CoChat
Sign Up
Cited by 982Social Media and PoliticsElectoral Systems and Political ParticipationComputational and Text Analysis Methods

Birds of the Same Feather Tweet Together: Bayesian Ideal Point Estimation Using Twitter Data

Open access
Barberá, Pablo·2015·Political Analysis
DOI 10.1093/pan/mpu011
Open this paper in CoChat — free accountOpen PDF

Abstract

Politicians and citizens increasingly engage in political conversations on social media outlets such as Twitter. In this article, I show that the structure of the social networks in which they are embedded can be a source of information about their ideological positions. Under the assumption that social networks are homophilic, I develop a Bayesian Spatial Following model that considers ideology as a latent variable, whose value can be inferred by examining which politics actors each user is following. This method allows us to estimate ideology for more actors than any existing alternative, at any point in time and across many polities. I apply this method to estimate ideal points for a large sample of both elite and mass public Twitter users in the United States and five European countries. The estimated positions of legislators and political parties replicate conventional measures of ideology. The method is also able to successfully classify individuals who state their political preferences publicly and a sample of users matched with their party registration records. To illustrate the potential contribution of these estimates, I examine the extent to which online behavior during the 2012 US presidential election campaign is clustered along ideological lines.

AI snapshot

Written from the paper’s full text.

TL;DR

The paper develops a Bayesian Spatial Following model that infers Twitter users’ and political actors’ latent ideology from whom they follow, under a homophily assumption. Applied to the US and five European countries, the estimates replicate conventional elite and mass ideology measures and reveal high polarization in 2012 US election retweet networks.

Key findings

  • US congressional Twitter ideal points correlate strongly with DW-NOMINATE: ρ = 0.941 in the House and ρ = 0.954 in the Senate; within-party correlations are ρ = 0.546 for Republicans and ρ = 0.610 for Democrats. European party placements are broadly congruent with Chapel Hill Expert Survey estimates, though the UK shows lower accuracy and some extreme parties appear more centrist.
  • Mass validation: self-identified conservatives are to the right of independents, who are to the right of liberals; state median Twitter ideal points correlate ρ = 0.876 with Lax and Phillips’s liberal-opinion estimates, higher than with the 2012 Obama vote (ρ = 0.785). Among matched campaign contributors, 90% of 1,058 right-of-median users donated to Republicans and 98% of 1,649 left-of-median users donated to Democrats; correlation with contribution-record ideal points is ρ = 0.80.
  • In the 2012 US election application, 65 million tweets mentioning Obama or Romney were sent by 8.8 million users; the analysis focuses on 15 million tweets from the 301,537-user sample. Moderates with ideal points between −0.5 and 0.5 are almost 40% of the sample but sent only 21% of tweets; conservative users (0.5 to 1.5) averaged 118 tweets versus 82 for users between −1.5 and −0.5.
  • Retweet interactions are highly polarized: 85% occur among users with similar ideological positions. For Obama retweets, 20% are in the bottom-left quadrant and 69% in the top-right; for Romney retweets, 49% are bottom-left and 31% top-right. Liberal users’ retweets include 18% from moderate/conservative users, while conservative users’ retweets include only 7% from liberal/moderate users.

Methods

  • The Bayesian Spatial Following model treats ideology as latent and models the probability that user i follows political actor j as logit⁻¹(α_j + β_i − ||θ_i − φ_j||²), where θ_i is the user’s ideal point, φ_j is the actor’s ideal point, α_j is actor popularity, and β_i is user political interest. Estimation uses hierarchical priors, a No-U-Turn/Hamiltonian Monte Carlo first stage and a random-walk Metropolis-Hastings second stage, implemented in Stan and R.
  • Data cover the US, UK, Spain, Germany, Italy, and the Netherlands. Target political actors with more than 5,000 followers in the US or 2,000 elsewhere number m = 318 US, 244 UK, 298 Spain, 214 Italy, 273 Germany, and 118 Netherlands. After filtering inactive, spam, out-of-country, and low-following accounts, final samples are n = 301,537 US, 135,015 UK, 123,846 Spain, 150,143 Italy, 49,142 Germany, and 96,624 Netherlands; the US sample is further restricted to accounts tweeting at least three times about Obama or Romney in the three months before the 2012 election.
  • Validation compares Twitter estimates with DW-NOMINATE, Chapel Hill Expert Survey placements, self-identified ideology, Lax and Phillips’s state opinion estimates, campaign contribution records, and Ohio voter registration records. The polarization application collects all public tweets mentioning Obama or Romney from August 15 to Election Day, totaling over 65 million tweets by 8.8 million users, and analyzes the 15 million tweets sent by the 301,537-user sample.

