{ "paper_id": "2021", "header": { "generated_with": "S2ORC 1.0.0", "date_generated": "2023-01-19T06:07:10.005424Z" }, "title": "Partisanship and Fear are Associated with Resistance to COVID-19 Directives", "authors": [ { "first": "Mike", "middle": [], "last": "Lindow", "suffix": "", "affiliation": { "laboratory": "", "institution": "University of Utah", "location": {} }, "email": "michael.lindow@eccles.utah.edu" }, { "first": "David", "middle": [], "last": "Defranza", "suffix": "", "affiliation": { "laboratory": "", "institution": "University of Utah", "location": {} }, "email": "david.defranza@eccles.utah.edu" }, { "first": "Arul", "middle": [], "last": "Mishra", "suffix": "", "affiliation": { "laboratory": "", "institution": "University of Utah", "location": {} }, "email": "arul.mishra@eccles.utah.edu" }, { "first": "Himanshu", "middle": [], "last": "Mishra", "suffix": "", "affiliation": { "laboratory": "", "institution": "University of Utah", "location": {} }, "email": "himanshu.mishra@eccles.utah.edu" } ], "year": "", "venue": null, "identifiers": {}, "abstract": "Ideological differences have had a large impact on individual and community response to the COVID-19 pandemic in the United States. Early behavioral research during the pandemic showed that conservatives were less likely to adhere to health directives, which contradicts a body of work suggesting that conservative ideology emphasizes a rule abiding, loss aversion, and prevention focus. We reconcile this contradiction by analyzing semantic content of local press releases, federal press releases, and localized tweets during the first month of the government response to COVID-19 in the United States. Controlling for factors such as COVID-19 confirmed cases and deaths, local economic indicators, and more, we find that online expressions of fear in conservative areas lead to an increase in adherence to public health recommendations concerning COVID-19, and that expressions of fear in government press releases are a significant predictor of expressed fear on Twitter.", "pdf_parse": { "paper_id": "2021", "_pdf_hash": "", "abstract": [ { "text": "Ideological differences have had a large impact on individual and community response to the COVID-19 pandemic in the United States. Early behavioral research during the pandemic showed that conservatives were less likely to adhere to health directives, which contradicts a body of work suggesting that conservative ideology emphasizes a rule abiding, loss aversion, and prevention focus. We reconcile this contradiction by analyzing semantic content of local press releases, federal press releases, and localized tweets during the first month of the government response to COVID-19 in the United States. Controlling for factors such as COVID-19 confirmed cases and deaths, local economic indicators, and more, we find that online expressions of fear in conservative areas lead to an increase in adherence to public health recommendations concerning COVID-19, and that expressions of fear in government press releases are a significant predictor of expressed fear on Twitter.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Abstract", "sec_num": null } ], "body_text": [ { "text": "Over the last decade, politics in the United States have become increasingly polarized (Balz, 2019; Dimock et al., 2014; Lockhart et al., 2020) . This phenomenon is manifest in the increasing politicization of historically non-partisan government agencies (Cooper, 2020; Mulgan, 2007; Peters, 2004) , the divide on Ebola preparedness (Nyhan, 2014) , and most recently, reactions and responses to the COVID-19 pandemic (Rothwell and Makridis, 2020) . A growing body of evidence suggests that, during the pandemic, conservatives in the United States have been less likely to restrict movement during shelter-in-place directives (van Holm et al., 2020; Clinton et al., 2021) , less likely to engage in social distancing (Painter and Qiu, 2020) , and less likely to search for information about COVID-19 (Barrios and Hochberg, 2020) .", "cite_spans": [ { "start": 87, "end": 99, "text": "(Balz, 2019;", "ref_id": "BIBREF2" }, { "start": 100, "end": 120, "text": "Dimock et al., 2014;", "ref_id": "BIBREF14" }, { "start": 121, "end": 143, "text": "Lockhart et al., 2020)", "ref_id": "BIBREF36" }, { "start": 256, "end": 270, "text": "(Cooper, 2020;", "ref_id": "BIBREF11" }, { "start": 271, "end": 284, "text": "Mulgan, 2007;", "ref_id": "BIBREF39" }, { "start": 285, "end": 298, "text": "Peters, 2004)", "ref_id": "BIBREF45" }, { "start": 334, "end": 347, "text": "(Nyhan, 2014)", "ref_id": "BIBREF40" }, { "start": 418, "end": 447, "text": "(Rothwell and Makridis, 2020)", "ref_id": "BIBREF52" }, { "start": 626, "end": 649, "text": "(van Holm et al., 2020;", "ref_id": "BIBREF24" }, { "start": 650, "end": 671, "text": "Clinton et al., 2021)", "ref_id": "BIBREF10" }, { "start": 717, "end": 740, "text": "(Painter and Qiu, 2020)", "ref_id": "BIBREF43" }, { "start": 791, "end": 828, "text": "COVID-19 (Barrios and Hochberg, 2020)", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "In addition, conservatives seem more likely to refuse to wear masks, view the pandemic as a hoax, and question or protest against health directives (Van Green and Tyson, 2020) .", "cite_spans": [ { "start": 148, "end": 175, "text": "(Van Green and Tyson, 2020)", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "Though reactance to COVID-19 mitigation efforts among conservatives is well documented, it may have been difficult to predict from existing theory and research. Indeed, a sizeable body of literature suggests that, instead of reacting against state-suggested and mandated precautions, conservatives might instead be more inclined to comply than their liberal counterparts (Jost et al., 2003b (Jost et al., , 2007 Sales, 1973) . Indeed, in previous periods of crisis, conservatives have been more inclined to seek safety (Sales, 1973; Thorisdottir and Jost, 2011) . This is associated with an increased likelihood among conservatives to respond to threats in the environment and appeals to fear (Block and Block, 2006; Jost et al., 2003a,b; Oxley et al., 2008; Pliskin et al., 2015) .", "cite_spans": [ { "start": 371, "end": 390, "text": "(Jost et al., 2003b", "ref_id": "BIBREF28" }, { "start": 391, "end": 411, "text": "(Jost et al., , 2007", "ref_id": "BIBREF29" }, { "start": 412, "end": 424, "text": "Sales, 1973)", "ref_id": "BIBREF54" }, { "start": 519, "end": 532, "text": "(Sales, 1973;", "ref_id": "BIBREF54" }, { "start": 533, "end": 561, "text": "Thorisdottir and Jost, 2011)", "ref_id": "BIBREF56" }, { "start": 693, "end": 716, "text": "(Block and Block, 2006;", "ref_id": "BIBREF5" }, { "start": 717, "end": 738, "text": "Jost et al., 2003a,b;", "ref_id": null }, { "start": 739, "end": 758, "text": "Oxley et al., 2008;", "ref_id": "BIBREF42" }, { "start": 759, "end": 780, "text": "Pliskin et al., 2015)", "ref_id": "BIBREF47" } ], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "In other words, both evidence from research and reports from the media suggest that conservatives demonstrate less compliance to COVID-19 directives than liberals while, at the same time, past research suggests conservatives might be more responsive if presented with an objective threat or when consumed by a sense of fear. That said, it is clear the divergent and partisan responses to mitigation efforts which have been observed to date pose a serious threat to communities in the United States (Rothwell and Makridis, 2020) . Therefore, in this research we explore what might motivate conservatives to adhere to health directives. Empirically, we utilize community level mobility data to understand changes (or lack thereof) in behavior, analyze millions of tweets and a set of official press releases to measure expressed fear, and seek ways in which public health and other officials might responsibly and effectively apply fear appeals to motivate behavior. Analytically, we first measure fear in press releases and tweets using word embeddings and distributed dictionaries, we identify the factors most likely to contribute to expressions of fear with gradient boosted trees, identify the words most common in press releases expressing fear, and explore the associations between expressions of fear and behavior changing with random effects models.", "cite_spans": [ { "start": 498, "end": 527, "text": "(Rothwell and Makridis, 2020)", "ref_id": "BIBREF52" } ], "ref_spans": [], "eq_spans": [], "section": "Introduction", "sec_num": "1" }, { "text": "As stated in the previous section, recent work suggesting increased skepticism and reluctance to adhere to public health directives among conservatives (Barrios and Hochberg, 2020; Iyengar and Massey, 2019; Painter and Qiu, 2020; Rothwell and Makridis, 2020; Van Green and Tyson, 2020) is surprising. Indeed, a long stream of research has found conservatives to be more rule abiding, risk averse, and prevention focused (Jost et al., 2003a (Jost et al., , 2007 Sales, 1973) . In times of crisis, conservatives are more likely than liberals to seek safety (Sales, 1973; Thorisdottir and Jost, 2011) . But given the current tug-of-war between political rhetoric and health risk (Barrios and Hochberg, 2020; Cruwys et al., 2020; Rothgerber et al., 2020; van der Linden et al., 2020) some work suggests conservatives have come to rely more on political identity (van Holm et al., 2020; Allcott et al., 2020; Gadarian et al., 2020; Painter and Qiu, 2020) , which could be influenced or exacerbated by conservative officials downplaying the health risks of the coronavirus (Bursztyn et al., 2020; Peters, 2020) . One specific attribute commonly associated with a conservative mindset may explain why some have diminished or