사리 나오는 블로그

Conservative Facebookusers are more likely to capitalize words than liberals

Data from five groups of Facebook users (500 users) whodeclared their political views on their user pages (‘very liberal’, ‘liberal’, ‘moderate’, ‘conservative’, and ‘very conservative’) show that the moreconservative users are, the more likely they are to capitalize words. Thiscorrelation is thought to be that liberals are less formal than conservativesin their thinking when it comes to rules and conventions. Conservatives (andliberals) send nonverbal messages to other users by capitalizing (anddecapitalizing) words, as we use nonverbal cues (gestures, facial expressions,etc.) when we talk.

Introduction

When we talk, we use nonverbal messages suchas intonation, gesture, facial expression, and so on.  These messages are important features ofcommunication and we use these all the time, intentionally or unintentionally.When we write, we cannot use these kinds of nonverbal messages to expressthoughts or emotions. Instead, we change styles of words or even wordsthemselves. In formal writings, italicizing, underlining or bolding is oftenused to emphasize certain words. In informal writings, several distinct stylesof writing are used: change of spellings (dafor the, nuthin for nothing, ma for my, etc.), shortening orabbreviations (thnx for thanks, lol for laugh out loud, u for you, ‘cuz for because, etc.), lengthening (borrrrrrrred for bored, ummmmfor um, etc.), capitalizing or decapitalizing (‘I do NOT want it’ instead of ‘I do not want it’, ‘i am’ instead of ‘I am’). Some of themare to mean something that is very difficult for us to convey by words only;for example, to emphasize something (bolding, lengthening, or capitalizing).People who use words such as da(the), ma (my), lol (laugh out loud) are to show others that they share a commonsubculture (hip-hop culture or cyber-culture).

I will call these changes in writing stylesand spellings ‘nonverbal’. Whenpeople write, especially write something informally, they are more likely touse these nonverbal messages in their writings. Internet writing would be agood example for informal writing; postings on blogs, personal pages of socialnetworking sites (MySpace, Facebook, etc.), and discussion forums, all havethese features of nonverbal messages. One of them is capitalization of words.On personal pages of Facebook[1]users – called Info page – peopleleave information about themselves, such as sex, birthday, hometown, politicalviews, interests, favorite movies, etc. On PersonalInformation section, people writetheir interests, favorite activities, favorite music/movies/books, and soforth, for example, “Interests: Painting,Drawing, Hiking”, or “Activities:Sleeping, Vegan baking, Watching TV, Bargain hunting”. In contrast to thefavorites sections (music/TV shows/movies/books) where words and titles areusually written capitalized (e.g. Little Miss Sunshine, Sex and the City), Activities and Interests sections are more likely to be written in lower case or,in other words, decapitalized (e.g. eating, working out, hanging out withfriends): words in titles are usually capitalized, such as ‘Pulp Fiction’ or ‘DreamsFrom My Father’, which makes people use capitalized words, but words thatgo into ‘Activities’ and ‘Interests’ sections – words like sleeping or ski - are not necessarily to be capitalized, even thoughcapitalizing the words is conventional in Facebook (Facebook suggests words aswe are typing in, and these are capitalized).

Dictionaries define conservatism asdisposition to preserve what is established, the tendency to prefer atraditional situation to change, or a political philosophy based on traditionand social stability[2].Core aspects of conservative ideology can be defined as resistance to change[3].If these are true for Facebook users and if it is true that capitalizing isconventional (or traditional) in Facebook, one can imagine that conservativeusers are more likely to use capitalized words than liberal users.

I hypothesize that, for Facebook users whoexplicitly specified their political views, the more conservative they are, themore likely they are to use capitalized letters. Facebook users label theirpolitical views as ‘very liberal’,‘liberal’, ‘moderate’, ‘conservative’, ‘very conservative’, or they do notdeclare it - leave it blank[4].In this study, random Facebook users in Stanford and Silicon Valley network[5]are studied as a control group, and users in the same networks but with theirpolitical views listed are studied as an experimental group (five experimentalgroups, from very liberal to very conservative). I will compare theexperimental groups with the control to test my hypothesis; whether or notthere is a correlation between political views and capitalization. If Facebookusers use capitalization as a way ofcommunication in writings, as nonverbal messages do in speaking, they wouldactively use this method, and it will be shown in the analysis.

Methods

Experiment1: Control group

This experiment is to check basicdemographics of the two Facebook networks and how many Facebook users declaretheir political views and activities/interests on their pages. In thisexperiment, 166 randomly chosen subjects were analyzed. All subjects were inStanford and/or Silicon Valley networks. Information under Info tab of their Facebook pages have been checked. Political Views from Basic Information section and Activities/Interests sections have beenchecked and stored for analysis. Subjects who have both capitalized anddecapitalized words in their Activities/Interests sections are considereddecapitalized cases. Subjects who support Democratic Party are considered‘liberal’, and subjects who support Republican Party are considered‘conservative’. Subjects who support other political parties or declaredthemselves as apathetic, not interested, independent, etc. are considered‘other’.