Limitations

  • Paper-stated: Twitter users are not a representative sample of the voting-age population; the sample is self-selected, more educated and politically interested, and restricted to users following at least three political accounts, so inferences do not represent all Twitter users. The paper argues this is not critical for estimating politicians’ ideal points.
  • Paper-stated: results across countries are not directly comparable because estimation is performed independently and the resulting dimension lacks a homogeneous cross-country scale. The method as formulated estimates ideology only for individuals who follow at least one political account, and the latent dimension may collapse different dimensions, especially in two-dimensional European party spaces; the UK shows lower accuracy and some extreme parties appear more centrist.
  • Paper-stated: the retweet polarization conclusion may be partly driven by the assumption that users are more likely to retweet someone they follow, though the text notes retweets between non-followers are frequent. Evident from the text: country location is inferred from profile fields, with country identified in 90% of cases and US state in 71%; the US sample is further restricted to accounts tweeting about Obama or Romney.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1093/pan/mpu011

Cites

88 of 88 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

18 of 56 cited sources checked · 9 supported · 2 flagged · accuracy not yet measured

  • Interpretationreasoning

    Ideal points can be multiplied by any nonzero constant, with the normalizing constant divided by its square, leaving the model predictions unchanged.

    “similarly i or j can be multiplied by any nonzero constant, with divided by its square, leaving the model predictions unchanged.”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites Bafumi et al. 2005 source

      Section 2.2.2 describes multiplicative aliasing, but its compensating scale factor is linear, not squared. It says multiplying the c parameters by a constant while dividing both the a and b parameters by that same constant preserves the likelihood; it does not support dividing a normalizing constant by the constant’s square. (LLM, conf=0.88)

      Checked by a reasoning model — may be wrong.

      Experimental — accuracy not yet measured
    • cites Londregan 1999 source

      Section 3.2 (“First-Order” Identifying Restrictions; page not provided) supports the need to normalize the scale of legislators’ ideal points in the Ladha and linear-probability models. It does not state that dividing a normalizing constant by the square of any nonzero multiplier leaves predictions unchanged; for the Poole–Rosenthal model, it instead says β can be pinned down. (LLM, conf=0.82)

      Experimental — accuracy not yet measured
  • Premisecitation

    Choosing starting values consistent with the expected direction of the scale (e.g., liberals with -1 and conservatives with +1) is sufficient to ensure global identification in most cases.

    “As Jackman (2001) shows, choosing starting values that are consistent with the expected direction of the scale (e.g., liberals with À1 and conservatives with þ1) is sufficient to ensure global identification in most cases.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Jackman 2001 source

      Section 4 says that starting the Gibbs sampler with Democrats on the left and Republicans on the right is a practical way to select one posterior mode in most unidimensional roll-call settings. It explicitly distinguishes this from global identification: the posterior retains a mirror-image mode, and the priors ensure local identification, not global identification. (LLM, conf=0.90)

      Checked by a reasoning model — may be wrong.

      Experimental — accuracy not yet measured
  • Interpretationreasoning

    Any constant can be added to all ideal point parameters without changing the model's predictions.

    “any constant can be added to all the parameters i and j without changing the predictions of the model”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites Bafumi et al. 2005 source
    • cites Londregan 1999 source

      Section 3.2, “First-Order” Identifying Restrictions (page unavailable), treats the location of the ideological scale as arbitrary and normalizes one legislator’s position to zero. But Eq. (10) in Section 2.2 shows that voting depends on the difference between a legislator’s ideal point and a proposal’s midpoint. Adding a constant to legislators’ ideal points alone would change that difference; the prediction is unchanged if the proposal locations shift with them. (LLM, conf=0.84)

      Experimental — accuracy not yet measured
  • Interpretationreasoning

    The model described by equation (1) is unidentified.

    “The model described by equation ( 1) is unidentified: any constant can be added to all the parameters i and j without changing the predictions of the model; and similarly i or j can be multiplied by any nonzero constant, with divided by its square, leaving the model predictions unchanged.”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites Bafumi et al. 2005 source
    • cites Londregan 1999 source
  • Premisecitation

    Multiple studies have observed patterns of homophilic segregation consistent with these models in networks of interactions between Twitter users.

    “Multiple studies have observed patterns of homophilic segregation consistent with these models in networks of interactions between Twitter users (Wu et al. 2011; Conover et al. 2012).”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites Wu et al. 2011

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Conover et al. 2012 source
  • Classified as probably not a substantial claim. 39
  • May combine two claims. 40
  • The quote only partly states this claim. 1

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    These indeterminacies are common to item-response theory models and are called additive aliasing and scaling invariance.

    “These indeterminacies, which are common to item-response theory models, are sometimes called "additive aliasing" and "scaling invariance" (see, e.g., Bafumi et al. 2005; Londregan 1999).”
    Quote in the paperNot a substantial claim2 of 2 sources checked
    • cites Bafumi et al. 2005 source
    • cites Londregan 1999 source

      Section 3, “Issues in Model Identification” (page unavailable), says spatial voting models and their psychometric cousins, Rasch models, are not identified without further restrictions. Section 2.2 equates a spatial-model voting condition with one from an educational-testing model. The copy does not identify the indeterminacies as “additive aliasing” and “scaling invariance,” so it supports the shared identification problem but not those specific characterizations. (LLM, conf=0.82)

      Experimental — accuracy not yet measured

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationcitation

    Given the dynamic nature of social media, its large size, and individuals' finite ability to process incoming information, Twitter users should maximize the value of their online experience by choosing to follow political actors who can provide information of higher value to them.