ignored the importance of COVID-19 health directives. That attribute is fear.", "cite_spans": [ { "start": 181, "end": 206, "text": "Iyengar and Massey, 2019;", "ref_id": "BIBREF25" }, { "start": 207, "end": 229, "text": "Painter and Qiu, 2020;", "ref_id": "BIBREF43" }, { "start": 230, "end": 258, "text": "Rothwell and Makridis, 2020;", "ref_id": "BIBREF52" }, { "start": 259, "end": 285, "text": "Van Green and Tyson, 2020)", "ref_id": null }, { "start": 420, "end": 439, "text": "(Jost et al., 2003a", "ref_id": "BIBREF27" }, { "start": 440, "end": 460, "text": "(Jost et al., , 2007", "ref_id": "BIBREF29" }, { "start": 461, "end": 473, "text": "Sales, 1973)", "ref_id": "BIBREF54" }, { "start": 555, "end": 568, "text": "(Sales, 1973;", "ref_id": "BIBREF54" }, { "start": 569, "end": 597, "text": "Thorisdottir and Jost, 2011)", "ref_id": "BIBREF56" }, { "start": 676, "end": 704, "text": "(Barrios and Hochberg, 2020;", "ref_id": "BIBREF3" }, { "start": 705, "end": 725, "text": "Cruwys et al., 2020;", "ref_id": null }, { "start": 726, "end": 750, "text": "Rothgerber et al., 2020;", "ref_id": null }, { "start": 751, "end": 779, "text": "van der Linden et al., 2020)", "ref_id": "BIBREF35" }, { "start": 858, "end": 881, "text": "(van Holm et al., 2020;", "ref_id": "BIBREF24" }, { "start": 882, "end": 903, "text": "Allcott et al., 2020;", "ref_id": "BIBREF1" }, { "start": 904, "end": 926, "text": "Gadarian et al., 2020;", "ref_id": "BIBREF18" }, { "start": 927, "end": 949, "text": "Painter and Qiu, 2020)", "ref_id": "BIBREF43" }, { "start": 1067, "end": 1090, "text": "(Bursztyn et al., 2020;", "ref_id": "BIBREF6" }, { "start": 1091, "end": 1104, "text": "Peters, 2020)", "ref_id": "BIBREF46" } ], "ref_spans": [], "eq_spans": [], "section": "Background and related work", "sec_num": "2" }, { "text": "Past research suggests that conservatives, more than liberals, display a reaction to fear in response to threats in the environment (Block and Block, 2006; Jost et al., 2003a,b; Oxley et al., 2008; Pliskin et al., 2015) . That is, conservatives have a stronger reaction to threats and new experiences (Oxley et al., 2008) and express stronger emotional reactions to negative outcomes (Joel et al., 2014) . Such a response could be driven by a greater need for control over the environment and greater impulse to reduce uncertainty (Jost et al., 2003a) . Since conservatives generally respect authority and want the hierarchical structure to remain in place, they are more fearful of change to this structure (Adorno et al., 1950; Jurgert and Duckitt, 2009) . These conflicting streams of research can be reconciled if conservatives who do experience or express fear of coronavirus are more likely to adhere to health directives as compared to conservatives who do not experience fear of coronavirus. Such a pattern would not only explain the response of conservatives to COVID-19 directives and recommendations but would also suggest a path forward for policy makers intent on motivating greater adherence to health directives.", "cite_spans": [ { "start": 132, "end": 155, "text": "(Block and Block, 2006;", "ref_id": "BIBREF5" }, { "start": 156, "end": 177, "text": "Jost et al., 2003a,b;", "ref_id": null }, { "start": 178, "end": 197, "text": "Oxley et al., 2008;", "ref_id": "BIBREF42" }, { "start": 198, "end": 219, "text": "Pliskin et al., 2015)", "ref_id": "BIBREF47" }, { "start": 301, "end": 321, "text": "(Oxley et al., 2008)", "ref_id": "BIBREF42" }, { "start": 384, "end": 403, "text": "(Joel et al., 2014)", "ref_id": "BIBREF26" }, { "start": 531, "end": 551, "text": "(Jost et al., 2003a)", "ref_id": "BIBREF27" }, { "start": 708, "end": 729, "text": "(Adorno et al., 1950;", "ref_id": null }, { "start": 730, "end": 756, "text": "Jurgert and Duckitt, 2009)", "ref_id": "BIBREF30" } ], "ref_spans": [], "eq_spans": [], "section": "Background and related work", "sec_num": "2" }, { "text": "To explore this phenomenon, we utilize two sources of data representing community level expressions of sentiment. First, as a proxy for the attitudes of local and federal officials, we collect press releases. Press releases are a common communication method used to inform a large number of citizens of a problem. Similar to past public health crises, local and federal government offices and agencies have used press releases to communicate official information to the public. Press releases have been used in past research as representations of the overall attitude of government agencies and officials (Fairbanks et al., 2007; Grossman et al., 2020; Lee and Basnyat, 2013; Levi and Stoker, 2000; Mayhew, 1974) . Second, to measure the fear expressed by citizens in each community, we collect tweets related to COVID-19.", "cite_spans": [ { "start": 605, "end": 629, "text": "(Fairbanks et al., 2007;", "ref_id": "BIBREF16" }, { "start": 630, "end": 652, "text": "Grossman et al., 2020;", "ref_id": "BIBREF23" }, { "start": 653, "end": 675, "text": "Lee and Basnyat, 2013;", "ref_id": "BIBREF32" }, { "start": 676, "end": 698, "text": "Levi and Stoker, 2000;", "ref_id": "BIBREF33" }, { "start": 699, "end": 712, "text": "Mayhew, 1974)", "ref_id": "BIBREF37" } ], "ref_spans": [], "eq_spans": [], "section": "Background and related work", "sec_num": "2" }, { "text": "In this section we first present the press release, Twitter, and behavioral data sets. We then elaborate the method used to measure the degree of fear expressed in text data. Finally, we outline the modeling framework for assessing the factors influencing the expression of fear, the nature of observed communications, and the impact on behavior changes in response to state directives.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Methodology", "sec_num": "3" }, { "text": "All data was collected over a 30-day period between February 29, 2020 and March 29, 2020. February 29 was selected as the starting point of observation because the first COVID-related death in the United States was announced on this day (CDC, 2020). We collected press releases from federal government agencies and offices in addition to those from the local governments of the 53 most populous metropolitan areas in the United States. In total, we collected 166 national and 1232 local press releases across all metropolitan areas during the observation period. However, not every local government issued a press release each day. In total, there were 291 observations which included both local and national press releases.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Press release data", "sec_num": "3.1" }, { "text": "Using the Twitter API, tweets were collected which contained any occurrence (including but not limited to hashtags) of the words coronavirus, covid, covid-19, or sars-cov2. Tweets were filtered to include only those from authors whose profile indicated they reside in one of the 53 metropolitan areas. In total, more than two million relevant tweets were collected during the observation period.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Twitter data", "sec_num": "3.2" }, { "text": "To measure movement during the pandemic, we utilized mobility data provided by Google, aggregated at the level of the county and collected from individuals who have opted in to location sharing features in Android or Google services (Google, 2020) . This data is subset into a variety of location types and represents the percent change in the number of visits and length of stay at each location as compared to an out of period baseline. We calculated the mean value for public locations (retail, grocery and pharmacy, workplaces, and transit stations) as our metric of local movement. County level data was aggregated to the level of the metropolitan area using a population-weighted mean.", "cite_spans": [ { "start": 233, "end": 247, "text": "(Google, 2020)", "ref_id": "BIBREF22" } ], "ref_spans": [], "eq_spans": [], "section": "Pandemic response data", "sec_num": "3.3" }, { "text": "To determine the aggregate political identity of the metropolitan area, we collected the results of presidential elections that occurred between 2000 and 2016 (MIT, 2018). Of course, it is possible that communities may have deviated from this baseline since 2016. We encourage future researcher to explore the ways in which community level political identification has changed since 2016. The average votes for Republican and Democrat candidates were calculated. If a metropolitan area cast more votes on average for Republican candidates over this period it was labelled conservative, otherwise it was classified as liberal. This categorical variable was then dummy coded for analysis. Across the 53 metropolitan areas, 17 were classified as conservative and 36 were classified as liberal.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Pandemic response data", "sec_num": "3.3" }, { "text": "In addition, we collected control variables including the number of COVID-19 cases and deaths reported for each metropolitan area, for each day, local income and poverty metrics, day of the week, and more. For the sake of parsimony, only variables that had a significant impact on the models are reported and discussed.