Advances in Food Technology and Nutritional Sciences

Open journal

ISSN 2377-8350

The “I”s have it: Sex and Social Status Differences on Twitter

Jennifer C. Beach, Sheila Brownlow*, Melissa D. Greene and N. Clayton Silver

Sheila Brownlow, PhD

Professor and Chair, Department of Psychology, Catawba College
Salisbury, NC 28144, USA. E-mail:


Because people usually focus on content words in conversation and written text, they may miss the significance of function words that more accurately signal people’s motivations, personality, and psychological states.1 For example, Barack Obama’s November 12, 2012 Tweet Four more years, one of the most shared and re-tweeted posts ever, may tell followers that the re-election was confirmed. However, that Tweet does not have the same meaning as I won or We won, which would have told followers something different about the person sending the Tweet while conveying the same content.

The LIWC analyzes text and classifies word use according to different parts and types of speech, using a built-in dictionary of approximately 6000 words.2 The ability of the LIWC program to accurately identify psychological states of language users through their language has been validated in several ways, using many types of writing samples in multiple contexts.3 The LIWC program categorizes different aspects of word use that tap social processes (e.g., relationships), affect (e.g., positive and negative emotion), cognitive mechanisms (e.g., causality, discrepancy), and linguistic dimensions, such as pronouns. It is these linguistic “function” words that actually say more about a speaker’s emotional state, cognitive sophistication, and demographic background than do content words.1

Pennebaker1 argued that pronoun use is an important social marker for age, sex, and social status; its use also provides insight into a speaker’s psychological state. Although “I” is used more than any other word, comprising 3.64% of our speech1 its use does not signal narcissism.4,5,6 However, the use of “I” is common among those who are anxious, insecure, neurotic, or self-conscious in a situation.7,8 In contrast, use of “you” in social media is a marker of people who are open and conscientious.9 First-person pronouns are more common in the language of people of lower status,10 and they also decrease judgments of a speaker’s competence.11

Enhanced vocal use of “I” is more common in women, because of higher levels of interpersonal focus in comparison to men.12,13 This sex difference in the use of first-person singular pronouns is seen across a wide variety of contexts.14 Younger people are more likely to use first-person singular,12,15 but older people use fewer first-person pronouns, more future tense verbs, and their word and sentence length increases as they age.16 Moreover, “I” use is prevalent in the dynamic style of speaking that focuses on using narrative or storytelling to convey points, with fewer complex words that show cognitive mechanisms.17

Research using methods other than the LIWC as its primary analytical tool, nevertheless, supports the LIWC findings. Specifically, the open vocabulary method used by Eichstaedt and colleagues18,19 demonstrated a clear relationship between the content of speech (i.e., words used) in social media and personality characteristics,18 as well as health and well-being.19 Several of these characteristics differ reliably across age and sex, confirming that women and younger people show openness and the use of articles in their speech. Therefore, content words, as well as function words provide clues as to people’s motivations and psychological states15 in addition to their age and sex.

Given that much communication is in a short, written form such as texts, Facebook posts, and Snapchat captions, linguistic patterns contained in various types of planned writing may also be seen in online writing. One such online source is Twitter, a social media site that is open-access (i.e., every Tweet still exists and is easily accessed, regardless of whether you are on Twitter), and which allows only 140-characters per Tweet, forcing users to be succinct in their expression. Reuters20 estimated that 500 million Tweets are sent each day. These communications afford an opportunity to examine self-referencing in a natural, daily setting among people who have a large number of followers. Therefore, our study examined how short language bursts reveal differences in self-reference through personal pronoun use, according to popularity (a measure of social status), age, and sex of actors. Specifically, we predicted that women and actors with a lower social status on Twitter would demonstrate more personal pronoun (“I”) use in their Tweets.



Tweets from a 30-day period (July 27, 2015 to August 26, 2015) were taken from 50 different famous people. Men and women actors (n=25 each) who were ranked on as the “most popular” in 2015 were targeted for our sample if they had an active (i.e., having tweeted in the past 30 days at least one time) Twitter account and used Twitter in English. We chose the most popular 25 women actors and 25 most popular men actors from separate ranking lists. Because famous people sometimes have others pretend to be them, we verified the Twitter account by either checking the blue circle on the account to look for a white checkmark (which means that the Twitter corporation verified to whom it belonged), or we went to the target person’s website and located the Twitter account link. The final sample included 2128 Tweets.

