A Long-Term Analysis of Polarization on Twitter

Kiran Garimella, Ingmar Weber · ICWSM / arXiv · 2017 · political-science · record garimella2017-polarization
tag layer · index · concepts · authors · עברית

Where this record's text came from. faceted 2026-08 by reading the locally held text (pdftext); every evidence string verified as an exact substring of that text before merging. Short ICWSM pre-print (2.6k words), which is why the record is comparatively sparse. Claim type is 'existence' — the paper's contribution is the demonstration that an increase occurred and its magnitude, not a mechanism for it; the authors explicitly decline to adjudicate the causal debate. The 'method' vocabulary supplied does not include a term for large-scale observational social-media data analysis, so 'observational-data-analysis' was coined; the design is longitudinal and correlational, with statistical testing (augmented Dickey-Fuller, t-test on a linear slope) but no experiment or simulation. The paper reports a significance test on the hashtag time series only; no reliability or robustness statistics are given for the follow/retweet measures. 2026-08-14 and 2026-08-15 schema-0.3 passes: every quote verified as an exact substring of the staged text and anchored at merge time; concept terms verified present; every edge endpoint an existing node. Update, 16 September 2026: EXCEPTION TO THE DATED PASS ABOVE, on this record only: this record carries 1 composite quote field - a quote assembled from two or more passages joined by an ellipsis. A COMPOSITE IS NOT AN EXACT CONTIGUOUS SUBSTRING of the staged text and is deliberately not anchored: each of its parts is verified separately at the offsets in evidence composite spans, and a part occurring more than once carries span null rather than a guessed offset. So the 'every quote verified as an exact substring' clause of the dated pass above DOES NOT HOLD for those fields, and never did - the parts were verified, the whole was not. Nothing is deleted, re-grounded or withdrawn: this writes down an exception that was always true of composites and had not been recorded, so a reader was told more than the pass had checked. 2026-09-16 the project CORRECTION TO THE EXEMPTION ABOVE, SAME DAY: THE EXEMPTION DOES NOT APPLY TO THIS RECORD. Its evidence composite spans row(s) are FOSSILS of a 2026-08-25 composite marking that was superseded by a 2026-08-26 re-slice; the quote is an exact contiguous substring of the staged text, so the dated substring claim above holds for it WITHOUT exception. The clause is left in place with this correction beside it rather than removed, so that the error and its repair are both on the record.

1 quote on this page was cut from a different copy of this article. This project holds two extractions of this paper: the plain text every other quote here is anchored in, and the publisher’s own HTML edition, which we are not licensed to republish. It is verbatim in the second and cannot be found in the first — not because the tagging is wrong, but because the extraction damaged the passage: a page number dropped into the middle of a number, a formula moved out of its sentence, a ligature where two letters belong. The record names the copy for each one. The quote is the paper’s; the extraction is not.

How to read this page. The article itself is not reproduced here — this paper is not cleared for redistribution, so what follows is the tag layer alone. The evidence sentences are withheld with the article. The table at the foot lists every tag the record carries; each row names the facet it belongs to, and the whose words column says whether the tag records the paper's own wording or the tagger's sentence about the paper. Every tag links to each other paper in the corpus carrying it. The machine-readable edition of the same thing is the TEI file beside this one, where the tags live in standOff, and the article is at its publisher via the link below. When we have changed our mind. Some entries in the table begin with a word in capitals and the reason for it. EVIDENCE CORRECTED — the finding still stands, but we had quoted the wrong sentence from the paper, and the quotation has been replaced with the right one. RETRACTED — the entry itself was wrong; it stays on the page, marked, with the reason, rather than being deleted. WITHDRAWN AND REPLACED — the entry was wrong, and a corrected version of it appears in the entries below. Each mark carries its own reason, in brackets, because the reasons differ from one entry to the next. An entry we have WITHDRAWN is removed from this page and kept in the record, with its reason. A line reading Audit and a date is a later re-reading of the entry against its whole sentence: DOUBTFUL means it was kept, and the reason says what is loose about it. Quotation marks inside the tagger's sentence. Where a tag is the tagger's own sentence, a phrase in quotation marks inside it is borrowed from the paper — the tagger's wording around it is not. The paper's own sentence is the evidence quote for that row; on this page it is withheld with the article, and the article is at its publisher. Nothing inside quotation marks anywhere on this page is ours: if it is quoted, it is the paper's. Whose words a concept carries. A concept entry records a term the paper itself uses, so the term, the sense given for it, and its evidence are the paper's own words, quoted where the paper fixes them. This is the opposite of a sense entry in the tag table, which is the tagger's sentence about how the paper uses a term. Where a concept entry could not be grounded in the paper's words, the entry says so on its own row.

