Audience’s retention of the framing of media coverage of russian constitutional reform in 2020

Algorithm for searching for framing bias in online media on the example of online coverage of constitutional reform in Russia. Specific features of problem lighting. The relationship between impact trends and incentives based on audience retention.

Рубрика Журналистика, издательское дело и СМИ
Вид дипломная работа
Язык английский
Дата добавления 10.08.2020
Размер файла 428,2 K

Отправить свою хорошую работу в базу знаний просто. Используйте форму, расположенную ниже

Студенты, аспиранты, молодые ученые, использующие базу знаний в своей учебе и работе, будут вам очень благодарны.

Размещено на http://www.allbest.ru/

FEDERAL STATE AUTONOMOUS EDUCATIONAL INSTITUTION

FOR HIGHER PROFESSIONAL EDUCATION

NATIONAL RESEARCH UNIVERSITY HIGHER SCHOOL OF ECONOMICS

Faculty of Communications, Media and Design

Audience's retention of the framing of media coverage of russian constitutional reform in 2020

Syagova Valeriya

BACHELOR'S THESIS

ADVERTISING AND PUBLIC RELATIONS

Educational Program 42.03.01 Advertising and Public Relations

Supervisor

Lecturer,

Chmel Kirill

Moscow, 2020

INTRODUCTION

online media framing constitutional reform

Stories news media deal with on a daily basis are often quite complex, to an extent that it is impossible to convey the entirety of the situation in just one article. Hence, it is only natural that some aspects of reality end up being more emphasized in a text, while others remain overlooked. Framing is a recognized phenomenon of constructing a message with more attention devoted to some parts of the story over the others, whether it is done consciously or unconsciously (Entman, 1993). The usage of framing assumes making a certain perspective more salient in a text communicated to the audience, and thus has both sociological (Goffman, 1974) and psychological implications (Kahneman & Tversky, 1984). Exposure to framing is known to have effects on the process of decision-making. Decision-maker's conception of the choice and what it entails is influenced by the way the problem is formulated. Studies show that even seemingly insignificant changes in the formulation of choice problems can cause preferences to shift (Tversky, Kahneman, 1981). And precisely for this reason, it is critical to identify framing bias in situations that deal with decision making, such as media coverage during elections and referendums.

There have been many studies aimed at identifying different kinds of media bias. Generally, they represent two main approaches: audience-based and content-based methods. As for the content-based approach, some researchers chose to pay close attention to the variety of think tanks newspapers and politicians were citing, and the way they related to each other on a political spectrum (Groseclose & Milyo, 2004). Others focused on the choice of topics and their prevalence in different media publications, while still considering their political alignment (Puglisi, 2004).

In general, the use of language (found in citations, topics or headlines) has been considered to be a prevalent indicator of media bias in many studies. For example, Gentzkow and Shapiro suggested a new index for measuring media slant, that estimates how often media outlets use language meant to sway the audience to a side of a political spectrum in regard to a given issue. By amplifying methods used in previous research, they have been able to analyze a larger amount of media outlets, than before. They have identified so-called `partisan phrases', those for which the frequency of use differs depending on whether they are used by a Republican or a Democrat. Measuring the usage frequency of these expressions in newspapers provided the basis for a creation of said index. And though they still relied on the Republican vs Democratic dichotomy, they pointed out that the index does not offer a category that would be considered entirely objective or `unbiased' (Gentzkow, Shapiro, 2010). Utilizing audience-based approach, this study also takes into consideration the role of the media audience and how its beliefs and ideological leanings correspond with slant detected in newspapers. Those featuring liberal (or conservative) language generate more sales in ZIP code areas that tend to be more liberal (or conservative). The factor of outlet ownership, on the other hand, was found to be insignificant, as the comparative slant of co-owned media outlets was deemed inconsequential.

In another study dedicated to measuring the influence of ownership on media bias, it has been proposed that focusing on issues that uncover media outlets' different interests is more effective than conducting the analysis from the standpoint of political leanings. The reason for that is that there is no objective standard that could help us determine the benchmark of a politically impartial publication (Gilens, Hertzmann, 2000).

There has also been a study focusing entirely on issue-filtering through following which stories from wire services such as Reuters and Associated Press were chosen by several news outlets to present (Baum and Groeling, 2008).

Media bias could also be measured by means of headline attention. In their study of news coverage, Lott and Hassett classified economics themed newspaper headlines as positive, negative, mixed or neutral and compared them to the actual data presented in the articles, while also distinguishing between different political leaderships contemporary to each publication (Lott, Hassett, 2004, revised in 2013).

However, content-based studies tend to focus on a preselected portion of news articles, which significantly confines the volume of relevant material. For instance, language indicative of strong party sympathies appears only in a minority of all stories (Budak, Goel, Rao, 2016). Recognizing these limitations, Budak, Goel and Rao conducted a research that combined machine learning and traditional manual article assessment. Larger scope of articles allowed them to analyze the instances of issue-filtering and issue-specific slant, alongside the usual outlet-level slant. The results diverge from those of previous research - the reporting style of major news outlets of different political leanings was found to be quite neutral and homogenous. It is mostly due to the fact that, in this case, even articles that did not feature strong indicators of political leaning were included in the sample.

Headline attention research has also been reintroduced with the usage of NLP, that allowed to automate the process of bias detection (Reddy, Dugenpudi, Mamidi, 2019).