Determining Social Status

We took a median split of number of followers, our operationalization of social status, for each Twitter user (Mdn=1,063,500 followers, SD=3,610,545), and their age (Mdn=45, SD=14.19). The rationale for using a median split (as opposed to dividing the sample into thirds to examine potential quadratic effects) was to avoid potentially discrepant and significantly small within cell sample sizes that could severely diminish power. A t-test confirmed that the median split for number of followers yielded two groups (low, high) differing significantly in the number of followers, t(48)=4.63, p<.0001. However, there were no statistically significant differences in the number of followers as a function of sex, t(48)=1.76, p>.05, or age t(48)=.70, p>.05. The final sample of Tweets classified by user sex, age, and social status is seen in Table 1.


Table 1: Tweets by user sex, age, and social status.

Number of tweets








<45 years

>45 years



Social status via followers





Note: Age and social status divisions were determined by median split of the sample of Twitter users.

Dependent Measure

Each person’s Tweet was copied and pasted into a separate word document and entered into the LIWC software, using its standard dictionaries that have been validated by the comparison of judges analysis of written text.2 We then calculated the percentage of “I” first person personal pronouns in each Tweet.


Preliminary Analysis

We first determined if there were outliers in the percentage of “I” tweets. Using a criterion of z=3.29, p<.001, 34 tweets from 16 different people (18 low social status and 16 high social status; 21 men and 13 women) were observed. If one or two persons dominated the outliers, then they would be removed from the analysis. However, because the outliers were somewhat uniform in distribution across people and believing that people may phrase their tweets differently and quickly, we opted to include all tweets given that only 1.5% were outliers. Next, we found that age via the median split did not influence “I” use, t(2126)=.89, p>.05, therefore we excluded age from subsequent analysis.

Sex Differences

Percentage of “I” language was examined in a 2×2 (sex × social status) between-subjects analysis of variance (ANOVA). The results produced a statistically significant main effect of sex, F(1, 2124)=4.05, p<.045, with an observed power of .52. As seen in Table 2, women used significantly more “I” language than did men.


Table 2: Percentage of “I” use as a function of sex and social status (number of followers).





Above Median

Below Median


2.00 (SD=4.81) [884]

4.19 (SD=6.60)  [141]


1.99 (SD=4.37) [189]

2.96 (SD=5.28) [914]


2.00 (SD=5.49)

3.12 (SD=4.74)

Total 2.30(SD=5.15)   2.79(SD=5.15)

The number in brackets represents the number of tweets for the particular cell.


Social Status Differences

As shown in Table 2, those with lower social status (i.e., fewer followers) used a significantly higher percentage of “I” language compared to their peers with more followers, F(1, 2124)=26.14, p<.0001. The observed power was .99.

Sex × Social Status

A statistically significant sex × social status interaction was obtained, F(1, 2124)=3.90, p<.049, with an observed power of .51. As illustrated in Figure 1, tests of simple effects revealed that men with few followers used “I” significantly more than did men with many followers F(1, 2124)=25.11, p<.0001. A similar pattern was seen among women. That is, women with fewer followers used significantly more self-focused language than did those with many followers F(1, 2124)=4.92, p<.028. More importantly, men with few followers showed a significantly greater percentage of “I” in their Tweets than did women with few followers, F(1, 2124)=7.95, p<.006, but that among those with many followers (i.e., high social status) there were no statistically significant sex differences in “I” language, F(1, 2124)=.01, p>.05.


Figure 1: Percentage of “I” language according to sex and social status

SBRPOJ-1-104 Fig 1


Additional Caveat

The cell means from the foregoing analysis are shown in Table 2. We note that the standard deviations of the “I” percentage are twice as large as the means, suggesting a lack of normality in the distribution, which was confirmed by a Shapiro-Wilk test (W=.565, df=2128, p<.0001). Although this might result in a loss of power, all hypothesis tests were still statistically significant.


Our results showed that those of lower social status in our sample of famous Twitter users used “I” more than did those of higher status, and that women did so more than men, but that social status obviated sex differences in “I” language. Men with lower social status used more self-referencing than did women with lower social status, but among high-status (i.e., heavily-followed) Twitter users, no sex differences were seen. It is noted, however, that there was quite a disparity in the number of tweets within each level of sex and social status group. Although our findings that women and low-status people used “I” more is consistent with the literature1,10,12,14 with regard to pronoun use, the data also suggested that self-referencing among people of high status may reflect a different communication style than the typical linguistic profile. Specifically, only men with lower social ranking used more than the typical amount of “I” language 3.64%,1 and everyone else used less. Whether this outcome was a function of medium (i.e., 140-character bursts), target sample (50 most famous celebrities), or both, is in question. Nonetheless, our data suggest that examination of language patterns might include social status as a variable that may mitigate typical sex- and age-related patterns.