    “Given the dynamic nature of social media, its large size, and individuals' finite ability to process incoming information (Oken Hodas and Lerman 2012), we should expect Twitter users to maximize the value of their online experience by choosing to follow political actors who can provide information that can be of higher value to them.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Oken Hodas and Lerman 2012 source

  • May combine two claims.

    Premisecitation

    Existing studies on ideal point estimation use two strategies to identify the model: applying two linearly independent restrictions on ideal point parameters, or applying a unit variance restriction.

    “Existing studies on ideal point estimation employ two different strategies to identify the model.One is to apply two linearly independent restrictions on the ideal point parameters, u or f in this case.”
    Quote in the paperNot a substantial claim1 of 3 sources checked
    • cites Londregan 1999 source

      Section 3.2 (“First-Order” Identifying Restrictions; page not provided) describes identifying models by imposing two restrictions on legislators’ preferred-point locations, including setting one location to zero and another to one in the Ladha and linear-probability models. The passages do not describe a unit-variance restriction as a second strategy, so that part of the attributed claim is unsupported. (LLM, conf=0.82)

      Experimental — accuracy not yet measured
    • cites Clinton, Jackman, and Rivers 2004 source
    • cites Gelman and Hill 2007

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The usual procedure for the first identification strategy is to constrain the ideal points of two legislators (liberal and conservative) at arbitrary positions, such as -1 and +1.

    “In particular, the usual procedure is to constrain the ideal points of two legislators (liberal and conservative) at arbitrary positions, such as À1 and þ1 (see, e.g., Londregan 1999; Clinton, Jackman, and Rivers 2004).”
    Quote in the paperNot a substantial claim1 of 2 sources checked
    • cites Londregan 1999 source
    • cites Clinton, Jackman, and Rivers 2004 source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductcitation

    The samplers in both stages are run using two chains with as many iterations as necessary to ensure that all ideology parameters have an effective number of simulation draws of at least 200.

    “The samplers in both stages are run using two chains with as many iterations as necessary to ensure that all ideology parameters have an effective number of simulation draws (Gelman and Rubin 1992) of at least 200.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Gelman and Rubin 1992 source

      The citation is attached to “effective number of simulation draws.” Section 2.2 (p. 5) describes monitoring convergence with potential scale reduction, while Section 4.5 (pp. 12–13) uses 200 iterations in an example; neither passage defines an effective number of draws or recommends a threshold of 200. Searches for “effective sample size” and “number of effectively independent simulations” found no passage establishing that metric. (LLM, conf=0.76)

      Experimental — accuracy not yet measured

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    The topology of the network of right-leaning Twitter users facilitates the rapid and broad dissemination of political information and increases their potential to manipulate the public agenda.

    “As Conover et al. (2012) argue, the topology of the network of right-leaning Twitter users facilitates the rapid and broad dissemination of political information, and increases their potential to manipulate the public agenda.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Conover et al. 2012 source

      The Introduction and Conclusion state that right-leaning users’ network topology facilitates the rapid and broad dissemination of political information. Section 5.3 also suggests more opportunities for political frame-making, but the paper does not say that this increases their potential to manipulate the public agenda. (LLM, conf=0.85)

      Experimental — accuracy not yet measured

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    How individuals gather political information affects the quality of political representation, the policy-making process, and the stability of the democratic system.

    “Given that a substantial proportion of citizens now rely mostly on the Internet to gather political information, 21 the policy implications of this issue are obvious: how individuals gather political information affects the quality of political representation, the policy-making process, and the stability of the democratic system (Mutz 2002).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Mutz 2002 source

      The Discussion (page not specified) reports modest consequences of cross-cutting political communication networks for political tolerance and democratic legitimacy. This supports the general idea that political communication matters for democracy, but the paper does not establish effects on the quality of political representation, policy-making, or the stability of the democratic system. (LLM, conf=0.83)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingown data

    The method successfully classifies individuals who state their political preferences publicly and a sample of users matched with their party registration records.

    “The method is also able to successfully classify individuals who state their political preferences publicly and a sample of users matched with their party registration records.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: Twitter users prefer to follow politicians whose positions on the latent ideological dimension are similar to theirs.

  • May combine two claims.

    Findingown data

    Online behavior during the 2012 US presidential election campaign is clustered along ideological lines, supporting the echo-chamber theory and high levels of political polarization at the mass level.

    “To illustrate a potential use of these estimates, I examine the extent to which online behavior during the 2012 US presidential election campaign is clustered along ideological lines, finding support for the so-called”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: Twitter users prefer to follow politicians whose positions on the latent ideological dimension are similar to theirs.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingnone given

    This figure is an almost exact replication of Fig. 5 in Bafumi and Herron (2010).