The study this work drew most inspiration from analyzed news articles examining a polarizing topic expected to elicit framing bias. All of the main topics were manually assessed and coded, for researchers to later identify the polarity with which every topic was described in an article using a machine learning classifier (Morstatter, Wu, Yavanoglu, Corman, Liu 2018).

However, technological advancement allowed not only for new automated methods to thrive, it may have also contributed to the polarization of media audiences. The emergence of a wide variety of online media outlets makes it possible for readers to search only for those articles that coincide with their already held views (Iyehngar, Hahn, 2009).

Empirical studies of biased exposure to information, labeled confirmation bias, predate the rise of online media. A lot of them focused on sequential information processing, which suggests acquiring and processing information in a sequence, until the need for information is stalled (Geller and Pitz, 1968), (Godden, 1976).

A more recent study was comprised of a number of experiments, during which participants were exposed to a set of articles exploring different opinions on the same policy in order to arrive at a decision (Jonas, Schulz-Hardt, Frey, Thelen, 2001).

This research is aimed at applying best practices of identifying framing bias in online media as well as confirmation bias among readers in order to analyze the coverage of the upcoming constitutional reform in Russia and audiences' awareness. It is likely to evoke framing bias, due to the fact that the coverage is bound to be selective considering the complexity of the issue, as the reform comprises of many controversial initiatives.

The goal of this research is to identify framing bias, or lack thereof, in online media coverage of the constitutional reform in Russia and find out whether audience's selective exposure tendencies affect their retention of information.

The objectives are as follows:

1) To develop the theoretical framework in accordance with which we can search for the instances of framing bias in online media;

2) To explore the specific characteristics of the coverage of our issue of choice;

3) To see if exposure tendencies relate to audiences' retention based incentives.

Therefore, the problem of the present research is the tendency of both framing bias and selective exposure bias to influence readers' retention of information in the face of political decision-making.

Research Question.

Does the online media coverage of the constitutional reform in Russia feature instances of framing bias, while audience in turn experience selective exposure bias?

H1. Framing bias is present in the coverage of the constitutional reform in Russia.

H2. Audiences' selective exposure is connected to their judgements regarding political decision-making.

Methods

This research is aimed at identifying both framing bias in online media using computational language analysis and selective exposure among readers through analyzing the connection between the tendency to exhibit it and making judgement related to politics.

First, the headline dataset was compiled using the ParseHub software. Said database consists of publications dealing with the upcoming constitutional reform in Russia published from January 15th until April 15th, 2020. Articles have been collected from 8 online media outlets that are in the top-30 most cited Russian online media outlets (according to media analysis company `Медиалогия' for 2019). Those publications comprise of: Fontanka.ru, 47news.ru, Gazeta.ru, Rg.ru, Lenta.ru, МБХ Медиа, News.ru and Rbc.ru. Then the headlines' content has been analyzed using Latent Dirichlet Allocation (LDA) method, specifically Gensim library using Python, to identify their underlying themes and topics.

The connection between audience's selective exposure tendencies and their judgement making has been studied with the help of the data provided by WCIOM. From 29th until 31st of March, 2020 they held a survey 1600 Russians above the age of 18 participated in. It was a telephone-based survey developed with two stage stratified random sampling. The phone numbers have been retrieved from the exhaustive list of functional Russian phone numbers. The data has been weighted in accordance with demographical parameters.

This data has been analyzed with the usage of R software for statistical computing.

Theoretical Basis

As the present study is aimed at analyzing the occurrence and possible effects of media framing, first we have to define the term itself. The concept of framing assumes that the manner in which a certain issue is presented in the news can influence the way it is understood by the audience (Scheufele & Tewksbury, 2007). Essentially, it is the way of shaping a message with more attention devoted to some parts of the story at the expense of others (Entman, 1993). Framing is a widely recognized phenomenon in psychology and sociology alike, as its effects and influence lie in the areas of interest of both sciences (Goffman, 1974; Kahneman & Tversky, 1984). Since this research deals with the analysis of media coverage preceding a popular vote, the one effect of framing that is of most interest to us is its ability to influence decision-making. According to psychological studies, decision-maker's conception of the choice presented to him is largely affected by the way the choice problem is formulated. Studies show that even slight alteration of the formulation of choice problems can change person's preferences (Tversky, Kahneman, 1981). Precisely due to the role framing takes in the decision-making process, numerous studies have been conducted with the aim to identify it, especially during elections and referendums. Since there have been two main actors in this process, media and their audience, two main approaches emerged: content-based and audience-based respectively.

However, these studies vary even within a single approach. There have always been different views as to what type of content is to be subjected to analysis. Some researchers decided to focus on think tanks, that politicians and newspapers that side with them, tend to cite (Groseclose & Milyo, 2004). Others payed close attention to media's choice of topics, while also dividing those topics in accordance with bipartisan political alignment, since different variations of the topics comprise both parties' political agendas (Puglisi, 2011). The `partisan' approach has been widely utilized in general. For instance, Gentzkow and Shapiro presented an index for measuring media bias that estimated the frequency with which media outlets used language meant to sway the audience to one side of a political spectrum. By identifying so-called `partisan phrases', those prevalence of which differs depending on whether they are used by a Republican or a Democrat. However, they relied on the bipartisan dichotomy, and pointed out that the index does not offer a benchmark for an `unbiased' publication (Gentzkow, Shapiro, 2010). They have also incorporated an audience-based approach in this very study. Analyzing newspaper sales by ZIP code areas, they have found out that residents of areas, that tend to skew conservative, gravitate towards conservative media, same goes for the supporters of Democratic party. The factor of outlet ownership has also been taken into consideration but has been deemed insignificant after all.