One possible explanation for high status Twitter users writing “I” less often is that its use may be a function of more formal, complete language expressions, and a lack of comfort with the medium, despite the social popularity (via followers) of the Twitter user. Twitter users with a higher social status may have used their following as a platform to Tweet about various causes, thus removing themselves from their Tweets. A second explanation is that higher-status users with many followers may have been more personally (or media) secure.7 Alternately, the use of “I” is more common in a storytelling or dynamic communication style that uses narrative, rather than linguistic devices with complex cognitive and causal mechanisms.17 Therefore, lower social status may have led to users to affect a chattier and informal style designed to garner followers. The downside, however, is that such personable language may have reinforced the Twitter user’s lack of social status and called his or her competence into question.10,11

 Although women used “I” more than did men, our results showed no meaningful differences in “I” use between men and women actors with many followers, a finding that is in contrast to documented literature using multiple samples indicating that women use the first person-pronoun more often than do men.1,3 Although many LIWC studies use populations not limited to college-aged students,8,18 most include a preponderance of persons in the college-age range whereas our sample did not. This suggests that social status, as it manifests itself in a spontaneous medium, may be more essential to expression than sex. In regard to age, differentiations between “older” and “younger” was a function of the sample and, on a more practical level, somewhat artificial, as our Twitter users (ranging from 22 to 84-years old) were generally middle-aged. Thus, it is not surprising that we found no evidence of age-related differences in “I” use.

Although our findings shed light on the use of first-person pronouns in quick language bursts, the sample we used does not reflect the typical sample of Twitter users. We operationalized social status according to the number of Twitter followers, which in our sample ranged from around 25,000 to over two million, whereas the average number of Twitter followers for any given user is 208.21 Moreover, Twitter posts may change in mood over the course of a day,22 becoming less positive (and probably less “I” oriented) as the day goes on. Finally, it is possible that although the accounts belonged to the persons in question, the Tweets themselves may have been written by publicists posting on social media for the famous persons.

A second limitation of our study is that the Twitter users were primarily American, and mostly White. Thus, our findings are generated from a culture where the use of first-person pronouns and self-focus predominate, in contrast to other more collectivist cultures. It is important, therefore, to examine similar sex and status-related language differences in people other cultures and among a sample that is more inclusive of persons of different races and ethnicities.

In sum, this study of “I” use in Twitter posts by popular actors revealed that language use on social media mirrors some aspects of natural language use, but in other aspects it may be different from that which is written or spoken. Moreover, Twitter language may be different from language on social media that affords longer thoughts to be posted, such as Facebook. Further research might include other measures of social status outside of followers to verify, whether other measures of status qualify “I” use patterns that are a function of age and sex. Our results do, however, contribute to the body of research that provides insight as to how public utterances may reveal private selves and motivations.


The authors are grateful to Justin Adkins and Erich Guebert for their assistance with data collection and study suggestions.


Authors did not add median split of age as a two-level factor in the design along with social status and sex because there were no Tweets by older women with many followers and only 25 tweets from older men with few followers.


The authors have no disclosure or competing interests and this research was unfunded.


There is no consent required for research with public documents, and research used public-domain material (Twitter), which does not require Institutional Review Board (IRB) approval.

1. Pennebaker J. The Secret Life of Pronouns: What Our Words Say About Us. New York, NY, USA: Bloomsbury Press/Bloomsbury Publishing; 2011.

2. Pennebaker JW, Chung CK, Ireland M, Gonzales A, Booth RJ. The development and psychometric properties of LIWC2007. LIWC. Web site. Accessed July 20, 2016

3. Tausczik Y, Pennebaker J. The psychological meaning of words: LIWC and computerized text analysis methods.J Lang Soc Psychol. 2010; 29(1): 24-54. doi: 10.1177/0261927X09351676

4. Carey A, Brucks M, Mehl M, et al. Narcissism and the use of personal pronouns revisited. J Pers Soc Psychol. 2015; 109(3): e1-e15. doi: 10.1037/pspp0000029

5. Golbeck J. Negativity and anti-social attention seeking among narcissists on Twitter: A linguistic analysis. First Monday. 2016; 21(3): 7. doi: 10.5210/fm.v0i0.6017

6. DeWall C, Buffardi L, Bonser I, Campbell W. Narcissism and implicit attention seeking: Evidence from linguistic analyses of social networking and online presentation. Pers Individ Dif. 2011; 51(1): 57-62. Web site. Accessed July 20, 2016

7. Hirsh J, Peterson J. Personality and language use in self-narratives. J Res Pers. 2009; 43(3): 524-527. Web site. Accessed July 20, 2016

8. Pennebaker J, King L. Linguistic styles: Language use as an individual difference. J Pers Soc Psychol. 1999; 77(6): 1296-1312. doi: 10.1037/0022-3514.77.6.1296

9. Golbeck J, Robles J, Edmondson M, Turner K. Predicting personality from Twitter. 2011 IEEE international conference on privacy, security, risk, and trust, and IEEE International conference on social computing. 2011; 149-156. Web site. Accessed July 20, 2016