Experiment2: Experimental groups

This experiment is to see whether people’spolitical views are related to their punctuation habits, especiallycapitalization. Pages of 100 Facebook users have been checked and stored foreach of five political views (‘very liberal’, ‘liberal’, ‘moderate’,‘conservative’, ‘very conservative’) – total of 500 Facebook users – and theirrate of capitalization of words in Activities/Interests sections have beenanalyzed. All subjects are in Stanford and/or Silicon Valley networks. This isdone by Profile Search feature ofFacebook, which allows people to search users by various criteria. 

Results

Experiment1

Among 166subjects, 53 subjects (32%) declared their political views and the rest 113subjects (68%) left it blank.

Figure 1. About 1/3 of the subjectsdeclared their political views.  

 Among those whodeclared their political views, 54.7% were liberal.

According to an article by Inside Facebook[6],the percentage of California users who declared their political views onFacebook is 33.2% and among those, liberals are 52.3% and conservatives are18.8%. The political spectrum of the subjects analyzed in this study is quitesimilar to the one of whole California Facebook users.

Experiment2

The data from 500 subjects (100 from eachpolitical view) were analyzed. As seen in Figure 3, the percentage of users whocapitalize words in their Activities/Interests sections (%Cap) increases aspolitical view gets conservative. The percentage of users who did notcapitalize words (or decapitalize words) (%Decap) decreases as political viewgets conservative. To see this trend more clearly, the differences between thetwo (%Cap - %Decap) are shown below, in Figure 5.

Figure 3. For subjects who listed their politicalviews, 35.1% capitalized words in Activities/Interests sections, 29.5%decapitalized them, and 39.6% left the sections blank.

Figure 4. Percentage of users who capitalized wordsand those who did not capitalize words are shown. There is a trend in bothcases.

As seen in Figure 5, the more conservativethe users are, the more likely they are to capitalize words. People who are very conservative are exceptions in thistrend. But in this graph, people who did not list their activities andinterests (%NotReported) are not included. To see how certain pool of peopleare prone to use capitalized words, I created an index called Capitalization Index.

CapitalizationIndex =

(Num. People who capitalize words –Num. People who decapitalize words) / Total Number of People

The index ranges from -1 to 1. If it isclose to 1 (or -1), most of the subjects that are tested use capitalized (ordecapitalized) words. If it is close to zero, it is either %Cap and %Decap aresimilar, and/or most of the subjects did not list their activities/interests.

Figure 6. CapitalizationIndex: a control group and five groups with different political views.

The Capitalization Index of the control group from Experiment 1 (166subjects) was -0.096. These 166 subjects were selected randomly, and theanalysis from Experiment 1 shows that they are more liberal than conservative;54.7% were liberal.

In addition, the control group subjects (subjects from Experiment 1) areprobably less frequent visitors of Facebook. Althoughdata is not shown in the study, more subjects in the control group have onlybasic information on their pages; name, sex, education, and contactinformation. But users who actively show themselves online, such as subjectsfrom the experimental group (Experiment 2) who actively show their politicalviews, are more likely to be active Facebook users. Thus, I assume that theywould be more likely to capitalize words to make their pages look nicer.

The four experimental groups (‘very liberal’to ‘conservative’) show a very interesting trend: the more they getconservative, the more capitalized words they use. Conservative users are 16percent more likely to capitalize words in their activities/interests sections.In contrast, very liberal users are slightly more likely (2%) to decapitalizewords in the same two sections. It is interesting that the two variables thatseem to be quite unrelated (political view and capitalization index) arecorrelated with each other strongly. But veryconservative users were different. They did not use capitalized words asmuch as conservatives did; they capitalized words as liberals did. And thepercentage of users who decapitalize words is higher than very liberal users (see Figure 4), which requires further analysis.

Discussion

Political views andcapitalization

The data shows that conservative people are more likely to use capitalizedwords than liberals, except the people identified themselves as very conservative. One’s political viewis correlated with one’s habit of using capitalized words. This is probablybecause of the fact that liberals are less formal in their thinking thanconservatives, and generally value traditional beliefs less than conservatives.

The capitalization rules of Facebook are not explicit, but it suggestswords as people type in. For example, if I type ‘swi’ in Activities section, Facebook suggests a list of words: Swimming, Switchfoot,Swingers, etc. (not swimming, switchfoot, swingers). Facebook did not have thisfeature (suggesting words) before – it is a relatively new feature – butcapitalizing of words is a Facebook convention since drop-down menus have allcapitalized words, for example, ‘Sex:Male/Female’, or ‘RelationshipStatus: Single/Married/etc.’ So using decapitalized words is a violation ofthis convention. People often violate rules or convention in informal settingsas seen in writings on an Internet discussion forum. A sample discussion forumis like this (Figure 8). Only ungrammatical posts were excerpted from theforum.

Figure8. A thread from a discussion forum, ‘best advice you ever received’,Thoughts.com. The red color indicates the violation of grammatical rules orconvention.

The users in Figure 8 did not capitalize the first words or the pronoun I,and used multiple dots (….) and ain’t.These violations give us the impression about the writer: e.g. the writer mightbe lazy (since the pronoun I was not capitalized), the writer might belaid-back or easy-going (since s/he didn’t care much about the grammar), and soforth, regardless of the writer’s intention to give the impression.