However, in another study concerning the influence of media ownership on bias it generates, it has been proposed that focusing on polarizing issues, that uncover media outlets' different interests is more effective than analyzing their political leanings. It is mostly due to the fact that there seems to be no tangible standard in accordance with which we could identify a politically impartial publication (Gilens, Hertzman, 2000).

Some researchers have decided to focus their content-based studies on headlines rather than on full articles, due to the fact that they create a stronger image in reader's mind. In a study of economics news coverage, Lott and Hassett classified newspaper headlines as positive, negative, mixed or neutral, while comparing them to the actual content of the articles. In accordance with previous research, they have also payed attention to the difference in political leadership, that was contemporary to each publication (Lott, Hassett, 2004, revised in 2013). With the emergence of Natural Language Processing (NLP) methods, headline attention approach has been revolutionized, as the process of analysis became largely automated and efficient (Reddy, Dugenpudi, Mamidi, 2019).

It is true not only for headline-based studies, but for other kinds of content-oriented research as well. For instance, a study of issue-filtering and issue-specific slant that applied machine learning alongside traditional manual article assessment produced unusual results. It showed that the reporting style of major news outlets is somewhat neutral and homogenous, despite the difference in their political leanings. It is mostly due to the fact that new methods have allowed for bigger samples of articles. And those samples have not included the language of strong political leaning exclusively, as it was done before to improve the accuracy of assessment (Budak, Goel, Rao, 2016).

As this study deals with decision-making process, it is important to pay attention to the way audience's process information they seek to make said decision. The emergence of online media has greatly added to the number of existing media outlets, presenting audiences with an unprecedented variety of options. However, it is precisely what makes it possible for readers to pick only those that correspond with their own beliefs (Iyengar, Hahn, 2009). In other words, makes them subjected to exposure bias, the studies of which predate online media. Many of them took into account the concept of sequential information processing, which suggests acquisition and processing of information in a sequence, that takes place until the need for information is fulfilled (Geller and Pitz, 1968; Godden, 1976).

СHAPTER ONE

The conceptual basis of this research relies heavily on such theories and concepts of psychology, sociology and communication as `framing theory', `media bias theory' and the phenomenon of `sequential information processing'. Both theories describe factors the construction of a message is influenced by. The concept, in turn, refers to the specific nature of the way people process information when searching for new knowledge when provided with choice. A detailed analysis of all of these concepts follows.

Framing

The studying of framing in the context of political communications in media has undergone some visible changes in perspective. The most recent and relevant ones deal with the transformation that mass media experienced in the later 20th century. Previously concerned with questions of ideology influencing mass media content, scholars began paying more attention to the process of constructing a news message. The unique process of such a construction makes for a very certain image of `reality' that does not necessarily portray the issue in its entirety. Quite the opposite, it is comprised of the elements of the factual basis of information, that are connected within the story through the usage of a certain frame. The formation of such a frame also depends on the genre of a publication as well as on a political leaning tendency of the content producer. All parts of the news story that are highlighted within a frame are given a certain value within the text, they are prioritized. Through construction of the story in media comes the construction of the story within the perception of a single person, and the way they may view reality. Here framing enters the realm of political effects, due to the fact that the studying of framing was based on a concept of mass media being a strong and powerful influence on one's views and attitudes but also being under the influence of pre-existing values and attitudes of their audiences. The latter is very influential because it affects the manner in which media audiences select and process information offered them by media. (McQuail, 2005).

As many other theories of mass communication, framing theory has sociological roots. It was established that it is people's natural incentive to make sense of all the important developments and occurrences that are happening in the society's public life at large. Processing such vast amounts of information is a challenging task, so, in order to raise efficiency, people employ different `schemas' and `frameworks' that are meant to make the information more orderly. It also makes for a better understanding of the meaning of such information (Goffman, 1974).

Essentially, the theory of framing states that the way different elements of the reality are portrayed in the news if going to have an effect on the way media audiences make sense of the story (Pan & Kosicki, 1993).

Framing is also recognized as an important psychological phenomenon, due to the fact that it strongly influences decision making process. For example, Kahneman and Tversky introduced a `decision problem'. It implies the existence of a set of options to choose from, the overview of the possible outcomes and consequences of a decision. `Decision frame' is, in turn, referring to a general idea that actors have about the decision they have to make: what results a certain choice is going to yield. Such a frame is influenced both by the formulation of the decision problem and by the specific characteristics of person's background and belief system. The formulation of a decision problem can often be executed in multiple ways. And this choice of formulation will ultimately influence the decision making process and thus the decision that is being made (Kahneman and Tversky, 1981).

Thus, framing exists at both micro- and macrolevels (Scheufele, 1999). Macrolevel framing suggests a way the information is presented in the media. However, this choice is often based on the knowledge of `schema' their audiences operate with. Framing of a message is usually executed in accordance with this knowledge (Shoemaker & Reese, 1996). However, it does not necessarily mean that a medium has a certain agenda that influences its content or that it goes beyond the boundaries of journalistic integrity to cater to their audiences. What it does mean though, is that framing is a useful instrument for making a message digestible, especially when it comes to complicated political and societal issues. The usage of framing makes for an efficient presentation that works with their audiences `schemas' in order to highlight the crucial parts of the message at hand (Gans, 1979).