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Pandemic response data", "sec_num": "3.3" }, { "text": "To measure expressed fear in press releases and tweets, we constructed a dictionary to represent the construct. First, we collected 35 common synonyms for the target word fear. Next, we extracted a vector representation for each word from a pretrained language model (Pennington et al., 2014) . The cosine similarity between each target-synonym pair was calculated and words were clustered based on these values. Finally, the synonyms forming the tightest cluster around the word fear were selected as the construct dictionary. This dictionary consisted of 25 words. Construct vectors were aggregated by taking the sum of individual vectors divided by their Euclidean norm (Garten et al., 2018) , resulting in a single construct vector of 200 dimensions. Formally, a construct vector C is calculated as:", "cite_spans": [ { "start": 267, "end": 292, "text": "(Pennington et al., 2014)", "ref_id": "BIBREF44" }, { "start": 673, "end": 694, "text": "(Garten et al., 2018)", "ref_id": "BIBREF19" } ], "ref_spans": [], "eq_spans": [], "section": "Measuring expressed fear", "sec_num": "3.4" }, { "text": "C = w\u2208D R R(w) || w\u2208D R R(w)|| 2 (1)", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Measuring expressed fear", "sec_num": "3.4" }, { "text": "Where w is a word, R is an embedding representation for w, and D R is the set of words in the construct dictionary representing C (Garten et al., 2018) .", "cite_spans": [ { "start": 130, "end": 151, "text": "(Garten et al., 2018)", "ref_id": "BIBREF19" } ], "ref_spans": [], "eq_spans": [], "section": "Measuring expressed fear", "sec_num": "3.4" }, { "text": "Next, press releases and tweets were normalized. Tweets and press release sentences were tokenized into word-grams, made lowercase, and single character words and stop words were removed (Bird et al., 2009; Symeonidis et al., 2018) . Then, vectors for tweet and sentence tokens were aggregated as described in equation 1 (Garten et al., 2018; Pennington et al., 2014) . Cosine similarity provided a measure of similarity between the construct vector and each aggregated tweet or sentence vector (Caliskan et al., 2017; Garten et al., 2018) . Note that the method of aggregation described in equation 1 pre-normalizes the vector representation. As such, similarity S between a construct representation, say for fear, C f , and a given document (e.g., a single tweet), T i , is calculated as the dot product of the respective aggregate vectors:", "cite_spans": [ { "start": 187, "end": 206, "text": "(Bird et al., 2009;", "ref_id": "BIBREF4" }, { "start": 207, "end": 231, "text": "Symeonidis et al., 2018)", "ref_id": "BIBREF55" }, { "start": 321, "end": 342, "text": "(Garten et al., 2018;", "ref_id": "BIBREF19" }, { "start": 343, "end": 367, "text": "Pennington et al., 2014)", "ref_id": "BIBREF44" }, { "start": 495, "end": 518, "text": "(Caliskan et al., 2017;", "ref_id": "BIBREF7" }, { "start": 519, "end": 539, "text": "Garten et al., 2018)", "ref_id": "BIBREF19" } ], "ref_spans": [], "eq_spans": [], "section": "Measuring expressed fear", "sec_num": "3.4" }, { "text": "S = C f \u2022 T i (2)", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Measuring expressed fear", "sec_num": "3.4" }, { "text": "The resulting document-level similarity measures were aggregated to the level of the day and metropolitan area.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Measuring expressed fear", "sec_num": "3.4" }, { "text": "To identify the factors most closely associated with increases in expressed fear, we employed gradi- ent boosted trees (Chen and Guestrin, 2016; Friedman, 2001 ). Gradient boosted decision trees employ a series of shallow trees to predict an outcome variable (Quinlan, 1986) . The assumption is that many weak learners will achieve a probably approximately correct (PAC) result (Valiant, 1984) . Each tree is evaluated based on the gradient of the error with respect to the prediction via functional gradient descent (Ruder, 2017) . Improvements in prediction accuracy for subtrees with a steeper gradient lead to larger overall improvements. Thus, gradient boosted trees result in the identification of variables that have the overall greatest influence on predictive accuracy. The method is amenable to analysis with relatively smaller sample sizes (Zhao and Duangsoithong, 2019) , is not impacted by multicolinearity (Ding et al., 2016) , and has been used in the past for prediction tasks involving social media in general and tweets in particular (Li et al., 2017; Ong et al., 2017) . We implement gradient boosted decision trees via extreme gradient boosting or XGBoost (Chen and Guestrin, 2016) .", "cite_spans": [ { "start": 119, "end": 144, "text": "(Chen and Guestrin, 2016;", "ref_id": "BIBREF9" }, { "start": 145, "end": 159, "text": "Friedman, 2001", "ref_id": "BIBREF17" }, { "start": 259, "end": 274, "text": "(Quinlan, 1986)", "ref_id": "BIBREF48" }, { "start": 378, "end": 393, "text": "(Valiant, 1984)", "ref_id": "BIBREF57" }, { "start": 517, "end": 530, "text": "(Ruder, 2017)", "ref_id": "BIBREF53" }, { "start": 851, "end": 881, "text": "(Zhao and Duangsoithong, 2019)", "ref_id": "BIBREF62" }, { "start": 920, "end": 939, "text": "(Ding et al., 2016)", "ref_id": "BIBREF15" }, { "start": 1052, "end": 1069, "text": "(Li et al., 2017;", "ref_id": "BIBREF34" }, { "start": 1070, "end": 1087, "text": "Ong et al., 2017)", "ref_id": "BIBREF41" }, { "start": 1176, "end": 1201, "text": "(Chen and Guestrin, 2016)", "ref_id": "BIBREF9" } ], "ref_spans": [], "eq_spans": [], "section": "Identifying factors influencing fear", "sec_num": "3.5" }, { "text": "To assess the type of language used in high-versus low-fear press releases, we counted each word in each press release (65,237 total words). Press releases were categorized based on their average cosine similarity (equation 2) with the fear construct (equation 1). Those that were on average more closely associated with fear than the median measure (0.5713) were classified as high fear and those below the median were classified as low fear. A log-odds ratio was then calculated for each word in each category of high-or low-fear communications, indicating the probability of a word occurring in a press release. Words with a positive (negative) log-odds ratio are more likely to appear in a high (low) fear press release.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Press release content analysis", "sec_num": "3.6" }, { "text": "To model the impact of fear and political identity on pandemic responses, we utilized a random effects model. The data is panel in nature, with individual observations being made each day, for each subject or metropolitan area. Thus, a random effects specification assumes that time invariant variables are uncorrelated with the time varying predictors, which enables an examination of time invariant variables (such as political identity) on the outcome variable (local movement). Errors were clustered at the level of the metropolitan area, allowing us to control for community differences. We used a one-day lag for the expression of fear on Twitter, allowing us to assess the relationship between this expression and subsequent changes in movement behavior. Past work has also used a one-day lag when analyzing tweets (Kaminski, 2016; Li et al., 2017; Zhang et al., 2013) based on evidence that 75% of tweet replies are made within 17 minutes and the majority of twitter users are passive (Ye and Wu, 2013; Romero et al., 2011) .", "cite_spans": [ { "start": 822, "end": 838, "text": "(Kaminski, 2016;", "ref_id": "BIBREF31" }, { "start": 839, "end": 855, "text": "Li et al., 2017;", "ref_id": "BIBREF34" }, { "start": 856, "end": 875, "text": "Zhang et al., 2013)", "ref_id": "BIBREF61" }, { "start": 993, "end": 1010, "text": "(Ye and Wu, 2013;", "ref_id": "BIBREF60" }, { "start": 1011, "end": 1031, "text": "Romero et al., 2011)", "ref_id": "BIBREF49" } ], "ref_spans": [], "eq_spans": [], "section": "Modeling the impact on pandemic response", "sec_num": "3.7" }, { "text": "The gradient boosted model predicts the next period (day) fear expression on Twitter based on the fear expressed in the current period in both local and national press releases, the number of confirmed COVID-19 cases and deaths in the current period, and the majority political identity, poverty rate, and median household income in a metropolitan area. Due to the sparsity in press release data described previously, the current analysis does not distinguish between metropolitan area and instead considers the influence of variables on the expression of fear across the entire data set. The model was trained on 80% of the available data and tested against the remaining 20% holdout sample, resulting in 235 training observations and 56 test observations. Each iteration was cross-validated with 10 folds for the purposes of hyperparameter tuning. After 700 iterations, the best performing model employed a learning rate of 0.025, a max tree depth of two, a minimum child weight of one, and a gamma of zero. The final model improves on the root mean square error (RMSE) of the untuned model from 0.0059 to 0.0055 and increases the R 2 value from 0.3388 to 0.3931. As shown in Figure 1 , the most important variable for predicting the amount of fear expressed in tweets was the amount of fear expressed in national press releases the previous day, followed by the number of COVID-19 cases reported for a given metropolitan area on the previous day. The poverty rate, political identity, median household income, and amount of fear expressed in local press releases for each metropolitan area also contributed to the predictive accuracy, though to a lesser degree.", "cite_spans": [], "ref_spans": [ { "start": 1178, "end": 1186, "text": "Figure 1", "ref_id": "FIGREF0" } ], "eq_spans": [], "section": "Factors influencing the expression of fear", "sec_num": "4.1" }, { "text": "In considering the most likely words to appear in high-fear versus low-fear press releases from the local government, it is clear that the former tend to emphasize language directly related to COVID-19 like fever, self-quarantine, flu, shortness, breath, and recover while the latter lacks such a focus and contains words like tax, loan, sales, bank, survey, art, and