Other fields define these differently the same word, defined another way in another discipline
echo chamber 4 fields · social media 4 fields · polarization 2 fields

Tags on this paper

Every tag the record carries, each linking to every other paper in the corpus that carries it. A tag appears here whether or not its words occur in the article — the spans below are only the ones a sentence in this paper licenses. Of the 8 concept tags on this paper, 5 are grounded in a quoted sentence.

political polarization echo chamber social media network analysis polarization leaning hashtag polarization the time a user began following is ESTIMATED, from the reverse-chronological order the API returns followers in, lower-bounded by each follower's account creation date each user's leaning is a beta-binomial coin-bias estimate - a uniform prior over the two sides, then one increment per follow or retweet polarization is defined as DISTANCE FROM BALANCE rather than as leaning, so a strongly left and a strongly right user score identically a hashtag inherits its leaning from the users who used it that week, and users who did not retweet a seed account that week contribute nothing two explicit noise controls chosen because the early years are thin: a minimum of five users per hashtag, and a five-week moving average automated and bad-faith accounts are handled BY THE PASSAGE OF TIME rather than by detection - the study relies on Twitter having suspended them in the years before collection Twitter users political retweet hashtags network partisan content Bayesian mean mean of p-value t-test observational-study

Method, as described: observational-data-analysis

Legend

relation to other work	`claim_relations`	how this paper stands to another: agrees, contradicts, extends, qualifies
claim type	`claim_type`	the shape of the claim: existence, mechanism, method, critique, prediction, null
concept edge	`concept_edges`	an asserted relation between two concepts, with the sentence asserting it
concept	`concepts`	an idea the paper works with, quoted where the paper uses it
content tag	`content_tags`	a finding, definition, quantity, statistic or variable named in the paper
control	`controls`	an experiment run to eliminate a rival explanation
limitation	`limitations`	what the paper says its evidence cannot support
mechanism	`mechanisms`	how the reported effect is said to come about
method	`method`	how the paper produced its evidence
phenomenon	`phenomenon`	the effect or regularity under study
sense	`senses`	a distinct use of a term, quoted where the paper fixes it
transfer	`transfers`	a method or concept carried in from another field

The paper

The full text is not reproduced here. This article is not cleared for redistribution, so the page carries the tag layer only. Its evidence sentences are withheld; the grounding status remains with each tag, and the article itself is at https://arxiv.org/abs/1703.02769. The reading view exists for the papers whose licence permits it.

Every tag on this page

claim_relations 4 · claim_type 1 · concept_edges 2 · concepts 5 · content_tags 10 · controls 1 · limitations 9 · mechanisms 6 · method 1 · phenomenon 1 · senses 4 · transfers 1

Relations on this page carry a quotation for the claim. Where a relation was separately evidenced — and so tested against the vocabulary for its type — the row says so; none of 2 relations here carry that mark. The rest are the tagger’s reading of the sentence quoted beside them.