Microlevel framing refers to the way audiences themselves process the given information to form opinions and make decisions. The acceptance of a frame presented by the media is largely influenced by cultural norms and other aspects of social life (Gamson & Modigliani, 1987). This process is commonly referred to as `frame building' and it unfolds within a certain culture and is shaped by it. The creation of a news message is only a part of a framing chain and is not the single most influential element of it (Scheufele, 1999).

The processing of the given information plays a crucial part in this process. So much so, that information processing has a theoretical basis of its own. It is established that audiences are most likely to internalize a message when presented with information that they actively engage with. Thus, the presence of an issue is more visible when media audiences actively engage with the messages highlighting it (Eveland, 2004).

However, this effect may not always be in place, due to the fact that media audiences are not homogenous in their views, beliefs, interests and media preferences. Consistent engaging with framed messages is more likely to have an influence on audiences that are not especially interested in a discussed issue. Knowledgeable audiences, on the other hand, prefer the systematic processing of information, within which they can compare different frames to form an educated opinion. Moreover, the usage of certain frames may yield the results opposite to the ones previously anticipated. It can go as far as to cause audiences to side with an oppositional frame, especially if the original frame is weak (Chong, Druckman, 2007).

Other studies have shown that the framing process is incredibly complex, so much so, that readers' retention of information may even be the result of internalizing multiple or even competing frames. A study describing such a phenomenon was conducted regarding the perception of of 9/11 terrorist attacks and the war in Afghanistan. Audiences' frames did not just align in a sequence of separate frames: a one describing a national tragedy and the other one referring to warfare. Both of those frames morphed into each other, creating a context in which processing of the September 11 events was crucial for the formulation of an opinion about the Afghan War. The original view on framing suggests that the influence of a framed message is only defined by the way an issue is presented in the media. However, pre-existing interpretative `schemas' audiences' possess tend to influence the effect of a framed message in a very significant way. The effectiveness is dictated not only by whether we encounter a certain issue in media for a significant period of time, the crucial element is how this encounter unfold depending on our pre-existing notions and beliefs (Eddy, Meirick, 2007).

There is also another angle when it comes to working with multiple frames. Two concepts connected in one message may be perceived as interconnected irrespective of the context they are built in (Price & Tewksbury, 1997). To better illustrate this, Scheufele and Tewksbury present the interconnected concepts of tax policy and unemployment rate. We may stumble upon the connection of these two terms in the news media. The message presented to you may suggest an opinion that determining whether or not higher or lower taxes should be more preferable depends on whether you strive for lower or higher unemployment rate. By doing this it establishes that questions relevant to unemployment are also relevant to the issues of tax policy. And it does not necessarily require deeper knowledge about unemployment. Sufficient knowledge of unemployment issues would contribute to the overall understanding of the topic, if the news message itself does not provide this information. Therefore, if media audiences are not familiar with the concept and the news message does not give exhaustive explanation of it, a frame that is build is not going to be very effective. The effects of a framed message are going to differ depending on whether the concepts it consists of already exist in audiences' memory or not (Scheufule, Tewksbury, 2007).

Information about isolated issues, however, are more likely to be processed differently, with more focus to the actual content of a news message and its factual basis, as well as questions and solutions presented in the article itself (Zaller, 1992).

Media bias

Along with framing, media bias is another widely recognized phenomenon in media research. And just like framing it can be motivated by several different incentives. Generally, media bias is referred to as a disturbance of balance that results in certain aspects of a news story appearing more salient than the others. However, just like agenda setting, media bias can have an ideological component as well, relating mostly to the political leanings of news producers themselves. Media bias can also be a result of distortion of information, whether motivated or not. However, it is difficult to navigate what information can be considered `distorted' and which one should be seen ad objective and unbiased. For example, public evaluation of the nature of bias concluded, that people consider bias to be the manner in which a news message is constructed and reported, but only if it is a highly charged one. Many also see media bias as an attempt at persuasion. The studies have also found that the more knowledge audiences possess about a certain issue, the more scrutiny all the other messages on this topic are under (ASNE, 1999).

There have been many theories speculating about the origin of media bias. For instance, engaging with certain topics is only possible if there is a platform provided for their presentation. And the control of such platforms usually belongs to news organizations. From an economical perspective it is more efficient to provide a platform for issues that are of interest to the public. Incomplete nature of information does not always play the crucial part in the formation of bias (Stromberg, 2004).

Another theory that focuses on reader preferences suggests that publishers may succumb to media bias by attempting to highlight certain aspects of a news story in order to appeal to their main audiences and their previously held beliefs, thus confirming them (Mullainathan & Shleifer, 2003).

In general, financial incentives as well as a structure of news production process are of great importance when it comes to formation of media bias. Ideological stance of the owner of a publication also plays a significant part when it comes to bias in media (Bovitz, 2002).

Thus, media bias is constructed within a complicated system of actors. And the most active ones are bound to influence the direction a news story will ultimately take, especially when it comes to bipartisan issues, dealing with which journalists usually tend to engage with a single influential group (Groseclose & Milyo, 2004).

When it comes to reporting on electoral politics, media outlets exercise media bias that is especially visible, because even if a ruling party changes, partisan media bias foes not (Puglisi, 2004).