cultural (see Figure 2) . Similarly, in national press releases, those with higher expressions of fear are more likely to contain words like tests, CDC, testing, spread, protect, support, and health Figure 2 : The most likely words to appear in high-fear (log-odds > 0) versus low-fear (log-odds < 0) local press releases. Figure 3 : The most likely words to appear in high-fear (log-odds > 0) versus low-fear (log-odds < 0) national press releases.", "cite_spans": [], "ref_spans": [ { "start": 382, "end": 391, "text": "Figure 2)", "ref_id": null }, { "start": 567, "end": 575, "text": "Figure 2", "ref_id": null }, { "start": 691, "end": 699, "text": "Figure 3", "ref_id": null } ], "eq_spans": [], "section": "Language used to express fear", "sec_num": "4.2" }, { "text": "as opposed to words like eligible, nonprofit, office, designations, and assessments (see Figure 3) .", "cite_spans": [], "ref_spans": [ { "start": 89, "end": 98, "text": "Figure 3)", "ref_id": null } ], "eq_spans": [], "section": "Language used to express fear", "sec_num": "4.2" }, { "text": "Finally, we consider whether an increase in expressed fear is associated with changes in behavior with a random effects model. When the base model is fit, we see both the previous day's fear as expressed on Twitter (b = 3.578, p < 0.01) and classification of a community as conservative (b = 1.880, p < 0.01) are associated with an increase in movement in public places over the prepandemic baseline (see Table 1 ). Importantly, however, the interaction between conservative identity and the expression of fear is significant and nega- tive (b = \u22122.113, p < 0.01), suggesting that for majority conservative communities an increase in expressed fear is associated with a subsequent decrease in movement in public places over the baseline. Similarly, when control variables are added we see both the previous day's fear as expressed on Twitter (b = 3.337, p < 0.01) and classification of a community as conservative (b = 1.573, p = 0.012) are associated with an increase in movement in public places over the pre-pandemic baseline (see Table 1 ). Again, the interaction between conservative identity and the expression of fear is significant and negative (b = \u22122.464, p = 0.019), suggesting that for majority conservative communities an increase in expressed fear is associated with a subsequent decrease in movement in public places over the baseline.", "cite_spans": [], "ref_spans": [ { "start": 405, "end": 412, "text": "Table 1", "ref_id": "TABREF1" }, { "start": 1034, "end": 1041, "text": "Table 1", "ref_id": "TABREF1" } ], "eq_spans": [], "section": "Relationship with behavior change", "sec_num": "4.3" }, { "text": "Past research suggests that conservatives are more likely to comply with authority (Jost et al., 2003a (Jost et al., ,b, 2007 Sales, 1973; Thorisdottir and Jost, 2011) and are likely to take preventative measures when they are fearful of a situation (Joel et al., 2014) . However, a more recent stream of research argues the opposite (Barrios and Hochberg, 2020; Iyengar and Massey, 2019; Painter and Qiu, 2020; Rothwell and Makridis, 2020; Van Green and Tyson, 2020) including evidence which suggests conservative non-adherence to COVID-19 health directives (Allcott et al., 2020; DeFranza et al., 2020; Gollwitzer et al., 2020) . We offer evidence that may help to reconcile this contradiction. The present research suggests that while conservative communities in general ignore public health guidelines, this behavior is not monolithic. Instead, we see an association with the community's expressions of fear when publicly discussing the pandemic and subsequent behaviors. When conservative communities express fear of the pandemic, subsequent movement in public places decreases as compared to pre-pandemic baselines. This is not the case in majority liberal communities, who exhibit no change in behavior in association with expressed fear. Moreover, we see that a dominant antecedent of local expressions of fear in conservative communities is the clear expression of fear from federal agencies and officials. Importantly, increasing confirmed COVID-19 case counts also predicted an increase in the expression of fear among members of majority conservative communities. This finding may be especially helpful in early phases of a pandemic when decisive action is essential.", "cite_spans": [ { "start": 83, "end": 102, "text": "(Jost et al., 2003a", "ref_id": "BIBREF27" }, { "start": 103, "end": 125, "text": "(Jost et al., ,b, 2007", "ref_id": null }, { "start": 126, "end": 138, "text": "Sales, 1973;", "ref_id": "BIBREF54" }, { "start": 139, "end": 167, "text": "Thorisdottir and Jost, 2011)", "ref_id": "BIBREF56" }, { "start": 250, "end": 269, "text": "(Joel et al., 2014)", "ref_id": "BIBREF26" }, { "start": 363, "end": 388, "text": "Iyengar and Massey, 2019;", "ref_id": "BIBREF25" }, { "start": 389, "end": 411, "text": "Painter and Qiu, 2020;", "ref_id": "BIBREF43" }, { "start": 412, "end": 440, "text": "Rothwell and Makridis, 2020;", "ref_id": "BIBREF52" }, { "start": 441, "end": 467, "text": "Van Green and Tyson, 2020)", "ref_id": null }, { "start": 559, "end": 581, "text": "(Allcott et al., 2020;", "ref_id": "BIBREF1" }, { "start": 582, "end": 604, "text": "DeFranza et al., 2020;", "ref_id": null }, { "start": 605, "end": 629, "text": "Gollwitzer et al., 2020)", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "Discussion", "sec_num": "5" }, { "text": "Taken together, these results suggests that majority conservative communities might benefit from early regular updates from local and federal officials. Because discussions of COVID-19 have be-come partisan (Allcott et al., 2020) , the role of local governments cannot and should not be ignored. It is essential that official communications present details of the current situation clearly and honestly, but also that they emphasize the seriousness and severity of the pandemic (Wood and Schulman, 2021) . At the same time, statements that work to diminish perceptions of risk or severity of the pandemic could be a detriment to efforts to manage infection rates in communities. The implications of these results extend beyond COVID-19 and help to inform policy communications at the local and federal level, specifically those targeted at majority conservative communities and for messages which may be unpopular but timely.", "cite_spans": [ { "start": 207, "end": 229, "text": "(Allcott et al., 2020)", "ref_id": "BIBREF1" }, { "start": 478, "end": 503, "text": "(Wood and Schulman, 2021)", "ref_id": "BIBREF59" } ], "ref_spans": [], "eq_spans": [], "section": "Discussion", "sec_num": "5" }, { "text": "We have presented an analysis of the degree of fear expressed in association with COVID-19 in both official government press releases and by Twitter users in the 53 most populous metropolitan areas in the United States. In doing so, we have identified key factors that influence expressions of fear among members of communities: strong expressions of fear in national communications and increasing confirmed COVID-19 case counts. In addition, we have identified the words most likely to occur in both high-and low-fear official communications. Finally, we have provided evidence that expressions of fear are associated with different prophylactic behaviors in conservative and liberal communities. In the latter, increased expressions of fear are not associated with behavior change. However, in the former, increased expressions of fear are associated with an overall decrease in movement and activity in public spaces.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Conclusion", "sec_num": "6" }, { "text": "These results suggest that restricted testing and under-reporting case counts could be detrimental to safe individual behaviors and compliance with public health policy recommendations. Similarly, minimizing the severity or seriousness of COVID-19 poses a particular danger in majority conservative communities. Moving beyond COVID-19 mitigation, the present research emphasizes the importance of timely, relevant, and clear information capable of communicating the authentic seriousness of a situation, especially in majority conservative communities.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Conclusion", "sec_num": "6" } ], "back_matter": [], "bib_entries": { "BIBREF0": { "ref_id": "b0", "title": "Else Frenkel-Brunswik, Daniel Levinson, and Nevitt Sanford. 1950. The authoritarian personality", "authors": [ { "first": "Theodor", "middle": [ "W" ], "last": "Adorno", "suffix": "" } ], "year": null, "venue": "Harpers", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Theodor W. Adorno, Else Frenkel-Brunswik, Daniel Levinson, and Nevitt Sanford. 1950. The authori- tarian personality. Harpers.", "links": null }, "BIBREF1": { "ref_id": "b1", "title": "Polarization and public health: Partisan differences in social distancing during the coronavirus pandemic", "authors": [ { "first": "Hunt", "middle": [], "last": "Allcott", "suffix": "" }, { "first": "Levi", "middle": [], "last": "Boxell", "suffix": "" }, { "first": "Jacob", "middle": [], "last": "Conway", "suffix": "" }, { "first": "Matthew", "middle": [], "last": "Gentzkow", "suffix": "" }, { "first": "Michael", "middle": [], "last": "Thaler", "suffix": "" }, { "first": "David", "middle": [ "Y" ], "last": "Yang", "suffix": "" } ], "year": 2020, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Hunt Allcott, Levi Boxell, Jacob Conway, Matthew Gentzkow, Michael Thaler, and David Y. Yang. 2020. Polarization and public health: Partisan dif- ferences in social distancing during the coronavirus pandemic. NBER:w26946.", "links": null }, "BIBREF2": { "ref_id": "b2", "title": "Americans hate all the partisanship, but they're also more partisan than they were. Washington Post", "authors": [ { "first": "Dan", "middle": [], "last": "Balz", "suffix": "" } ], "year": 2019, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Dan Balz. 2019. Americans hate all the partisanship, but they're also more partisan than they were. Wash- ington Post.", "links": null }, "BIBREF3": { "ref_id": "b3", "title": "Risk perception through the lens of politics in the time of the COVID-19 pandemic", "authors": [ { "first": "M", "middle": [], "last": "John", "suffix": "" }, { "first": "Yael", "middle": [], "last": "Barrios", "suffix": "" }, { "first": "", "middle": [], "last": "Hochberg", "suffix": "" } ], "year": 2020, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "John M. Barrios and Yael Hochberg. 