    “In fact, this figure is an almost exact replication of Fig. 5 in Bafumi and Herron (2010).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Bafumi and Herron 2010 source

  • May combine two claims.

    Findingown data

    In Spain, the Socialist Party (PSOE) is located to the left of the Conservative Party (PP), with the Center Party (UPyD) between them, and the Communist Party to the left of all three, although overlapping with the PSOE.

    “In Spain, the Socialist Party (PSOE) is located to the left of the Conservative Party (PP), with the recently created Center Party (UPyD) between them, and the Communist Party to the left of all three, although overlapping with the PSOE.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    With few exceptions, all Twitter accounts from the same party are clustered together, and parties with similar ideologies are located in similar positions on the latent ideological dimension.

    “With few exceptions, all Twitter accounts from the same party are clustered together, and parties with similar ideologies are located in similar positions on the latent ideological dimension.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    A total of 2707 Twitter users were matched to campaign contribution records and 2462 to Ohio voter registration records.

    “A total of 2707 and 2462 Twitter users were successfully matched in each case, based on perfect nonduplicated matches of first and last name, and zip code or county.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Either of the two identification strategies can be applied in this case to resolve the indeterminacies, as shown in the supplementary materials.

    “As I show in the supplementary materials, either of these strategies can be applied in this case to resolve the indeterminacies.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Both distributions of ideal points for political actors and ordinary citizens are bimodal.

    “Four features stand out: both distributions are bimodal, liberal citizens represent a majority of the sample population, political actors are more polarized than mass voters, and the distribution of mass ideology has longer tails.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The most conservative voters in Ohio tend to consistently register as Republicans, and vice versa.

    “The panel on the right, which classifies voters according to their registration history (the number of times they registered as Democrat minus the number of times they registered as Republican), indicates that this is indeed the case.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The correlation between Twitter-based and roll-call-based ideal points is 0.941 in the House and 0.954 in the Senate.

    “The correlation between Twitter-and roll-call-based ideal points is ¼ 0:941 in the House and ¼ 0:954 in the Senate.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    In the Netherlands, all parties are accurately scaled, with the exception of the Socialist Party (SP) and the right-wing Party for Freedom (PVV), which appear slightly more centrist in the Twitter estimates than in the expert surveys.

    “In the Netherlands, all parties are accurately scaled, with the exception of the Socialist Party (SP) and the right-wing Party for Freedom (PVV), which appear slightly more centrist in my Twitter estimates than in the expert surveys.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    In the UK, the Labor Party and the Liberal Democrats are located to the left of the Conservative Party on average, but the Liberal Democrats are classified as right-wing, almost overlapping with the conservatives.

    “The results show the lower degree of accuracy in the UK: both the Labor Party and the Liberal Democrats are located to the left of the Conservative Party on average, but the latter is classified as right-wing, almost overlapping with the conservatives, perhaps indicating their status as coalition partners.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Within-party correlations between Twitter-based and roll-call-based ideal points are 0.546 for Republicans and 0.610 for Democrats.

    “Within-party correlations are also relatively high: ¼ 0:546 for Republicans, ¼ 0:610 for Democrats.17”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    In Germany, the locations of the three right-wing parties (CDU, CSU, and FDP) on the ideological dimension are indistinguishable, which is consistent with the experts' assessments, but Die Linke appears to be more centrist.

    “In Germany, the locations of the three right-wing parties (CDU, CSU, and FDP) on the ideological dimension are indistinguishable, which is consistent with the experts' assessments, but Die Linke appears to be more centrist.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims.

    Findingnone given

    This minority includes virtually all political parties and candidates, public officials, interest groups, media outlets, journalists, and many others.

    “Of course, Twitter users are a self-selected minority of the population, but this minority includes virtually all political parties and candidates, public officials, interest groups, media outlets, journalists, and many others for whom we do not necessarily have good measures of ideology.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Right-leaning users form a cluster of highly motivated individuals who dominate public conversations.

    “Using the 2012 US presidential election campaign as a case study, I have shown that public exchanges on Twitter take place predominantly among users with similar viewpoints, and also that right-leaning users form a cluster of highly motivated individuals, who dominate public conversations.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingnone given

    The sample of users includes 3.5% of all unique users in the collection of tweets but sent 23% of all tweets.

    “Note that although this sample of users includes 3.5% of all unique users in the collection of tweets, they sent 23% of all tweets, and thus are a good representation of the political discussion that takes place on Twitter.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    It is possible to identify the location of most Twitter users, as well as their gender, and to track their behavior over time.

    “Furthermore, as I have shown in this article, it is possible to identify the location of most Twitter users, as well as their gender, and to track their behavior over time, which could help recover the representativeness of this sample of individuals.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    The analysis focuses on 15 million tweets sent by 301,537 individuals in the sample of politically interested users.

    “In the analysis below, I focus on the 15 million tweets that were sent by the 301,537 individuals in my sample of politically interested users.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingnone given

    The correlation between the two sets of estimates is 0.999 in all examples.