The rule-breaking (or convention-violating) writing style sends informationabout the writer to readers. It does not change the content of words orsentences, but it could change nuance or impression of words and sentences.

Figure 9.  Activities and Interestssections written in capitalized words.                

Figure10. Activities and Interests sections written in decapitalized words.

The impression I get from the two figures above is that: figure 10, whichis not written in capitalized words, looks free from rules or traditions.Figure 9, which is written in capitalized words, looks like it is written by aperson who likes to get things organized. Other people might feel theimpression different from mine or do not care about the style of writing atall, but some people do care about how to write things on their Facebook pages:they care about the writing styles, especially capitalization in this example.Therefore, one might imagine that liberals are more likely to use the writingstyle of Figure 10, since liberals are less likely to want to conform to rules,and that includes rules of grammar. Words like ‘Running’ or ‘Clarinet’ doesn'trequire capitals, but in general I would say that using capitals reflects aview of the world that is formal and hierarchical. The increasing use of no capitalsin places where capitals used to be required also reflects a less formal, lessdeferential view of the world.

Thecontrol group vs. the experimental groups

But subjects who declared their political viewstend to capitalize more frequently than the control group, who did not listtheir political views (Figure 6). The users from the five experimental groupsare the ones who visit Facebook more frequently than the control group, so theyprobably spent more time polishing their Facebook pages, which may result inusing more capitalized words. It might be that using capitals make the userpages look neater, nicer, and more polished. The control group used lesscapitalized words than the very liberals.This could be the effect of frequency of loyalty to the site. I could notcheck this possibility because I do not have the data of how many visitsFacebook users make, but I assume that the difference between the control andexperimental groups results from the frequent visitor effect, in other words,frequent visitors want their user pages look more refined, which meanscapitalized words. I also assume that the users from the five experimentalgroups visit Facebook as frequently as the others.

Conservatives capitalizewords more often than liberals

The data drawn from 500 Facebook users showthat the hypothesis is true; the more conservative people are, the morecapitalized words they are likely to use. The data is very compelling: 5.2%more liberals capitalize words than very liberals; 4.5% more moderates capitalize words than liberals; 8.7% more conservatives capitalize words than moderates. Surprisingly, 18.4% more conservatives capitalize words than very liberals (Figure 5).

The‘very conservatives’

The veryconservatives do not show the effect of capitalization. Less percentage ofthem capitalized words than the liberals.One of the hidden assumptions I made was that the demographics of the fiveexperimental groups are similar to each other; in other words, thedistributions of age, sex, race, religion, etc. among users are similar to eachother. And maybe this hypothesis does not hold for the very conservative experimental group.

The subjects in this study were drawn fromStanford and Silicon Valley networks, strong Democratic Party orientednetworks. According to the exit poll conducted on the recent presidentialelection day, 75% of Stanford students supported Obama[7].  73% of voters in San Mateo county voted forObama, and 70% of voters in Santa Clara county voted for Obama[8].So statistically, people who can explicitly list themselves as ‘veryconservative’ are rare, and the peer pressure would make the possibility evenrarer. The 166 randomly chosen subjects from Experiment 1 had no very conservative user, which means thatthe percentage of very conservativeFacebook users in Stanford/Silicon Valley networks is at most 0.6%.

Thus, peer pressure would be one of thereasons why very conservatives showless capitalization. They feel the negative peer pressure in this liberal orDemocratic environment, so they need a way to tell people that they are not the“stereotypical” conservatives (e.g. people who drives a pickup with a “supportthe troops” sticker on the bumper, enjoys watching NFL, lives in suburban areain a red state, listens to country music, wearing cowboy jeans, etc.), whichare considered “uncool” among collegestudents and unprofessional in Silicon Valley Hi-Tech companies. So, I thinkthat very conservatives use decapitalization to tell people thatthey are not that kind of stereotypical conservatives that people mightimagine.

Limitationsand further research

One of the limitations of this study is thatthe underlying assumptions might not be true. I assumed that the demographicsof the five experimental groups are similar (The distributions of age, sex,race, and others of the five groups are similar). Because I didn’t and couldn’tcontrol for other factors (e.g. age) when selecting subjects, this might haveaffected the obtained results. But my hypothesis is that there is a correlationbetween political views and capitalization rates, so regardless of age or otherfactors, the two variables should correlate with each other. But anotherpossibility is that there is a mediating variable (one suspect would be age because one’s political view iscorrelated with one’s age), and this variable might have driven the results.So, when possible, age (along withother possible mediating variables) should be controlled to see more meaningfulresults.

This study is done with subjects fromStanford and Silicon Valley networks. The results may not hold for othernetworks or the whole Facebook users. ‘Veryconservatives’ in other networks - especially in red states if myhypothesis is true - may show a larger CapitalizationIndex, because conservatives in red states do not experience negative peerpressure from surrounding liberal Democrats. Therefore, when combined with thisstudy’s results, a study conducted for the whole Facebook users or networkswith more conservative users will show meaningful results.