facet	tag	field	conf.	whose words
claim_type	existence Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`claim_type`		the paper's words
phenomenon	political polarization on Twitter Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`phenomenon`		the paper's words
method	observational-study Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: observational-study	`method`		the paper's words
mechanisms	mechanisms Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: the time a user began following is ESTIMATED, from the reverse-chronological order the API returns followers in, lower-bounded by each follower's account creation date	`mechanisms[0]`	0.9	the tagger's words
mechanisms	the paper's REASON for the Bayesian treatment - that a single act from one side is a weak signal - is the sentence immediately before this one, and widening to reach it pulls in a footnote the extraction has interleaved into the body ('4Since the Twitter API restricts us to the last 3200 tweets'), so the tag is trimmed instead Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: each user's leaning is a beta-binomial coin-bias estimate - a uniform prior over the two sides, then one increment per follow or retweet	`mechanisms[1]`	0.9	the tagger's words
mechanisms	mechanisms Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: polarization is defined as DISTANCE FROM BALANCE rather than as leaning, so a strongly left and a strongly right user score identically	`mechanisms[2]`	0.9	the tagger's words
mechanisms	mechanisms Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: a hashtag inherits its leaning from the users who used it that week, and users who did not retweet a seed account that week contribute nothing	`mechanisms[3]`	0.9	the tagger's words
mechanisms	mechanisms Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: two explicit noise controls chosen because the early years are thin: a minimum of five users per hashtag, and a five-week moving average	`mechanisms[4]`	0.9	the tagger's words
mechanisms	mechanisms Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: automated and bad-faith accounts are handled BY THE PASSAGE OF TIME rather than by detection - the study relies on Twitter having suspended them in the years before collection	`mechanisms[5]`	0.9	the tagger's words
limitations	Twitter API 'most recent activity only' limitations hinder longitudinal study Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. what this limits: any longitudinal claim: API LIMITATIONS HINDER SUCH STUDIES - only most-recent activity is retrievable	`limitations[0]`		the tagger's words
limitations	Automated tweets and astroturfing were not detected or removed. Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. what this limits: the authenticity of the measured activity: WE DID NOT EXPLICITLY ATTEMPT TO DETECT ASTROTURFING AND OTHER TYPES OF AUTOMATED TWEETS	`limitations[1]`		the tagger's words
limitations	Hashtag hijacking could still contaminate the content measure. Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. what this limits: the content measure, with the boundary decision stated: HASHTAG HIJACKING COULD STILL AFFECT OUR ANALYSIS, THOUGH THESE EFFECTS ARE ARGUABLY ALSO PART OF THE POLITICAL LANDSCAPE AND SHOULD BE INCLUDED	`limitations[2]`		the tagger's words
limitations	The study cannot settle whether social media causes, reflects or reduces offline polarization. Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. what this limits: any causal reading: the study cannot settle whether social media causes, reflects or reduces offline polarization	`limitations[3]`		the tagger's words
limitations	The election-period pattern in hashtag polarization does not hold for the 2010 midterms, probably because of noisier early estimates. Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. what this limits: the election-period pattern: FOR THE 2010 MIDTERM ELECTIONS, HOWEVER, THIS OBSERVATION DOES NOT HOLD, POTENTIALLY DUE TO NOISIER ESTIMATES BASED ON FEWER ACTIVE USERS	`limitations[4]`		the tagger's words
limitations	Pre-September-2009 tweets were excluded because their volume was too low. Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. what this limits: the temporal coverage of the analysis: WE ONLY CONSIDER TWEETS FROM SEPTEMBER 2009 ONWARDS ... SINCE THE VOLUME FOR EARLIER TWEETS IS LOW	`limitations[5]`		the tagger's words
limitations	The follow times underlying the whole time series are estimated rather than observed, and the accuracy attributed to that estimator is reported second-hand and only for accounts with millions of followers. Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. what this limits: the year-by-year following curve: follow dates are estimated by the method of Meeder et al., whose accuracy is reported to hold for users with millions of followers, while the analysis uses all cases with estimated follow dates from January 2009 onwards	`limitations[6]`	1.0	the tagger's words
limitations	The content measure is not specified in this paper; its method is applied from another work and the reader is sent there for the details. Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. what this limits: the hashtag-polarization series: the procedure that produces it - leaning assignment, volume correction and smoothing - is taken from another paper, and this one refers the reader there for details	`limitations[7]`	1.0	the tagger's words
limitations	Each user's polarization value is a posterior belief that begins from an assumed balance between the two sides, not a summary of that user's observed behaviour alone. Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. what this limits: a per-user polarization value: it is the mean of a beta posterior that starts from a uniform prior over left and right and is updated by each observed follow or retweet, so a user with few observed actions is scored largely by the prior	`limitations[8]`	1.0	the tagger's words
transfers	social-network-structure analysis applied to measure political polarization on Twitter Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`transfers[0]`		the tagger's words
controls	three separate questions, each answered from A DIFFERENT TYPE OF INFORMATION, rather than one measure carried through Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. eliminates: a single data source's idiosyncrasies standing in for the phenomenon - polarisation measured only through, say, retweets is a fact about retweeting until a second and third instrument agree · outcome (stated by the tagger, no quotation): held	`controls[0]`	0.9	the tagger's words