    “In all the examples in this article, the correlation between these two sets of estimates is ¼ :999.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Users with ideal points between -0.5 and 0.5 represent almost 40% of the sample, but sent only 21% of the tweets.

    “Moderate individuals are less active than those with more extreme ideological positions, even if they are a majority of the population: users with ideal points between À0.5 and 0.5 represent almost 40% of the sample considered in the study, but they sent only 21% of the tweets included in the collection.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Findingnone given

    Twitter users are a self-selected minority of the population.

    “Of course, Twitter users are a self-selected minority of the population, but this minority includes virtually all political parties and candidates, public officials, interest groups, media outlets, journalists, and many others for whom we do not necessarily have good measures of ideology.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Users with ideal points between 0.5 and 1.5 sent an average of 118 tweets, whereas those between -1.5 and -0.5 sent 82 tweets on average.

    “Overall, conservative users are more active: users with ideal points between 0.5 and 1.5 sent an average of 118 tweets, whereas those in the interval between À1.5 and À0.5 sent 82 tweets on average.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationreasoning

    Following decisions create opportunity costs because they reduce the likelihood of being exposed to other messages, assuming constant time on Twitter.

    “Second, given the fast-paced nature of Twitter, it also creates opportunity costs, since it reduces the likelihood of being exposed to other messages, assuming the amount of time a user spends on Twitter is constant.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: The amount of time a user spends on Twitter is constant.

  • May combine two claims.

    Premisecitation

    Twitter is not only an online social network but also a news media.

    “However, Twitter is not only an online social network-it is also a news media (Kwak et al. 2010).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Kwak et al. 2010 source

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims.

    Interpretationnone given

    The structure of the Twitter network is far from static, which can facilitate the estimation of highly granular dynamic ideal points in real time.

    “Second, the structure of this network is far from static, which can facilitate the estimation of highly granular dynamic ideal points in real time.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    The assumption behind this model is similar in nature to that justifying how donations and roll call votes can be scaled onto a latent ideological dimension.

    “Although obviously less costly than campaign contributions or votes in a legislature, the assumption behind this model is similar in nature to that justifying how donations and roll call votes can be scaled onto a latent ideological dimension (Poole and Rosenthal 2007; Bonica 2014).”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites Poole and Rosenthal 2007

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Bonica 2014 source

  • May combine two claims. The quote only partly states this claim.

    Interpretationnone given

    The method allows estimating ideology for more actors than any existing alternative, at any point in time and across many polities, and for a larger set of political actors and individuals than any other method before.

    “This method allows us to estimate ideology for more actors than any existing alternative, at any point in time and across many polities.”
    • Also stated in: § 1 Introduction · plain
      “This method represents an additional measurement tool that can be used to estimate ideology, an important quantity of interest in political science, for a larger set of political actors and individuals than any other method before.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: Twitter users prefer to follow politicians whose positions on the latent ideological dimension are similar to theirs.

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation

    Individuals exhibit a preference for opinion-reinforcing political information and systematically avoid opinion challenges.

    “the existing literature on the selective exposure theory (Lazarsfeld, Berelson, and Gaudet 1944; Bryant and Miron 2004; Stroud 2008; Iyengar and Hahn 2009) that argues that individuals exhibit a preference for opinion-reinforcing political information and that they systematically avoid opinion challenges.”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites Lazarsfeld, Berelson, and Gaudet 1944

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Bryant and Miron 2004 source
    • cites Stroud 2008 source

  • May combine two claims.

    Interpretationreasoning

    Valid policy positions and ideal point estimates for ordinary users and political actors can be inferred or derived from the structure of the following links across these two sets of Twitter users.

    “I argue that valid policy positions for ordinary users and political actors can be inferred from the structure of the”
    • Also stated in: § 2.1 Assumptions · boosted
      “In this article, I demonstrate that valid ideal point estimates of individual Twitter users and political actors with a Twitter account can be derived from the structure of the”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: Twitter users prefer to follow politicians whose positions on the latent ideological dimension are similar to theirs.

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationnone given

    Following decisions can create cognitive dissonance if the content users are exposed to challenges their political views.

    “If the content of the messages users are exposed to as a result of their following decisions challenges their political views, it can create cognitive dissonance.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Interpretationnone given

    It is possible to link Twitter profiles to other data through name identification, which provides ways to examine differences between private and public political behavior.

    “Third, it is possible to link Twitter profiles to other data through name identification, which provides interesting ways to examine differences between private and public political behavior.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    When analyses include voters, they do so at the expense of strong bridging assumptions or only for self-selected population groups.

    “When the analysis also includes voters, it is done at the expense of strong”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites Jessee 2009 source
    • cites Bonica 2014 source

  • Classified as probably not a substantial claim.

    Premisereasoning

    The choice of the m target Twitter users who are political elites is a key challenge in implementing this method.

    “A key challenge in implementing this method is the choice of the m target Twitter users who are political elites: the set of users with "discriminatory" predictive power such that the decision to follow them (or not) provides information about an individual's ideology.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims.