The structure of a news article can also influence the salience of its message. For example, it has been noted that headlines tend do highlight more positive messaging than the articles themselves (Lott & Hassett, 2004).

Overreporting about an issue can also be considered a variation of media bias. Thus, an issue gains salience because of intensity and repetition of presence. Studies show that media audiences tend to follow the lead of disproportionate salience (Ansolabehere, 2004).

And though the consensus about the influencing power of media bias is consistent, multiple different methods have been utilized in order to identify it. Some researchers have followed the `audience based' approach, inspecting the reader-base of different media outlets (Munson & Resnick, 2010). Others have decided to pay more attention to media's sourced of information. Thus, media bias was identifiable through different `think tanks' outlets often cited (Groseclose & Milyo, 2005). Automated sentiment analysis has been applied not only to the articles themselves but also to the audiences' commentary. However, such research requires pre-labeled data, which is oftentimes difficult to obtain (Guerra & Veloso, 2011).

Overall, identifying the way a message is phrased in a news report or what issues have been made more salient may lead to the identification of media bias.

Selective exposure

Considering that mass media are not homogenous by nature, the specific nature of a person's media consumption may be connected to the nature of their political leanings, to influence it or to be an expression thereof. s than others. In particular, those viewing FOX news were more likely to believe in both the link and the weapons while those watching PBS and listening to NPR were less likely (Kull et al. 2003-4). The implications of this finding are troubling: Varying exposure to news media makes for a development of different views and beliefs regarding political and societal changes. If there is no widespread general consensus for an assessment of public policy matters, it leads to the existence of polarizing views and opinions shared among different societal groups. (Kull, 2003-4).

Media exposure influences the development of beliefs, however the factors that influence exposure itself are not as clear. However, one of the more visible and identifiable influences if the emergence of online media outlets that suggest a vast variety of presented views and opinions, never seen in the traditional media before (Iyengar, Hahn, 2009).

The variety of today's media is so unprecedent that one can successfully utilize media without encountering political messages or even specifically aim to gain knowledge about specific issue (Atre and Katz 2005, Galston 2003). It is often related to the phenomenon of partisan selective exposure which suggests focusing on media that supports one's already held political views, especially regarding the support of certain political parties or various polarizing issues. An ever-increasing supply of news media makes it much easier to search for an outlet that expresses beliefs similar to audiences' very own. Thus, the instances of selective exposure increase as well. In this case, audiences acts customers who tend to reach out for a media product that is more favourable to them (Mutz and Martin, 2001).

However, not everyone agrees that the effect of selective exposure is so influential. It has been noted that people themselves do not necessarily intentionally look for information that would support their beliefs (Kinder, 2003).

Some researchers also do not consider the effect of selective exposure to be the prevalent one when it comes to the choice of media outlets. However, they note that the instances of selective exposure occur under certain unique conditions that do not necessarily have something to do with the attempts at setting an agenda (Zaller, 1992).

It contrasts with the general consensus on the nature of information seeking, which suggests that looking for new information, one usually favours whichever one doesn't oppose his own vies and predispositions (Jonas, 2005).

Generally, the views on selective exposure have always varied. It is especially notable when it comes to research. There have been multiple studies that noted a connection between media consumption and audiences' beliefs and predispositions. However, the nature of the connection remained largely unclear (Best, 2005; Chaffee, 2001).

Though many researches have studied the instances, where it is more obvious that individuals preferred the sources of information that corresponded with their previously held beliefs and predispositions (Donsbach, 1991; Redlawsk, 2002; Taber & Lodge, 2006). There also have been some publications expressing opposing views on the subject at hand, stating that the influence of selective exposure cannot be considered crucial when it comes the choice of media outlets (Meffert, 2006). Such a difference may suggest the existence of other interfering factors that determine whether one is going to succumb to selective exposure or maintain a less rigid approach towards his media consumption habits.

As has been established before, theoretical basis describing the selective exposure phenomenon poses that people's previously held views determine their media consumption habits. However, not every such view or predispositions is capable of navigating the choice of a media outlet. Therefore, some of those predispositions have more capability to lead the decision-making process.

From a psychological standpoint, those views and beliefs that hold some kind of personal relevance are the ones that are most likely to pose as influential when it comes to formation of media preferences. Some of those beliefs have such a presence that they remain salient irrespective of the context they are accessed within (Price & Tewksbury, 1997).

For example, partisan political leanings can serve as an example of a belief with a very strong presence (Green, 2002). Unlike with other subjects, when it comes to partisan political leanings, audiences tend to reach out for those outlets that do not oppose their position on a political spectrum, precisely because it holds such a great personal relevance to them. It has also been noted that topics that evoke a strong emotional response may influence the choice process. It is mostly due to the fact that media audiences not only seek new information to gain knowledge, but also to find sentiments that resonate with their own emotions (Valentino, 2007). There is evidence that a stimulus capable of evoking a strong emotional response can make people with partisan views exhibit selective exposure patterns (Taber & Lodge, 2006). Moreover, political topics themselves are known to evoke strong emotional response in general. People with strong views on politics tend to engage with media that resonates with their beliefs in a positive way, even more so when a topic is relevant to one's identity or personality. Thus, the effects of selective exposure are stronger with those whose political views and predispositions are strong as well (Marcus, 2000).

The tendency to aim for information that stands in support of one's belief system predates the age of digital media and has been an object of research in cognitive psychology for many years. The incentive behind it was the one of avoiding dissonance and conflict of previously held beliefs with the ones present in media (Festinger, 1957).