concepts	echo chamber Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: echo chamber	`concepts[1]`		the paper's words
concepts	polarization Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: polarization	`concepts[4]`		the paper's words
concepts	leaning Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: leaning	`concepts[5]`		the paper's words
concepts	hashtag polarization Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: hashtag polarization	`concepts[6]`		the paper's words
concepts	polarization Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: polarization	`concepts[7]`		the paper's words
senses	polarization - STIPULATED AS RESTRICTION, not as hostility or extremity: a tendency to be restricted to one side of the spectrum when obtaining or engaging with political information Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[0]`	1.0	the tagger's words
senses	polarization - AS A NUMBER: twice the absolute deviation of a beta-posterior leaning from 0.5, so it measures distance from a balanced diet on a 0.0 to 1.0 scale and nothing else Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[1]`	1.0	the tagger's words
senses	polarization - AS LEXICAL DIVERGENCE: two sides that describe the same thing in completely different words, the paper's own example being global warming versus climategate Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[2]`	1.0	the tagger's words
senses	the term's scope - FIXED BY STIPULATION IN A FOOTNOTE: every later use of the bare word is declared to mean political polarization Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[3]`	1.0	the tagger's words
claim_relations	builds-on → the hashtag-leaning methodology of [15] Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`claim_relations[0]`		the tagger's words
claim_relations	builds-on → Meeder et al. [10] follow-time estimation method Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`claim_relations[1]`		the tagger's words
claim_relations	extends → Andris et al. [2] (partisanship of the US Congress) Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`claim_relations[2]`		the tagger's words
claim_relations	extends → Liu et al. [9] (seven-year longitudinal Twitter analysis) Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`claim_relations[3]`		the tagger's words
concept_edges	polarization —operationalised-as→ leaning Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. strength of the relation: 1 also in: polarization · leaning	`concept_edges[0]`		the tagger's words
concept_edges	hashtag polarization —is-a→ polarization Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. strength of the relation: 1 Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but endpoint by anaphora: the section is headed 'Hashtag Polarization'; the span writes 'polarization of hashtags'. also in: hashtag polarization · polarization	`concept_edges[1]`		the tagger's words
content_tags	679,000 — size of the longitudinal Twitter dataset Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`content_tags.quantities[0]`		the tagger's words
content_tags	10%-20% — relative increase in online polarization over eight years, depending on measure Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`content_tags.quantities[1]`		the tagger's words
content_tags	eight years — span of the longitudinal analysis (two presidential + two mid-term elections) Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. about this quote: QUOTE REPAIRED: the extractor split a hyphen-space and the quote was recorded faithfully with the damage in it, which made it unanchorable. The repaired string is an EXACT SUBSTRING of the staged text - that is why it was accepted, not because a rule produced it. The text as originally recorded is kept in evidence as extracted.	`content_tags.quantities[2]`		the tagger's words
content_tags	> 60% — share of Americans who get news from social media (Pew) Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`content_tags.quantities[3]`		the tagger's words
content_tags	Online political polarization increased over the past eight years, a relative change of 10-20% depending on the measure. Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`content_tags.findings[0]`		the tagger's words
content_tags	echo chamber Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: echo chamber	`content_tags.definitions[0]`		the tagger's words
content_tags	Bayesian Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. about this quote: QUOTE REPAIRED: the extractor split a hyphen-space and the quote was recorded faithfully with the damage in it, which made it unanchorable. The repaired string is an EXACT SUBSTRING of the staged text - that is why it was accepted, not because a rule produced it. The text as originally recorded is kept in evidence as extracted. THE QUOTE CONTINUES A SENTENCE THAT THIS COPY CUTS IN TWO, AND ITS FIRST WORDS ARE NOT PART OF THAT SENTENCE - a footnote stands between it and the words it follows, which read 'not necessarily a strong signal', and the quote opens with the end of that footnote's split URL. also in: Bayesian	`content_tags.statistics[0]`		the tagger's words
content_tags	mean Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. about this quote: QUOTE REPAIRED: the extractor split a hyphen-space and the quote was recorded faithfully with the damage in it, which made it unanchorable. The repaired string is an EXACT SUBSTRING of the staged text - that is why it was accepted, not because a rule produced it. The text as originally recorded is kept in evidence as extracted. also in: mean	`content_tags.statistics[1]`		the tagger's words
content_tags	p-value Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: p-value	`content_tags.statistics[2]`		the tagger's words
content_tags	t-test Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. about this quote: QUOTE REPAIRED: the extractor split a hyphen-space and the quote was recorded faithfully with the damage in it, which made it unanchorable. The repaired string is an EXACT SUBSTRING of the staged text - that is why it was accepted, not because a rule produced it. The text as originally recorded is kept in evidence as extracted. also in: t-test	`content_tags.statistics[3]`		the tagger's words

Interpretive choices

Tags that could not be placed in the text:

none — every tag was located in the text

What was deliberately not tagged, and why, is recorded per paper by the tagger. Where a paper has no such record, this page says so rather than implying that nothing was declined.