    Interpretationreasoning

    The self-selection of the sample should not affect the inference of politicians' ideal points, because these users can be considered authoritative when it comes to politics.

    “However, this should not affect the inference of politicians' ideal points, since these users can indeed be considered as "authoritative" when it comes to politics.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: Users who follow three or more political accounts are more likely to be knowledgeable and interested in politics than the average citizen.

  • May combine two claims.

    Interpretationcitation

    The reason the model performs well in the US case is probably that economic and social left-right positions are highly correlated and policy preferences are mostly unidimensional.

    “This is probably the reason why the model performs well in the US case, where economic and social left-right positions are highly correlated and policy preferences are mostly unidimensional (Poole and Rosenthal 2007).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Poole and Rosenthal 2007

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationown data

    The positions of key political actors within each cluster are what we would expect based on anecdotal evidence.

    “Note also that their positions within each cluster are what we would expect based on anecdotal evidence.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductcitation

    The sample is highly self-selected because Twitter users are not a representative sample of the population.

    “This is a highly self-selected sample because Twitter users are not a representative sample of the population.14”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Individuals are being increasingly exposed to only information that reinforces their existing views, avoiding challenging opinions.

    “As a result of this transformation, individuals are being increasingly exposed to only information that reinforces their existing views, thus avoiding challenging opinions (Sunstein 2001; Garrett 2009; Iyengar and Hahn 2009).”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites Sunstein 2001
    • cites Garrett 2009 source
    • cites Iyengar and Hahn 2009 source

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationreasoning

    The procedure is roughly analogous to an expert survey with many respondents where each respondent provides a small amount of information that, when aggregated, results in highly accurate policy estimates.

    “This procedure is roughly analogous to an expert survey with many respondents where each respondent provides a small amount of information that, when aggregated, results in highly accurate policy estimates.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Premisenone given

    Samples from the posterior density of each parameter in the model can be obtained using Markov Chain Monte Carlo methods.

    “However, samples from the posterior density of each parameter in the model can be obtained using Markov Chain Monte Carlo methods.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Interpretationreasoning

    Each parameter indexed by i can be estimated individually because of local independence conditional on the j parameters, allowing multi-core processors to run multiple samplers simultaneously and dramatically increase computation speed.

    “Note that each of these parameters can be estimated individually because I assume local independence, conditional on the j parameters, and therefore multi-core processors can be used to run multiple samplers simultaneously and dramatically increase computation speed.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: Individual decisions to follow are independent across users n and m, conditional on the estimated parameters (local independence).

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Study design and conductnone given

    The model-fitting strategy appears quite robust, and the results are largely insensitive to the choice of priors and initial values.

    “This modelfitting strategy appears to be quite robust, and my results are largely insensitive to the choice of priors and initial values.foot_3”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The increasing amount and heterogeneity of political information citizens have access to affects their political views.

    “A recurring theme in the literature on the Internet and politics is how the increasing amount and heterogeneity of political information citizens have access to affects their political views (Farrell 2012).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Farrell 2012 source

  • May combine two claims.

    Interpretationcitation

    Using a Hamiltonian Monte Carlo algorithm and a hierarchical setup improves the efficiency of the estimation procedure.

    “To improve the efficiency of this procedure, I use a Hamiltonian Monte Carlo algorithm (Gelman et al. 2013) and employ a hierarchical setup that considers each of the four sets of parameters as draws from four common population distributions:”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Gelman et al. 2013 source

  • May combine two claims.

    Premisecitation

    In the US, economic and social left-right positions are highly correlated and policy preferences are mostly unidimensional.

    “This is probably the reason why the model performs well in the US case, where economic and social left-right positions are highly correlated and policy preferences are mostly unidimensional (Poole and Rosenthal 2007).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Poole and Rosenthal 2007

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationnone given

    The set of target users considered will determine the interpretation of the latent scale where ideal points are located.

    “The set of users that are considered will determine the interpretation of the latent scale where ideal points are located and, for this reason, it is important to include identifiable figures with extreme ideological positions, beyond just partisans.8”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationnone given

    The estimation method can be applied to any country where a high number of citizens are discussing politics on Twitter.

    “The estimation method I propose in this article can be applied to any country where a high number of citizens are discussing politics on Twitter.9”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Study design and conductreasoning

    The results for different countries are not directly comparable, because the estimation was performed independently and the resulting dimension does not have a homogeneous scale across countries.

    “It is important to note that the results for different countries are not directly comparable, as the estimation was performed independently, and the resulting dimension does not have a homogeneous scale across countries.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    Samples from the i parameters in the second stage can be compared with those from the random sample in the first stage to ensure no errors in estimation.

    “Samples from the i parameters in the second stage can be compared with those obtained for the random sample in the first stage to ensure that there were no errors in the estimation.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisenone given

    In the US, partisanship and ideology are highly correlated.