However, such research was often held in regard with political campaigns and not necessarily political news content. Thus, it was often connected to the appeal of a particular candidate or a party and not the messaging itself (Sears & Freedman, 1967). Other researchers have considered this phenomenon to be the reasoning behind the repetitive nature of political campaigns and effects that they exhibit (Klapper, 1964).

However, the difference between exposure motivated by one's desire for political action and exposure explained by one's social situation remained largely out of focus. Exposure determined by social situation may be shaped by factors such as income, neighborhood's political leaning, family relations, friendly and work-related relations etc. (Cotton, 1985). When it comes to consciously avoiding exposure, researchers have found that it is not a strong enough motive when it comes to shaping media preferences and seeking new information (Sears, 1968). However, there have been studies identifying motivated selective exposure regarding bipartisan political leanings (Iyengar & McGrady, 2007). It is to be mentioned though, that other researchers have not identified the same effects in similar studies. (Chaffee & Miyo, 1983). As was mentioned before, the nature of media preference can also rely on other external factors that do not necessarily connect to one's political leanings. Such an effect was measured through presenting people with information they had to navigate through based on a previously made judgement about a topic (Jonas, Schulz-Hardt, Frey & Thelen, 2001). There is also evidence to suggest that media audiences are becoming more selective in their media preferences. Even audiences with strong partisan bias have grown to detect such bias in media they consume (Smith, Lichter & Harris, 1997). Thus, political news audiences are motivated to seek not only new information sources, but new media in general, outside of the realm of television and traditional press. These preferences can also suggest bipartisan differences, due to the fact that people may move on to formats more suitable to the expression of their own political beliefs, whether it is going to be a radio show or a talk show on television (Pfau, Houston & Semmler, 2007). Therefore, selective exposure bias it stronger when there is an element of partisanship involved, as well as an influence of strong emotional response to a given issue. There is also a notable effect of a newfound variety of media sources and formats that makes for a larger supply side thus creating an environment for making a choice of media tailored to one's views without sacrificing the general level of awareness.

Sequential Information Processing

As have been stated previously, the phenomenon of people choosing to engage with information that does not directly oppose their previously held views and beliefs have been recognized in the field of psychology for many decades of research work. But what exactly do people stand in support for when it comes to a certain media message? Research shows that it could be a certain expectations of a consequence, a shared attitude or belief, a stereotype or a cultural norm and an affirmation of one's predisposition. Looking for these aspects in media one consumes may lead to conservation of their beliefs even if sources of opposing information are accessible to them (Johnston, 1996; Pinkley, Griffith & Northcraft, 1995).

Recognizing these trends holds special importance when it comes to making decisions that are outside the realm of mundane choices. Dealing with a choice that suggests significant consequences may be complicated by such selectiveness due to the fact that it does not help to form an adequate judgement. Crucial details may be ignored when under the influence of a rigid belief system. The inability to see the overall picture when dealing with an issue may lead to a choice of a wrong decision and potential losses (Brockner & Rubin, 1985). Moreover, it has been noted that leaning on previously held beliefs and predispositions can yield less preferable results when it comes to solving decision problems and all in all lead to a higher risk of failure (Schulz-Hardt, 1997). Such an effect makes detecting such biases crucial for decision-making, what could serve as a driver for it and how it can be dealt with in a more efficient way.

As it has been stated before, selectivity is oftentimes a product of dissonance, or a desire to avoid it (Festinger, 1957). Dissonance theory poses that, when holding a certain position, a person is less likely to give it up for an opposing one to avoid dealing with a conflict in the aftermath of making a choice. This phenomenon has also been recognized as `confirmation bias'. The instances of such a bias are not only present when it comes to making a decision, it is also relevant to multiple stages that precede final decision-making and shaping of source preferences, as well as making initial judgements (Schulz-Hardt, 1997). Moreover, it if a phenomenon relevant not only to different stages of individual decision making, but most importantly, to decision-making in groups as well. That is an aspect that, in theory, makes it crucial for the analysis of electoral trends and politics in general (Schulz-Hardt & Frey, 2000).

However, it had been up to debate whether these findings are indeed applicable to media audiences' behaviour and are they valuable when me make judgements about audiences' actual political decisions and decisions regarding their informational preferences. Precisely for that reason, some researchers have compared the dynamics of controlled experiments with the assistance of students with the dynamics of an actual decision-making process in investment banking (Schulz-Hardt, Frey, 2000).

Thus, the applicability of the previous findings was confirmed once again. However, it was still unclear whether the nature of exposure to information within an experiment can be compared to that occurring in the more natural circumstances, native to media audiences.

In circumstances regarding making a political or a financial decision of significant importance one rarely has adequate, educated opinion at hand, available for inspecting (Vertzberger, 1990). The information found does not always meet the standards of relative objectivity or even basic credibility and it is usually portioned. The search process keeps reiterating to the moment when one finally arrives at a decision that he wants to make. Therefore, it remains unclear how many iterations or information sources it is required to provide to accurately measure the specifics of a decision-making process. What is more important, this process is in itself full of many micro-decisions that arise when a person is confronted with conflicting information, there is always a choice of whether to engage with it or whether to avoid it altogether. This process continues right until the moment a decision is made. This phenomenon is commonly referred to as `sequential information seeking', which states that the search for new information always comes in a form of a sequence.