2020. Risk per- ception through the lens of politics in the time of the COVID-19 pandemic. NBER:27008.", "links": null }, "BIBREF4": { "ref_id": "b4", "title": "Natural language processing with Python: Analyzing text with the natural language toolkit", "authors": [ { "first": "Steven", "middle": [], "last": "Bird", "suffix": "" }, { "first": "Ewan", "middle": [], "last": "Klein", "suffix": "" }, { "first": "Edward", "middle": [], "last": "Loper", "suffix": "" } ], "year": 2009, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Steven Bird, Ewan Klein, and Edward Loper. 2009. Natural language processing with Python: Analyz- ing text with the natural language toolkit. O'Reilly Media Inc.", "links": null }, "BIBREF5": { "ref_id": "b5", "title": "Nursery school personality and political orientation two decades later", "authors": [ { "first": "Jack", "middle": [], "last": "Block", "suffix": "" }, { "first": "Jeanne", "middle": [ "H" ], "last": "Block", "suffix": "" } ], "year": 2006, "venue": "Journal of Research in Personality", "volume": "40", "issue": "5", "pages": "734--749", "other_ids": {}, "num": null, "urls": [], "raw_text": "Jack Block and Jeanne H. Block. 2006. Nursery school personality and political orientation two decades later. Journal of Research in Personality, 40(5):734- 749.", "links": null }, "BIBREF6": { "ref_id": "b6", "title": "Misinformation during a pandemic", "authors": [ { "first": "Leonardo", "middle": [], "last": "Bursztyn", "suffix": "" }, { "first": "Aakaash", "middle": [], "last": "Rao", "suffix": "" }, { "first": "Christopher", "middle": [], "last": "Roth", "suffix": "" }, { "first": "David", "middle": [], "last": "Yanagizawa-Drott", "suffix": "" } ], "year": 2020, "venue": "", "volume": "", "issue": "", "pages": "2020--2064", "other_ids": {}, "num": null, "urls": [], "raw_text": "Leonardo Bursztyn, Aakaash Rao, Christopher Roth, and David Yanagizawa-Drott. 2020. Misinforma- tion during a pandemic. University of Chicago, Becker Friedman Institute for Economics Working Paper:2020-44.", "links": null }, "BIBREF7": { "ref_id": "b7", "title": "Semantics derived automatically from language corpora contain human-like biases", "authors": [ { "first": "Aylin", "middle": [], "last": "Caliskan", "suffix": "" }, { "first": "Joanna", "middle": [ "J" ], "last": "Bryson", "suffix": "" }, { "first": "Arvind", "middle": [], "last": "Narayanan", "suffix": "" } ], "year": 2017, "venue": "Science", "volume": "356", "issue": "6334", "pages": "183--186", "other_ids": {}, "num": null, "urls": [], "raw_text": "Aylin Caliskan, Joanna J. Bryson, and Arvind Narayanan. 2017. Semantics derived automatically from language corpora contain human-like biases. Science, 356(6334):183-186.", "links": null }, "BIBREF8": { "ref_id": "b8", "title": "Washington State report first COVID-19 death", "authors": [], "year": null, "venue": "Centers for Disease Control and Prevention", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "CDC. 2020. CDC, Washington State report first COVID-19 death. Technical report, Centers for Dis- ease Control and Prevention.", "links": null }, "BIBREF9": { "ref_id": "b9", "title": "XGBoost: A scalable tree boosting system", "authors": [ { "first": "Tianqi", "middle": [], "last": "Chen", "suffix": "" }, { "first": "Carlos", "middle": [], "last": "Guestrin", "suffix": "" } ], "year": 2016, "venue": "Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining", "volume": "", "issue": "", "pages": "785--794", "other_ids": {}, "num": null, "urls": [], "raw_text": "Tianqi Chen and Carlos Guestrin. 2016. XGBoost: A scalable tree boosting system. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pages 785--794. Association for Computing Machinery.", "links": null }, "BIBREF10": { "ref_id": "b10", "title": "Partisan pandemic: How partisanship and public health concerns affect individuals' social mobility during covid-19", "authors": [ { "first": "J", "middle": [], "last": "Clinton", "suffix": "" }, { "first": "J", "middle": [], "last": "Cohen", "suffix": "" }, { "first": "J", "middle": [], "last": "Lapinski", "suffix": "" }, { "first": "M", "middle": [], "last": "Trussler", "suffix": "" } ], "year": 2021, "venue": "Science Advances", "volume": "7", "issue": "2", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "J. Clinton, J. Cohen, J. Lapinski, and M. Trussler. 2021. Partisan pandemic: How partisanship and public health concerns affect individuals' social mobility during covid-19. Science Advances, 7(2).", "links": null }, "BIBREF11": { "ref_id": "b11", "title": "Politicization of the bureaucracy across and within administrative traditions", "authors": [ { "first": "Christopher", "middle": [ "A" ], "last": "Cooper", "suffix": "" } ], "year": 2020, "venue": "International Journal of Public Administration", "volume": "", "issue": "", "pages": "1--14", "other_ids": {}, "num": null, "urls": [], "raw_text": "Christopher A. Cooper. 2020. Politicization of the bureaucracy across and within administrative tradi- tions. International Journal of Public Administra- tion, pages 1-14.", "links": null }, "BIBREF12": { "ref_id": "b12", "title": "Greenaway. 2020. A social identity perspective on COVID-19: Health risk is affected by shared group membership", "authors": [ { "first": "Tegan", "middle": [], "last": "Cruwys", "suffix": "" }, { "first": "Mark", "middle": [], "last": "Stevens", "suffix": "" }, { "first": "Katharine", "middle": [ "H" ], "last": "", "suffix": "" } ], "year": null, "venue": "British Journal of Social Psychology", "volume": "59", "issue": "3", "pages": "584--593", "other_ids": {}, "num": null, "urls": [], "raw_text": "Tegan Cruwys, Mark Stevens, and Katharine H. Green- away. 2020. A social identity perspective on COVID-19: Health risk is affected by shared group membership. British Journal of Social Psychology, 59(3):584-593.", "links": null }, "BIBREF13": { "ref_id": "b13", "title": "Arul Mishra, and Himanshu Mishra. 2020. Religion and reactance to COVID-19 mitigation guidelines", "authors": [ { "first": "David", "middle": [], "last": "Defranza", "suffix": "" }, { "first": "Mike", "middle": [], "last": "Lindow", "suffix": "" }, { "first": "Kevin", "middle": [], "last": "Harrison", "suffix": "" } ], "year": null, "venue": "American Psychologist", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "David DeFranza, Mike Lindow, Kevin Harrison, Arul Mishra, and Himanshu Mishra. 2020. Religion and reactance to COVID-19 mitigation guidelines. American Psychologist, Advanced online publica- tion.", "links": null }, "BIBREF14": { "ref_id": "b14", "title": "Political polarization in the American public", "authors": [ { "first": "Michael", "middle": [], "last": "Dimock", "suffix": "" }, { "first": "Jocelyn", "middle": [], "last": "Kiley", "suffix": "" }, { "first": "Scott", "middle": [], "last": "Keeter", "suffix": "" }, { "first": "Carroll", "middle": [], "last": "Doherty", "suffix": "" } ], "year": 2014, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Michael Dimock, Jocelyn Kiley, Scott Keeter, and Car- roll Doherty. 2014. Political polarization in the American public. Technical report, Pew Research Center.", "links": null }, "BIBREF15": { "ref_id": "b15", "title": "Predicting short-term subway ridership and prioritizing its influential factors using gradient boosting decision trees", "authors": [ { "first": "Chuan", "middle": [], "last": "Ding", "suffix": "" }, { "first": "Donggen", "middle": [], "last": "Wang", "suffix": "" }, { "first": "Xiaolei", "middle": [], "last": "Ma", "suffix": "" }, { "first": "Haiyling", "middle": [], "last": "Li", "suffix": "" } ], "year": 2016, "venue": "Sustainability", "volume": "8", "issue": "11", "pages": "1100--1116", "other_ids": {}, "num": null, "urls": [], "raw_text": "Chuan Ding, Donggen Wang, Xiaolei Ma, and Haiyling Li. 2016. Predicting short-term subway ridership and prioritizing its influential factors us- ing gradient boosting decision trees. Sustainability, 8(11):1100-1116.", "links": null }, "BIBREF16": { "ref_id": "b16", "title": "Transparency in government communication", "authors": [ { "first": "Jenille", "middle": [], "last": "Fairbanks", "suffix": "" }, { "first": "Kenneth", "middle": [ "D" ], "last": "Plowman", "suffix": "" }, { "first": "Brad", "middle": [ "L" ], "last": "Rawlins", "suffix": "" } ], "year": 2007, "venue": "Journal of Public Affairs: An International Journal", "volume": "7", "issue": "1", "pages": "23--37", "other_ids": {}, "num": null, "urls": [], "raw_text": "Jenille Fairbanks, Kenneth D. Plowman, and Brad L. Rawlins. 2007. Transparency in government com- munication. Journal of Public Affairs: An Interna- tional Journal, 7(1):23-37.", "links": null }, "BIBREF17": { "ref_id": "b17", "title": "Greedy function approximation: A gradient boosting machine", "authors": [ { "first": "Jerome", "middle": [ "H" ], "last": "Friedman", "suffix": "" } ], "year": 2001, "venue": "Annals of Statistics", "volume": "29", "issue": "5", "pages": "1189--1232", "other_ids": {}, "num": null, "urls": [], "raw_text": "Jerome H. Friedman. 