    “In the United States, where partisanship and ideology are highly correlated, it is not as clear.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning

    Twitter data has the potential to generate much value added to explanatory models of political behavior at a low cost.

    “This new source of information has the potential to generate much value added to explanatory models of political behavior at a low cost.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Bafumi and Herron (2010) estimated ideology for the mass public using the Cooperative Congressional Election Study (N=33,000) with questions asking respondents to take positions on roll call votes as if they were members of Congress.

    “Bafumi and Herron (2010) estimate ideology for the mass public using the Cooperative Congressional Election Study (N ¼ 33,000), which included questions asking respondents to take positions on roll call votes as if they were members of Congress.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Bafumi and Herron 2010 source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisenone given

    The model is agnostic regarding the interpretation of the latent dimension, which depends on the set of m political actors considered.

    “Note that the model is agnostic regarding the interpretation of the latent dimension, which will depend on the set of m political actors that are considered.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    There is a high degree of polarization on Twitter, driven predominantly by conservative Twitter users.

    “To sum up, the picture that emerges points toward a high degree of polarization, which is driven predominantly by conservative Twitter users.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: Users are more likely to retweet someone they follow.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisenone given

    If two dimensions are orthogonal, the estimated latent space corresponds to the dimension that explains a larger proportion of the variance.

    “Note that this should not affect the estimation as long as these two dimensions are orthogonal to each other, in which case the estimated latent space would correspond to the dimension that explains a larger proportion of the variance.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    4.3 million US citizens made a political contribution in the 2012 electoral cycle.

    “In all, 4.3 million US citizens made a political contribution in the 2012 electoral cycle (Bonica 2013a), but 15% of online adults (almost forty million citizens) are active on Twitter.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Bonica 2013a

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    15% of online adults (almost forty million citizens) are active on Twitter.

    “In all, 4.3 million US citizens made a political contribution in the 2012 electoral cycle (Bonica 2013a), but 15% of online adults (almost forty million citizens) are active on Twitter.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Pew Research Center 2013

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • May combine two claims.

    Study design and conductown data

    The method developed in this article allows computing ideal points for a far larger sample of users than previous studies, including most active Twitter users in the US, with higher precision.

    “By using the method developed in this article, I am able to compute ideal points for a far larger sample of users than any of the previous studies, which now includes most active Twitter users in the United States, and with a much higher degree of precision.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    A substantial proportion of citizens now rely mostly on the Internet to gather political information.

    “Given that a substantial proportion of citizens now rely mostly on the Internet to gather political information, 21 the policy implications of this issue are obvious: how individuals gather political information affects the quality of political representation, the policy-making process, and the stability of the democratic system (Mutz 2002).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Pew Research Center 2011

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    This generates an echo-chamber environment that fosters social extremism and political polarization.

    “This generates a so-called echo-chamber environment that fosters social extremism and political polarization.”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites Sunstein 2001
    • cites Garrett 2009 source
    • cites Iyengar and Hahn 2009 source
68 of 68

  • The authors state this as an assumption.

    Twitter users prefer to follow politicians whose positions on the latent ideological dimension are similar to theirs.

    “The key assumption of this model is that Twitter users prefer to follow politicians whose positions on the latent ideological dimension are similar to theirs.”

  • The paper relies on this without support.

    Most studies estimate ideal points for legislators only.

    “However, most studies estimate ideal points for legislators only.”

  • The paper relies on this without support.

    Citizens who discuss politics on Twitter are more likely to be educated and politically interested, and that makes them a particularly useful source of information about elites' ideology.

    “Citizens who discuss politics on Twitter are more likely to be educated and politically interested, and that makes them a particularly useful source of information about elites' ideology.”

  • The authors state this as an assumption.

    Following decisions are costly signals that provide information about Twitter users' perceptions of both their ideological location and that of political accounts.

    “I consider following decisions to be costly signals about users' perceptions of both their ideological location and that of political accounts.”

  • The paper relies on this without support.

    Virtually every legislator, political party, and candidate in developed democracies has an active Twitter account.

    “Virtually every legislator, political party, and candidate in developed democracies has an active Twitter account.”

  • The paper relies on this without support.

    Twitter users are embedded in a common social network.

    “Most importantly, they are embedded in a common social network.”

  • The authors state this as an assumption.

    The amount of time a user spends on Twitter is constant.

    “assuming the amount of time a user spends on Twitter is constant”

  • The authors state this as an assumption.

    Individual decisions to follow are independent across users n and m, conditional on the estimated parameters (local independence).

    “Assuming local independence (individual decisions to follow are independent across users n and m, conditional on the estimated parameters), the likelihood function to maximize this model is as follows:”

  • The authors state this as an assumption.

    The decision to follow a political actor is a function of the squared Euclidean distance in the latent ideological dimension between the user and the actor.

    “For the reasons explained above, I expect this decision to be a function of the squared Euclidean distance in the latent ideological dimension 5 between user i and j: À jj i À j jj 2”

  • The paper relies on this without support.