The studies of sequential information processing have included those that dealt with the phenomenon of selective exposure. The nature of audiences' media consumption pattern was evaluated and later compared to political views they tend to hold. Their information seeking preferences have included preferences for entire media, such as tendency to watch television or read the news only in newspapers (Sweeney & Gruber, 1984).

Generally, aforementioned studies show that information seeking functions as a process that consists of many stages and is shaped by one's choice of media or even a preferred format.

CHAPTER TWO

Identifying Framing Bias: Design The goal of the present research is identifying media bias in publications regarding the constitutional reform in Russia. The presence of media bias does not necessarily suggest siding with an agenda. It is usually a result of making a complicated issue more digestible for media audiences by making certain points more salient in the text, than the others (Gentzkow, Shapiro, 2010). Therefore, we define a frame as a point that gained salience in the process of publishing a news story. In other words, it is this part of a complicated issue that we encounter more often and in a larger number of articles.

Every such point is essentially a topic, a set of which constitutes an entire issue that media tends to simplify. If we are treating these points like topics, we can identify them using methods specifically designed for discovering topics in media articles and texts at large. Thus, to identify a set of frames media publishers operate with while talking about an issue, we are going to use Latent Dirichlet Allocation (LDA) model, or topic modelling.

An LDA model is a statistical model that helps to identify underlying themes and topics in a corpus of texts. It suggests the existence of a certain structure in a text that determines the appearance of certain words on a regular basis compared to the others.

For this task we have utilized Gensim library, a Python based library commonly used for natural language processing, specifically unsupervised topic modelling. The choice of unsupervised topic modelling is partially determined by the inaccessibility of relevant labeled data. The entirety of workflow has been processed with the usage of Google Colab.

Identifying Framing Bias: Data

The dataset comprises of news article headlines related to the constitutional reform in Russia, dating from January 15th, 2020, the date of their announcement, and until April 15th, 2020. All of the articles mention the reform, which is referred to as `поправки в конституцию' in every single article.

Headline attention is a method that has been widely used in both computational and non-computational research regarding identifying framing bias (Lott, Hassett, 2004, revised in 2013).

The headlines have been collected from the articles published in 8 media outlets that are a part of top-30 most cited online media outlets in Russia in the year 2019. The rating is provided by a Russian media research company `Медиалогия'. The articles have been collected from the following online media sites:

- Fontanka.ru

- 47news.ru

- Gazeta.ru

- Rg.ru

- Lenta.ru

- МБХ Медиа

- News.ru

- Rbc.ru

Potential leaning and biases of each publication are not relevant to this body of work, due to the fact that the main factor is their widespread circulation among other media outlets because they are the most cited ones.

All of the news headlines related to the constitutional reform in Russia have been collected using ParseHub software. Later the datasets of each online media outlet have been compiled into a single CSV file comprised of 1544 entries overall.

Identifying Framing Bias: Analysis

1). We begin our data analysis with preprocessing the acquired data. First, the text should be rid of all punctuation.

2). We are also lowercasing all of the words, so they appear in their `normal form'. Then follows lemmatization, which is necessary for verbs changing into their present tense and nouns being changed into ones that are in the first person.

3). However, it is still required to remove the stop-words from the news article headlines. Stop-words are commonly used words that do not necessarily possess any meaning useful for the analysis that requires knowledge of the context in which the words are used. They usually include numerals, interjections and prepositions. However, for the sake of our analysis, a few additional ones were included due to the fact that during the initial run they occupied most of the context given without adding any meaningful information or helping to indicate a topic. Among those words are `конституция', the Constitution itself, a most used term that strongly shifts the overall results.

4). The next stage of preprocessing data is removing the numerals and tokenizing the given text. Tokenization is a process of tokenizing or dividing a text into a number of tokens. One token is a part of a larger body of text separated from it and functioning as a singular element in a list.

5). Then we proceed to access the Gensim library that we are going to use for identifying the underlying topics in the text. Moreover, we vectorize our text data to bring the meaning of words in the realm of data that is computable. The original dictionary was comprised of 2071 unique tokens, while the filtered one is comprised of roughly 330 unique tokens.

6). Then we can proceed to training the given model.

7). The preferred number of topics is selected to be extracted from out textual data.

Thus, we have gotten the number of words that identify the underlying topics. The first topic most likely relates to the initial stage of the reform, due to the fact that it includes such words as `government', `minister' and `resignation'. This may be connected to the resignation of Russian government that occurred during the same timeline as the initiation of the constitutional reform.

('кремль', 0.1265235),

('поддержать', 0.072394535),

('право', 0.039507896),

('отставка', 0.03684118),

('член', 0.028160019),

('создание', 0.026434278),

('решение', 0.023002647),

('призвать', 0.022744656),

('правительство', 0.022267027),

('министр', 0.021948531)])

The second topic refers to the task force organized around the constitutional reform to discuss its contents and propositions. Not only it mentions the task force itself, it also mentions the senator participating in its work, as well as the main electoral body of Russia.

[('группа', 0.09823255),

('рабочий', 0.07490455),

('срок', 0.059916195),

('президентский', 0.04426208),

('россиянин', 0.043276463),

...

Подобные документы

  • Theoretical basics of Internet advertising. The analysis of the media planning process. The establishing media objectives through developing media strategies and tactics. The effectiveness of the media planning in Internet. The example of the media plan.