2001. Greedy function approx- imation: A gradient boosting machine. Annals of Statistics, 29(5):1189-1232.", "links": null }, "BIBREF18": { "ref_id": "b18", "title": "Partisanship, health behavior and policy attitudes in the early stages of the COVID-19 pandemic", "authors": [ { "first": "Shana", "middle": [ "K" ], "last": "Gadarian", "suffix": "" }, { "first": "W", "middle": [], "last": "Sata", "suffix": "" }, { "first": "Thomas", "middle": [ "B" ], "last": "Goodman", "suffix": "" }, { "first": "", "middle": [], "last": "Pepinsky", "suffix": "" } ], "year": 2020, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Shana K. Gadarian, Sata W. Goodman, and Thomas B. Pepinsky. 2020. Partisanship, health behavior and policy attitudes in the early stages of the COVID-19 pandemic. SSRN:3562796.", "links": null }, "BIBREF19": { "ref_id": "b19", "title": "Dictionaries and distributions: Combining expert knowledge and large scale textual data content analysis", "authors": [ { "first": "Justin", "middle": [], "last": "Garten", "suffix": "" }, { "first": "Joe", "middle": [], "last": "Hoover", "suffix": "" }, { "first": "Kate", "middle": [ "M" ], "last": "Johnson", "suffix": "" } ], "year": 2018, "venue": "Behavior Research Methods", "volume": "50", "issue": "1", "pages": "344--361", "other_ids": {}, "num": null, "urls": [], "raw_text": "Justin Garten, Joe Hoover, Kate M. Johnson, Reihane Boghrati, Carol Iskiwitch, and Morteza Dehghani. 2018. Dictionaries and distributions: Combining ex- pert knowledge and large scale textual data content analysis. Behavior Research Methods, 50(1):344- 361.", "links": null }, "BIBREF21": { "ref_id": "b21", "title": "Partisan differences in physical distancing are linked to health outcomes during the COVID-19 pandemic", "authors": [ { "first": "Jay", "middle": [ "J" ], "last": "Knowles", "suffix": "" }, { "first": "", "middle": [], "last": "Van Bavel", "suffix": "" } ], "year": 2020, "venue": "Nature Human Behavior", "volume": "4", "issue": "", "pages": "1186--1197", "other_ids": {}, "num": null, "urls": [], "raw_text": "Knowles, and Jay J. Van Bavel. 2020. Partisan dif- ferences in physical distancing are linked to health outcomes during the COVID-19 pandemic. Nature Human Behavior, 4:1186--1197.", "links": null }, "BIBREF22": { "ref_id": "b22", "title": "COVID-19 community mobility report", "authors": [ { "first": "", "middle": [], "last": "Google", "suffix": "" } ], "year": 2020, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Google. 2020. COVID-19 community mobility report. Technical report, Google.", "links": null }, "BIBREF23": { "ref_id": "b23", "title": "Political partisanship influences behavioral responses to governor's recommendations for COVID-19 prevention in the United States", "authors": [ { "first": "Guy", "middle": [], "last": "Grossman", "suffix": "" }, { "first": "Kim", "middle": [], "last": "Soojong", "suffix": "" }, { "first": "Jonah", "middle": [], "last": "Rexer", "suffix": "" }, { "first": "Harsha", "middle": [], "last": "Thirumurthy", "suffix": "" } ], "year": 2020, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Guy Grossman, Kim Soojong, Jonah Rexer, and Har- sha Thirumurthy. 2020. Political partisanship in- fluences behavioral responses to governor's recom- mendations for COVID-19 prevention in the United States. SSRN:3578695.", "links": null }, "BIBREF24": { "ref_id": "b24", "title": "The impact of political ideology on concern and behavior during", "authors": [ { "first": "Eric", "middle": [], "last": "Joseph Van Holm", "suffix": "" }, { "first": "Jake", "middle": [], "last": "Monaghan", "suffix": "" }, { "first": "Dan", "middle": [ "C" ], "last": "Shahar", "suffix": "" }, { "first": "J", "middle": [ "P" ], "last": "Messina", "suffix": "" }, { "first": "Chris", "middle": [], "last": "Surprenant", "suffix": "" } ], "year": 2020, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Eric Joseph van Holm, Jake Monaghan, Dan C. Shahar, J. P. Messina, and Chris Surprenant. 2020. The im- pact of political ideology on concern and behavior during COVID-19. SSRN:3573224.", "links": null }, "BIBREF25": { "ref_id": "b25", "title": "Scientific communication in a post-truth society. Proceedings of the National Academy of", "authors": [ { "first": "Shanto", "middle": [], "last": "Iyengar", "suffix": "" }, { "first": "Douglas", "middle": [ "S" ], "last": "Massey", "suffix": "" } ], "year": 2019, "venue": "Sciences", "volume": "116", "issue": "16", "pages": "7656--7661", "other_ids": {}, "num": null, "urls": [], "raw_text": "Shanto Iyengar and Douglas S. Massey. 2019. Sci- entific communication in a post-truth society. Pro- ceedings of the National Academy of Sciences, 116(16):7656-7661.", "links": null }, "BIBREF26": { "ref_id": "b26", "title": "Conservatives anticipate and experience stronger emotional reactions to negative outcomes", "authors": [ { "first": "Samantha", "middle": [], "last": "Joel", "suffix": "" }, { "first": "Caitlin", "middle": [ "M" ], "last": "Burton", "suffix": "" }, { "first": "Jason", "middle": [ "E" ], "last": "Plaks", "suffix": "" } ], "year": 2014, "venue": "Journal of Personality", "volume": "82", "issue": "1", "pages": "32--43", "other_ids": {}, "num": null, "urls": [], "raw_text": "Samantha Joel, Caitlin M. Burton, and Jason E. Plaks. 2014. Conservatives anticipate and experience stronger emotional reactions to negative outcomes. Journal of Personality, 82(1):32-43.", "links": null }, "BIBREF27": { "ref_id": "b27", "title": "Exceptions that prove the rule-Using a theory of motivated social cognition to account for ideological incongruities and political anomalies: Reply to Greenberg and Jonas", "authors": [ { "first": "T", "middle": [], "last": "Jost", "suffix": "" }, { "first": "John", "middle": [], "last": "", "suffix": "" }, { "first": "Jack", "middle": [], "last": "Glaser", "suffix": "" }, { "first": "Arie", "middle": [ "W" ], "last": "Kruglanski", "suffix": "" }, { "first": "Frank", "middle": [ "J" ], "last": "Sulloway", "suffix": "" } ], "year": 2003, "venue": "Psychological Bulletin", "volume": "129", "issue": "3", "pages": "383--393", "other_ids": {}, "num": null, "urls": [], "raw_text": "T. Jost, John, Jack Glaser, Arie W. Kruglanski, and Frank J. Sulloway. 2003a. Exceptions that prove the rule-Using a theory of motivated social cognition to account for ideological incongruities and political anomalies: Reply to Greenberg and Jonas. Psycho- logical Bulletin, 129(3):383-393.", "links": null }, "BIBREF28": { "ref_id": "b28", "title": "Political conservatism as motivated social cognition", "authors": [ { "first": "T", "middle": [], "last": "Jost", "suffix": "" }, { "first": "John", "middle": [], "last": "", "suffix": "" }, { "first": "Jack", "middle": [], "last": "Glaser", "suffix": "" }, { "first": "Arie", "middle": [ "W" ], "last": "Kruglanski", "suffix": "" }, { "first": "Frank", "middle": [ "J" ], "last": "Sulloway", "suffix": "" } ], "year": 2003, "venue": "Psychological Bulletin", "volume": "129", "issue": "3", "pages": "339--375", "other_ids": {}, "num": null, "urls": [], "raw_text": "T. Jost, John, Jack Glaser, Arie W. Kruglanski, and Frank J. Sulloway. 2003b. Political conservatism as motivated social cognition. Psychological Bulletin, 129(3):339-375.", "links": null }, "BIBREF29": { "ref_id": "b29", "title": "Are needs to manage uncertainty and threat associated with political conservatism or ideological extremity?", "authors": [ { "first": "T", "middle": [], "last": "Jost", "suffix": "" }, { "first": "Jaime", "middle": [ "L" ], "last": "John", "suffix": "" }, { "first": "Hulda", "middle": [], "last": "Napier", "suffix": "" }, { "first": "Samuel", "middle": [ "D" ], "last": "Thorisdorrit", "suffix": "" }, { "first": "", "middle": [], "last": "Gosling", "suffix": "" }, { "first": "P", "middle": [], "last": "Tibor", "suffix": "" }, { "first": "Brian", "middle": [], "last": "Palfai", "suffix": "" }, { "first": "", "middle": [], "last": "Ostafin", "suffix": "" } ], "year": 2007, "venue": "Personality and Social Psychology Bulletin", "volume": "33", "issue": "7", "pages": "989--1007", "other_ids": {}, "num": null, "urls": [], "raw_text": "T. Jost, John, Jaime L. Napier, Hulda Thorisdorrit, Samuel D. Gosling, Tibor P. Palfai, and Brian Ostafin. 2007. Are needs to manage uncertainty and threat associated with political conservatism or ide- ological extremity? Personality and Social Psychol- ogy Bulletin, 33(7):989-1007.", "links": null }, "BIBREF30": { "ref_id": "b30", "title": "A motivational model of authoritarianism: Integrating personal and situational determinants", "authors": [ { "first": "Philipp", "middle": [], "last": "Jurgert", "suffix": "" }, { "first": "John", "middle": [], "last": "Duckitt", "suffix": "" } ], "year": 2009, "venue": "Political Psychology", "volume": "30", "issue": "5", "pages": "693--719", "other_ids": {}, "num": null, "urls": [], "raw_text": "Philipp Jurgert and John Duckitt. 2009. A motiva- tional model of authoritarianism: Integrating per- sonal and situational determinants. Political Psy- chology, 30(5):693-719.", "links": null }, "BIBREF31": { "ref_id": "b31", "title": "Nowcasting the Bitcoin market with Twitter signals", "authors": [ { "first": "Jermain", "middle": [], "last": "Kaminski", "suffix": "" } ], "year": 2016, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Jermain Kaminski. 2016. Nowcasting the Bitcoin mar- ket with Twitter signals. ArXiv:1406.7577.", "links": null }, "BIBREF32": { "ref_id": "b32", "title": "From press release to news: Mapping the framing of the 2009 H1N1 a influenza pandemic", "authors": [ { "first": "Iccha", "middle": [], "last": "Seow Ting Lee", "suffix": "" }, { "first": "", "middle": [], "last": "Basnyat", "suffix": "" } ], "year": 2013, "venue": "Health Communication", "volume": "28", "issue": "2", "pages": "119--132", "other_ids": {}, "num": null, "urls": [], "raw_text": "Seow Ting Lee and Iccha Basnyat. 