10. Kacewicz E, Pennebaker J, Davis M, Jeon M, Graesser A. Pronoun use reflects standings in social hierarchies. J Lang Soc Psychol. 2014; 33(2): 125-143. doi: 10.1177/0261927X13502654

11. Berry D, Pennebaker J, Mueller J, Hiller W. Linguistic bases of social perception. Pers Soc Psychol Bull. 1997; 23(5): 526-537. doi: 10.1177/0146167297235008

12. Pennebaker J. How author’s forgettable words reveal their personality and social behavior. Information Design Journal. 2013; 20(3): 267-272. Web site. Accessed July 20, 2016

13. Poole M. Social class, sex, and linguistic coding. Lang Speech. 1979; 22(1): 49-67. doi: 10.1177/002383097902200104

14. Newman ML, Groom CJ, Handelman LD, Pennebaker JW. Gender differences in language use: An analysis of 14,000 text samples. Discourse Processes. 2008; 45(3): 211-236. doi: 10.1080/01638530802073712

15. Kern M, EichstaedtJ, Seligman M, et al. From ‘Sooo excited!!!’ to ‘So proud’: Using language to study development. Dev Psychol. 2014; 50(1): 178-188. doi: 10.1037/a0035048

16. Pennebaker J, Stone L. Words of wisdom: Language use over the life span.J Pers Soc Psychol. 2003; 85(2): 291-301.doi: 10.1037/0022-3514.85.2.291

17. Pennebaker J, Chung C, Frazee J, Lavergne G, Beaver D. When small words foretell academic success: The case of college admissions essays.PLoS ONE. 2014; 9(12). doi: 10.1371/journal.pone.0115844

18. Schwartz HA, Eichstaedt JC, Kern ML, et al. Personality, gender, and age in the language of social media: The open vocabulary approach. PLoS ONE. 2013; 8(9). doi: 10.1371/journal.pone.0073791

19. Park G, Schwartz H, Seligman M, et al. Automatic personality assessment through social media language. J Pers Soc Psychol. 2015; 108(6): 934-952. doi: 10.1037/pspp0000020

20. Reuters. Twitter incorporated company profile. 2016. Web site.
symbol=TWTR.N. Accessed July 20, 2016

21. Muruganandam C. Average number of Twitter followers. 2016. Web site. Accessed July 20, 2016

22. Golder S, Macy M. Diurnal and season mood vary with work, sleep, and daylength across diverse cultures. Science. 2011; 333(6051): 1878-1881. doi: 10.1126/science.1202775


Practical Pointers for Drug Development and Medical Affairs

Gerald L. Klein*, Roger E. Morgan, Shabnam Vaezzadeh, Burak Pakkal and Pavle Vukojevic



Prevalence and Risk Factors of Subclinical Mastitis of Goats in Banadir Region, Somalia

Omar M. Salah*, Yasin H. Sh-Hassan, Moktar O. S. Mohamed, Mohamed A. Yusuf and Abas S. A. Jimale


Use of Black Soldier Fly (Hermetia illucens) Prepupae Reared on Organic Waste

Maggot Debridement Therapy: A Natural Solution for Wound Healing

Isayas A. Kebede*, Haben F. Gebremeskel and Gelan D. Dahesa,


Figure 11. Risk Map for the Introduction of Ruminant Diseases at Borders

Ovine Network in Morocco: Epizootics Spread Prevention and Identification of the At-Risk Areas for “Peste des Petits Ruminants” and “Foot and Mouth Disease”

Yassir Lezaar*, Mehdi Boumalik, Youssef Lhor, Moha El-Ayachi, Abelilah Araba and Mohammed Bouslikhane



The Impact of Family Dynamics on Palliative Care at the End-of-Life

Neil A. Nijhawan*, Rasha Mustafa and Aqeela Sheikh


Long-Term Follow-Up After Laparoscopic Radical Prostatectomy for Localized and Locally Advanced Prostate Cancer

Shrenik J. Shah*, Abhishek Jha, Chirag Davara, Rushi Mistry and Kapil Kachhadiya



Treatment and Control Methods of Bovine Mastitis: A Review

Isayas A. Kebede* and Gelan D. Dahesa



Prevalence (%) of Types of Anthropometric Failure among Previous and Present Studied Tribal Children

Original Research, peer reviewed

2024 Apr

Biswajit Mahapatra and Kaushik Bose*


2024 Apr

Gerald L. Klein*, Roger E. Morgan, Shabnam Vaezzadeh, Burak Pakkal and Pavle Vukojevic

Original Research

2024 Mar

Omar M. Salah*, Yasin H. Sh-Hassan, Moktar O. S. Mohamed, Mohamed A. Yusuf and Abas S. A. Jimale