false,"canonical_url":"","primary_terms":{"category":1},"custom_schema":[],"og_image":125959,"indexnow_submitted_at":null,"gsc_index_status":null},"prepublish_checks":{},"jetpack_sharing_enabled":true,"relatedPostsOrdered":[],"bylinesOrdered":[{"key":"_gdod74fws","termId":853},{"key":"_bjkdq94ze","termId":936}],"acknowledgementsOrdered":[],"_links":{"self":[{"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/decoded\/111574","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/decoded"}],"about":[{"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/types\/decoded"}],"author":[{"embeddable":true,"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/users\/655"}],"replies":[{"embeddable":true,"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/comments?post=111574"}],"version-history":[{"count":2,"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/decoded\/111574\/revisions"}],"predecessor-version":[{"id":138531,"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/decoded\/111574\/revisions\/138531"}],"wp:attachment":[{"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/media?parent=111574"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/categories?post=111574"},{"taxonomy":"bylines","embeddable":true,"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/bylines?post=111574"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/collection?post=111574"},{"taxonomy":"_post_visibility","embeddable":true,"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/_post_visibility?post=111574"},{"taxonomy":"decoded-category","embeddable":true,"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/decoded-category?post=111574"},{"taxonomy":"formats","embeddable":true,"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/formats?post=111574"},{"taxonomy":"_fund_pool","embeddable":true,"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/_fund_pool?post=111574"},{"taxonomy":"languages","embeddable":true,"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/languages?post=111574"},{"taxonomy":"regions-countries","embeddable":true,"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/regions-countries?post=111574"},{"taxonomy":"research-teams","embeddable":true,"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/research-teams?post=111574"},{"taxonomy":"workflow-status","embeddable":true,"href":"https:\/\/alpha.pewresearch.org\/pewresearch-org\/wp-json\/wp\/v2\/workflow-status?post=111574"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}