2013. From press release to news: Mapping the framing of the 2009 H1N1 a influenza pandemic. Health Communica- tion, 28(2):119-132.", "links": null }, "BIBREF33": { "ref_id": "b33", "title": "Political trust and trustworthiness", "authors": [ { "first": "Margaret", "middle": [], "last": "Levi", "suffix": "" }, { "first": "Laura", "middle": [], "last": "Stoker", "suffix": "" } ], "year": 2000, "venue": "Annual Review of Political Science", "volume": "3", "issue": "1", "pages": "475--507", "other_ids": {}, "num": null, "urls": [], "raw_text": "Margaret Levi and Laura Stoker. 2000. Political trust and trustworthiness. Annual Review of Political Sci- ence, 3(1):475-507.", "links": null }, "BIBREF34": { "ref_id": "b34", "title": "A hybrid model combining convolutional neural network with XGBoost for predicting social media popularity", "authors": [ { "first": "Liuwu", "middle": [], "last": "Li", "suffix": "" }, { "first": "Runwei", "middle": [], "last": "Situ", "suffix": "" }, { "first": "Junyan", "middle": [], "last": "Gao", "suffix": "" }, { "first": "Zhenguo", "middle": [], "last": "Yang", "suffix": "" }, { "first": "Wenyin", "middle": [], "last": "Liu", "suffix": "" } ], "year": 2017, "venue": "Proceedings of the 25th ACM international conference on Multimedia", "volume": "", "issue": "", "pages": "1918--1917", "other_ids": {}, "num": null, "urls": [], "raw_text": "Liuwu Li, Runwei Situ, Junyan Gao, Zhenguo Yang, and Wenyin Liu. 2017. A hybrid model combining convolutional neural network with XGBoost for pre- dicting social media popularity. In Proceedings of the 25th ACM international conference on Multime- dia, pages 1918--1917. Association for Computing Machinery.", "links": null }, "BIBREF35": { "ref_id": "b35", "title": "Coronavirus: New survey shows how Republicans and Democrats are responding differently", "authors": [ { "first": "Costas", "middle": [], "last": "Sander Van Der Linden", "suffix": "" }, { "first": "John", "middle": [], "last": "Panagopoulos", "suffix": "" }, { "first": "", "middle": [], "last": "Kerr", "suffix": "" } ], "year": 2020, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Sander van der Linden, Costas Panagopoulos, and John Kerr. 2020. Coronavirus: New survey shows how Republicans and Democrats are responding differ- ently. The Conversation.", "links": null }, "BIBREF36": { "ref_id": "b36", "title": "America's electorate is increasingly polarized along partisan lines about voting by mail during the covid-19 crisis", "authors": [ { "first": "Mackenzie", "middle": [], "last": "Lockhart", "suffix": "" }, { "first": "Seth", "middle": [ "J" ], "last": "Hill", "suffix": "" }, { "first": "Jennifer", "middle": [], "last": "Merolla", "suffix": "" }, { "first": "Mindy", "middle": [], "last": "Romero", "suffix": "" }, { "first": "Thad", "middle": [], "last": "Kousser", "suffix": "" } ], "year": 2020, "venue": "Proceedings of the National Academy of Sciences", "volume": "117", "issue": "40", "pages": "24640--24642", "other_ids": {}, "num": null, "urls": [], "raw_text": "Mackenzie Lockhart, Seth J. Hill, Jennifer Merolla, Mindy Romero, and Thad Kousser. 2020. Amer- ica's electorate is increasingly polarized along par- tisan lines about voting by mail during the covid-19 crisis. Proceedings of the National Academy of Sci- ences, 117(40):24640-24642.", "links": null }, "BIBREF37": { "ref_id": "b37", "title": "Congress: The electoral connection", "authors": [ { "first": "R", "middle": [], "last": "David", "suffix": "" }, { "first": "", "middle": [], "last": "Mayhew", "suffix": "" } ], "year": 1974, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "David R. Mayhew. 1974. Congress: The electoral con- nection. Yale University Press.", "links": null }, "BIBREF38": { "ref_id": "b38", "title": "County presidential election returns", "authors": [], "year": 2000, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": { "DOI": [ "10.7910/DVN/VOQCHQ" ] }, "num": null, "urls": [], "raw_text": "MIT. 2018. County presidential election returns 2000- 2016. Technical report, MIT Election Data and Sci- ence Lab.", "links": null }, "BIBREF39": { "ref_id": "b39", "title": "Truth in government and the politicization of public service advice", "authors": [ { "first": "Richard", "middle": [], "last": "Mulgan", "suffix": "" } ], "year": 2007, "venue": "Public Administration", "volume": "85", "issue": "3", "pages": "569--586", "other_ids": {}, "num": null, "urls": [], "raw_text": "Richard Mulgan. 2007. Truth in government and the politicization of public service advice. Public Ad- ministration, 85(3):569-586.", "links": null }, "BIBREF40": { "ref_id": "b40", "title": "The partisan divide on Ebola preparedness. The New York Times", "authors": [ { "first": "Brendan", "middle": [], "last": "Nyhan", "suffix": "" } ], "year": 2014, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Brendan Nyhan. 2014. The partisan divide on Ebola preparedness. The New York Times.", "links": null }, "BIBREF41": { "ref_id": "b41", "title": "Personality prediction based on Twitter information in Bahasa Indonesia", "authors": [ { "first": "Veronica", "middle": [], "last": "Ong", "suffix": "" }, { "first": "D", "middle": [ "S" ], "last": "Anneke", "suffix": "" }, { "first": "", "middle": [], "last": "Rahmanto", "suffix": "" }, { "first": "Derwin", "middle": [], "last": "Williem", "suffix": "" }, { "first": "Aryo", "middle": [ "E" ], "last": "Suhartono", "suffix": "" }, { "first": "Esther", "middle": [ "W" ], "last": "Nugroho", "suffix": "" }, { "first": "Muhamad", "middle": [ "N" ], "last": "Andangsari", "suffix": "" }, { "first": "", "middle": [], "last": "Suprayogi", "suffix": "" } ], "year": 2017, "venue": "Federated Conference on Computer Science and Information Systems (FedCSIS)", "volume": "11", "issue": "", "pages": "367--372", "other_ids": {}, "num": null, "urls": [], "raw_text": "Veronica Ong, Anneke D. S. Rahmanto, Williem, Der- win Suhartono, Aryo E. Nugroho, Esther W. Andan- gsari, and Muhamad N. Suprayogi. 2017. Person- ality prediction based on Twitter information in Ba- hasa Indonesia. In Federated Conference on Com- puter Science and Information Systems (FedCSIS), volume 11, pages 367--372. Annals of Computer Science and Information Systems.", "links": null }, "BIBREF42": { "ref_id": "b42", "title": "Political attitudes vary with physiological traits", "authors": [ { "first": "Douglas", "middle": [ "R" ], "last": "Oxley", "suffix": "" }, { "first": "Kevin", "middle": [ "B" ], "last": "Smith", "suffix": "" }, { "first": "John", "middle": [ "R" ], "last": "Alford", "suffix": "" }, { "first": "Matthew", "middle": [ "V" ], "last": "Hibbing", "suffix": "" }, { "first": "Jennifer", "middle": [ "L" ], "last": "Miller", "suffix": "" }, { "first": "Mario", "middle": [], "last": "Scalora", "suffix": "" }, { "first": "Peter", "middle": [ "K" ], "last": "Hatemi", "suffix": "" } ], "year": 2008, "venue": "Science", "volume": "321", "issue": "5896", "pages": "1667--1670", "other_ids": {}, "num": null, "urls": [], "raw_text": "Douglas R. Oxley, Kevin B. Smith, John R. Alford, Matthew V. Hibbing, Jennifer L. Miller, Mario Scalora, and Peter K. Hatemi. 2008. Political attitudes vary with physiological traits. Science, 321(5896):1667-1670.", "links": null }, "BIBREF43": { "ref_id": "b43", "title": "Political beliefs affect compliance with COVID-19 social distancing orders", "authors": [ { "first": "Marcus", "middle": [], "last": "Painter", "suffix": "" }, { "first": "Tian", "middle": [], "last": "Qiu", "suffix": "" } ], "year": 2020, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Marcus Painter and Tian Qiu. 2020. Political beliefs affect compliance with COVID-19 social distancing orders. SSRN:3569098.", "links": null }, "BIBREF44": { "ref_id": "b44", "title": "GloVe: Global vectors for word representation", "authors": [ { "first": "Jeffrey", "middle": [], "last": "Pennington", "suffix": "" }, { "first": "Richard", "middle": [], "last": "Socher", "suffix": "" }, { "first": "Christopher", "middle": [], "last": "Manning", "suffix": "" } ], "year": 2014, "venue": "Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP)", "volume": "", "issue": "", "pages": "1532--1543", "other_ids": {}, "num": null, "urls": [], "raw_text": "Jeffrey Pennington, Richard Socher, and Christopher Manning. 2014. GloVe: Global vectors for word representation. In Proceedings of the 2014 Confer- ence on Empirical Methods in Natural Language Processing (EMNLP), pages 1532--1543. Associ- ation for Computational Linguistics.", "links": null }, "BIBREF45": { "ref_id": "b45", "title": "Politicization in the united states", "authors": [ { "first": "Guy", "middle": [], "last": "Peters", "suffix": "" } ], "year": 2004, "venue": "Politicization of the civil service in comparative perspective", "volume": "", "issue": "", "pages": "125--1138", "other_ids": {}, "num": null, "urls": [], "raw_text": "Guy Peters. 2004. Politicization in the united states. In Politicization of the civil service in comparative perspective, pages 125-1138. Routledge.", "links": null }, "BIBREF46": { "ref_id": "b46", "title": "Alarm, denial, blame: The pro-Trump media's coronavirus distortion. The New York Times", "authors": [ { "first": "Jeremy", "middle": [ "W" ], "last": "Peters", "suffix": "" } ], "year": 2020, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Jeremy W. Peters. 