    курсовая работа [64,2 K], добавлен 25.03.2014

  • Consideration of the mass media as an instrument of influence on human consciousness. The study of the positive and negative aspects of the radio, television, press, magazines, Internet. Advantages and disadvantages of the media in the Great Britain.

    дипломная работа [2,3 M], добавлен 14.10.2014

  • "The Bauer media group". "The Bertelsmann" is a German multinational mass media corporation. "The Axel Springer Verlag". The German media industry. Company that is specialised in production and delivery of media in the form of digital, audio, video.

    реферат [18,9 K], добавлен 13.03.2014

  • The role of mass media in modern life. The influence of newspapers, magazines and television in mind and outlook of the mass of people. Ways to provide information and display the news of dramatic events, natural disasters, plane crash, murders and wars.

    презентация [730,5 K], добавлен 17.05.2011

  • Понятие, определение и специфика социальной журналистики в "small media". Анализ социальной тематики, базовой структуры малых медиа, линейной схемы коммуникации. Принципы существования малых медиа, их распространение по разным мультимедийным платформам.

    курсовая работа [228,8 K], добавлен 06.05.2018

  • Analysis of the publishing content. Relationship of international relations and the complexity of editorials in periodicals wider audience. The similarity between international relations and newspaper editorials in the western and communist countries.

    статья [21,3 K], добавлен 23.02.2010

  • Influence of television on modern political practice. Nature of media power and its impact on political system of society, its character, practice and institutions. Dangers of new mediated symbolic politics for the democratic political practices.

    реферат [25,0 K], добавлен 28.05.2012

  • Особенности тенденции конвергенции СМИ в целом и явления, к которым приводит эта тенденция. История и направления деятельности медиа-холдинга "РосБизнесКонсалтинг". Реализация концепции конвергенции издательским домом Independent Media и "КоммерсантЪ".

    курсовая работа [99,7 K], добавлен 12.11.2010

  • Описание явления социальных сетей и современной ситуации на соответствующем рынке. Изучение видов взаимодействия в интернете и взаимодействия различных типов аудитории в социальных сетях. Рекомендации по продвижению СМИ на примере журнала "Катрен-Стиль".

    дипломная работа [2,6 M], добавлен 20.06.2014

  • Прецедентні феномени як один із виявів інтертекстуальності. Ідіостиль електронного ЗМІ, його порівняння зі стилем друкованого видання. Прецедентні ситуації, висловлювання й імена, особливості їх використання в текстах інтернет-видання "Online-Експрес".

    дипломная работа [102,3 K], добавлен 24.05.2015

  • The Time is a British newspaper national newspaper. The Times has its sister paper Sunday Times. The Times is the originator of the ubiquitous Times Roman typeface, originally developed by Stanley Morison. Saturday supplements and online presence.

    презентация [3,7 M], добавлен 27.05.2014

  • Сущность понятия имидж политического деятеля, принципы и mass-media каналы его формирования, анализ зарубежного опыта. Имидж председателя Законодательного Собрания Краснодарского края: исследование краевых печатных СМИ, перспективы позиционирования.

    курсовая работа [87,9 K], добавлен 09.06.2013

  • Oprah as one of the longest-running daytime television talk shows in history. General characteristics of the Oprah Winfrey Show, acquaintance with guests. Relationship with the activity of one of the most influential in the world of american leaders.

    презентация [1,1 M], добавлен 28.05.2015

  • Характеристика муниципального унитарного предприятия г. Кунгура "Телерадиокомпания "Кунгур". Программы и рубрики телерадиокомпании. Правила внутреннего распорядка и режим работы. Особенности разработки, анализ рекламной кампании фирмы Nike "Play Russian".

    отчет по практике [19,5 K], добавлен 11.09.2015

  • Russian press for the young reader as opposed to the "adult" started with a magazine. History of child Journalism Beginning of a new era in the Children's journalism. Authors of children's creative destiny "Chizh", "Hedgehog" - to the brightest example.

    реферат [11,6 K], добавлен 28.02.2009

  • The characteristics and structure of constitutional law of Ukraine, factors affecting its formation and development, the current trend. Reform and the direction of change of the legal branch of the state. Principles of functioning of constitutional law.

    реферат [40,5 K], добавлен 13.02.2015

  • Concept of the constitutional justice in the postsoviet Russia. Execution of decisions of the Constitutional Court. Organizational structure of the constitutional justice. Institute of the constitutional justice in political-legal system of Russia.

    реферат [23,9 K], добавлен 10.02.2015

  • Russian mass media as the tool of democracy. The law on mass-media of 1991. Strengthening the rights of mass-media and their restriction. Role of the state in becoming. Latent forms of the state intervention. Monitoring by authority of regional editions.

    контрольная работа [16,4 K], добавлен 17.04.2011

  • The foundations of the constitutional system of the Russian Federation. The civil society as the embodiment of balance of private and public interests. Legal and functional character of the civil society. Institutional structure of constitutional system.

    реферат [19,5 K], добавлен 07.01.2015

  • The Constitutional Court about political parties and religious organizations as institutes of the civil society. The party political component of the constitutional system as the subject of the constitutional control. Creation of regional parties.

    реферат [24,9 K], добавлен 14.02.2015

Работы в архивах красиво оформлены согласно требованиям ВУЗов и содержат рисунки, диаграммы, формулы и т.д.
PPT, PPTX и PDF-файлы представлены только в архивах.
Рекомендуем скачать работу.