    Differences in the number of political accounts each user follows are related to the overall number of Twitter users they follow or their overall level of interest in politics.

    “Similarly, this parameter accounts for the differences in the number of political accounts each user i decides to follow, which could be related to the overall number of Twitter users they follow, or their overall level of interest in politics.”

  • The paper relies on this without support.

    Homophilic networks can be based not only on political traits, but also on other personal characteristics.

    “However, "homophilic" networks can be based not only on political traits, but also on other personal characteristics.”

  • The paper relies on this without support.

    High-quality ideology measures are available for a subset of all Twitter users in the US, UK, Spain, Germany, Italy, and the Netherlands.

    “However, in order to test the validity of the estimated parameters, I will focus on six countries where high-quality ideology measures are available for a subset of all Twitter users: the US, the UK, Spain, Germany, Italy, and the Netherlands.”

  • The paper relies on this without support.

    Users who follow three or more political accounts are more likely to be knowledgeable and interested in politics than the average citizen.

    “Precisely because they are more likely to be knowledgeable and interested in politics than the average citizen”
    “these users can indeed be considered as "authoritative" when it comes to politics.”

  • The paper relies on this without support.

    Many Twitter users define themselves politically in their profiles.

    “Here, I exploit the fact that many Twitter users define themselves politically in their profiles.”

  • The authors state this as an assumption.

    Users are more likely to retweet someone they follow.

    “It may be argued that this conclusion is driven at least in part by the key assumption behind the method to estimate ideology, since users are more likely to retweet someone they follow.”

  • The paper relies on this without support.

    As citizens spend a greater proportion of their time online, most of their political behavior also takes place through social networking platforms.

    “As citizens spend a greater proportion of their time online, most of their political behavior also takes place through social networking platforms.”

  • The paper relies on this without support.

    Most existing work relies on crude measures of political preferences.

    “Being able to estimate covariates of interest represents an important step in the study of such behavior, since most existing work relies on crude measures of political preferences.”
17 of 17

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • The methodology of Barberá (2015) measures users' political orientation.

    “We employ the methodology of Barberá (2015) to measure users' political orientation.”

    From Cross-Partisan Interactions on Twitter (2025) · cites this paper as Barberá 2015

  • Homophily in communication is a key tendency of users in information propagation media like Twitter and other platforms: people tend to communicate with others who hold similar beliefs.

    “Homophily in communication is a key tendency of users in information propagation media like Twitter and other platforms: people tend to communicate with others who hold similar beliefs (Barberá, 2015;Conover et al., 2011).”

    From Credibility of climate change denial in social media (2019) · cites this paper as Barberá, 2015

Cites

Flagged first, then by citations

Research works printed but not linked: 2

  1. It may be argued that this conclusion is driven at least in part by the key assumption behind the method to estimate ideology, since users are more likely to retweet someone they follow. However, note that retweets between users who do not follow each other are indeed frequent. For example, of the 460,000 users who retweeted at least one of Barack Obama's tweets between August 15, 2012, and November 6, 2012, 152,000 (33%) did not follow him as of November 5, 2012. In fact, 21.2 of his 21.5 million followers at the time did not retweet any of his tweets during this period.
  2. King A. , Orlando F. , and Sparks D. B. 2011. Ideological extremity and primary success: A social network approach. Chicago, Illinois: Paper presented at the 2011 MPSA Conference.

Cited by

Sorted by citations

Related

Authors
Barberá, P.
Affiliations
New York University (US)
Corresponding author
Barberá, Pablo
ORCID
Barberá, Pablo
Venue
Political Analysis
Publisher
Cambridge University Press (CUP)
Published
2015
Volume
23
Issue
1
Pages
76-91
ISSN
  • 1047-1987 (ISSN-L)
  • 1476-4989
DOI
10.1093/pan/mpu011
OpenAlex
W2161834943
Semantic Scholar
119306396df0ec0c751aca034c11a36c45338dde
Cited by
982
References
86
Citations by year
2026: 46 · 2025: 89 · 2024: 80 · 2023: 100 · 2022: 121 · 2021: 87 · 2020: 133 · 2019: 81 · 2018: 74 · 2017: 82 · 2016: 48 · 2015: 26 · 2014: 12 · 2013: 3
FWCI
165.45
Type
journal article
Language
en
Open access
Licence
www.cambridge.org/core/terms (unspecified, 734 d)
Locations
  • Political Analysis (publishedVersion)
  • SSRN Electronic Journal (acceptedVersion)
  • London School of Economics and Political Science Research Online (London School of Economics and Political Science) (acceptedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
Funders
National Science Foundation (1248077)
Topics
  • Social Sciences › Communication › Social Media and Politics
  • Social Sciences › Political Science and International Relations › Electoral Systems and Political Participation
  • Social Sciences › General Social Sciences › Computational and Text Analysis Methods
Keywords
Twitter, ideal points, political debate, legislators, political parties, party registration, 2012 presidential election, online political behavior, latent variable models

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-13.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-13.

Cited sources are checked by gpt-6-sol.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review