2020. Alarm, denial, blame: The pro-Trump media's coronavirus distortion. The New York Times.", "links": null }, "BIBREF47": { "ref_id": "b47", "title": "Running for your life, in context: Are rightists always less likely to consider fleeing their country when fearing future events", "authors": [ { "first": "Ruthie", "middle": [], "last": "Pliskin", "suffix": "" }, { "first": "Gal", "middle": [], "last": "Sheppes", "suffix": "" }, { "first": "Eran", "middle": [], "last": "Halperin", "suffix": "" } ], "year": 2015, "venue": "Journal of Experimental Social Psychology", "volume": "59", "issue": "", "pages": "90--95", "other_ids": {}, "num": null, "urls": [], "raw_text": "Ruthie Pliskin, Gal Sheppes, and Eran Halperin. 2015. Running for your life, in context: Are rightists al- ways less likely to consider fleeing their country when fearing future events? Journal of Experimen- tal Social Psychology, 59:90-95.", "links": null }, "BIBREF48": { "ref_id": "b48", "title": "Induction of decision trees. Machine Learning", "authors": [ { "first": "J", "middle": [], "last": "", "suffix": "" }, { "first": "Ross", "middle": [], "last": "Quinlan", "suffix": "" } ], "year": 1986, "venue": "", "volume": "1", "issue": "", "pages": "81--106", "other_ids": {}, "num": null, "urls": [], "raw_text": "J. Ross Quinlan. 1986. Induction of decision trees. Ma- chine Learning, 1(1):81-106.", "links": null }, "BIBREF49": { "ref_id": "b49", "title": "Influence and passivity in social media", "authors": [ { "first": "M", "middle": [], "last": "Daniel", "suffix": "" }, { "first": "Wojciech", "middle": [], "last": "Romero", "suffix": "" }, { "first": "Sitaram", "middle": [], "last": "Galuba", "suffix": "" }, { "first": "Bernardo", "middle": [ "A" ], "last": "Asur", "suffix": "" }, { "first": "", "middle": [], "last": "Huberman", "suffix": "" } ], "year": 2011, "venue": "Joint European Conference on Machine Learning and Knowledge Discovery in Databases", "volume": "", "issue": "", "pages": "18--33", "other_ids": {}, "num": null, "urls": [], "raw_text": "Daniel M. Romero, Wojciech Galuba, Sitaram Asur, and Bernardo A. Huberman. 2011. Influence and passivity in social media. In Machine Learning and Knowledge Discovery in Databases, pages 18--33. Joint European Conference on Machine Learning and Knowledge Discovery in Databases.", "links": null }, "BIBREF51": { "ref_id": "b51", "title": "Politicizing the COVID-19 pandemic: Ideological differences in adherence to social distancing", "authors": [ { "first": "Moore", "middle": [], "last": "", "suffix": "" }, { "first": "Allison", "middle": [], "last": "Bihl", "suffix": "" } ], "year": 2020, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Moore, and Allison Bihl. 2020. Politicizing the COVID-19 pandemic: Ideological differences in ad- herence to social distancing. PsyArXiv:k23cv.", "links": null }, "BIBREF52": { "ref_id": "b52", "title": "Politics is wrecking america's pandemic response", "authors": [ { "first": "Jonathan", "middle": [], "last": "Rothwell", "suffix": "" }, { "first": "Chistos", "middle": [], "last": "Makridis", "suffix": "" } ], "year": 2020, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Jonathan Rothwell and Chistos Makridis. 2020. Pol- itics is wrecking america's pandemic response. Brookings.", "links": null }, "BIBREF53": { "ref_id": "b53", "title": "An overview of gradient descent optimization algorithms", "authors": [ { "first": "Sebastian", "middle": [], "last": "Ruder", "suffix": "" } ], "year": 2017, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Sebastian Ruder. 2017. An overview of gradient de- scent optimization algorithms. ArXiv:1609.04747.", "links": null }, "BIBREF54": { "ref_id": "b54", "title": "Threat as a factor in authoritarianism: an analysis of archival data", "authors": [ { "first": "M", "middle": [], "last": "Stephen", "suffix": "" }, { "first": "", "middle": [], "last": "Sales", "suffix": "" } ], "year": 1973, "venue": "Journal of Personality and Social Psychology", "volume": "28", "issue": "1", "pages": "44--57", "other_ids": {}, "num": null, "urls": [], "raw_text": "Stephen M. Sales. 1973. Threat as a factor in author- itarianism: an analysis of archival data. Journal of Personality and Social Psychology, 28(1):44-57.", "links": null }, "BIBREF55": { "ref_id": "b55", "title": "A comparative evaluation of preprocessing techniques and their interactions for twitter sentiment analysis. Expert Systems with Applications", "authors": [ { "first": "Symeon", "middle": [], "last": "Symeonidis", "suffix": "" }, { "first": "Dimitrios", "middle": [], "last": "Effrosynidis", "suffix": "" }, { "first": "Avi", "middle": [], "last": "Arampatzis", "suffix": "" } ], "year": 2018, "venue": "", "volume": "110", "issue": "", "pages": "298--310", "other_ids": {}, "num": null, "urls": [], "raw_text": "Symeon Symeonidis, Dimitrios Effrosynidis, and Avi Arampatzis. 2018. A comparative evaluation of pre- processing techniques and their interactions for twit- ter sentiment analysis. Expert Systems with Applica- tions, 110:298-310.", "links": null }, "BIBREF56": { "ref_id": "b56", "title": "Motivated closed-mindedness mediates the effect of threat on political conservatism", "authors": [ { "first": "Hulda", "middle": [], "last": "Thorisdottir", "suffix": "" }, { "first": "John", "middle": [ "T" ], "last": "Jost", "suffix": "" } ], "year": 2011, "venue": "Political Psychology", "volume": "32", "issue": "5", "pages": "785--811", "other_ids": {}, "num": null, "urls": [], "raw_text": "Hulda Thorisdottir and John T. Jost. 2011. Mo- tivated closed-mindedness mediates the effect of threat on political conservatism. Political Psychol- ogy, 32(5):785-811.", "links": null }, "BIBREF57": { "ref_id": "b57", "title": "A theory of the learnable", "authors": [ { "first": "Leslie", "middle": [ "G" ], "last": "Valiant", "suffix": "" } ], "year": 1984, "venue": "Communications of the ACM", "volume": "27", "issue": "11", "pages": "1134--1142", "other_ids": {}, "num": null, "urls": [], "raw_text": "Leslie G. Valiant. 1984. A theory of the learnable. Communications of the ACM, 27(11):1134--1142.", "links": null }, "BIBREF58": { "ref_id": "b58", "title": "2020. 5 facts about partisan reactions to COVID-19 in the U.S. Pew Research Center", "authors": [ { "first": "Ted", "middle": [], "last": "Van Green", "suffix": "" }, { "first": "Alec", "middle": [], "last": "Tyson", "suffix": "" } ], "year": null, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Ted Van Green and Alec Tyson. 2020. 5 facts about partisan reactions to COVID-19 in the U.S. Pew Re- search Center.", "links": null }, "BIBREF59": { "ref_id": "b59", "title": "Beyond politics-Promoting COVID-19 vaccination in the United States", "authors": [ { "first": "Stacy", "middle": [], "last": "Wood", "suffix": "" }, { "first": "Kevin", "middle": [], "last": "Schulman", "suffix": "" } ], "year": 2021, "venue": "The New England Journal of Medicine", "volume": "", "issue": "", "pages": "1--8", "other_ids": {}, "num": null, "urls": [], "raw_text": "Stacy Wood and Kevin Schulman. 2021. Beyond politics-Promoting COVID-19 vaccination in the United States. The New England Journal of Medicine, pages 1-8.", "links": null }, "BIBREF60": { "ref_id": "b60", "title": "Measuring message propagation and social influence on Twitter", "authors": [ { "first": "Shaozhi", "middle": [], "last": "Ye", "suffix": "" }, { "first": "Felix", "middle": [], "last": "Wu", "suffix": "" } ], "year": 2013, "venue": "International Journal of Communication Networks and Distributed Systems", "volume": "11", "issue": "1", "pages": "59--76", "other_ids": {}, "num": null, "urls": [], "raw_text": "Shaozhi Ye and Felix Wu. 2013. Measuring message propagation and social influence on Twitter. com. International Journal of Communication Networks and Distributed Systems, 11(1):59--76.", "links": null }, "BIBREF61": { "ref_id": "b61", "title": "On predicting Twitter trend: Factors and models", "authors": [ { "first": "Peng", "middle": [], "last": "Zhang", "suffix": "" }, { "first": "Xufei", "middle": [], "last": "Wang", "suffix": "" }, { "first": "Baoxin", "middle": [], "last": "Li", "suffix": "" } ], "year": 2013, "venue": "Proceedings of the 2013 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining", "volume": "", "issue": "", "pages": "1427--1429", "other_ids": {}, "num": null, "urls": [], "raw_text": "Peng Zhang, Xufei Wang, and Baoxin Li. 2013. On predicting Twitter trend: Factors and models. In Proceedings of the 2013 IEEE/ACM International Conference on Advances in Social Networks Analy- sis and Mining, pages 1427--1429. Association for Computing Machinery.", "links": null }, "BIBREF62": { "ref_id": "b62", "title": "Empirical analysis using feature selection and bootstrap data for small sample size problems", "authors": [ { "first": "Yuying", "middle": [], "last": "Zhao", "suffix": "" }, { "first": "Rakkrit", "middle": [], "last": "Duangsoithong", "suffix": "" } ], "year": 2019, "venue": "16th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON)", "volume": "", "issue": "", "pages": "814--817", "other_ids": {}, "num": null, "urls": [], "raw_text": "Yuying Zhao and Rakkrit Duangsoithong. 2019. Em- pirical analysis using feature selection and boot- strap data for small sample size problems. In 16th International Conference on Electrical Engineer- ing/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON), pages 814-- 817. Curran Associates.", "links": null } }, "ref_entries": { "FIGREF0": { "num": null, "uris": null, "text": "The relative importance of variables for predicting fear expressed in tweets, as determined by the XG-Boost model.", "type_str": "figure" }, "TABREF1": { "html": null, "num": null, "text": "Relative changes in movement in public place as compared to an out of sample baseline. For parsimony, only covariates that have a significant influence on the model are included. Note:", "type_str": "table", "content": "
p < 0.1, * * p < 0.05, * * * p <
" } } } }