I looked at potential differences by the website, geographic part, and you may ethnicity playing with t-examination and you will analysis regarding variance (ANOVA) into the LIWC category rates. On one or two websites, half a dozen of the several t-evaluating was significant on the following the categories: first-people only one [t(3998) = ?5.61, p Second Desk dos to possess mode, important deviations, and you may contrasts ranging from cultural organizations). Contrasts shown significant differences when considering Light and all of other cultural groups in the five of one’s half dozen tall ANOVAs. Hence, i included ethnicity since the a beneficial dummy-coded covariate inside the analyses (0 = White, step one = Other ethnic groups).
Of a dozen ANOVA screening connected with geographic part, only a couple was indeed high (family and you can confident feelings). As distinctions weren’t commercially important, we don’t envision geographical region in the further analyses.
Overall performance
Volume off term use goes without saying from inside the detailed analytics (pick Table step 1) and through word-clouds. The word-cloud method portrays the most commonly used conditions along the whole attempt and also in all the age groups. The term-cloud system automatically excludes particular words, also posts (a, and you will, the) and you will prepositions (in order to, which have, on). The remainder stuff terms was scaled sizes in line with its volume, creating an user-friendly portrait of the most extremely commonplace posts terms and conditions around the the latest test ( Wordle, 2014).
Shape step 1 reveals the newest 20 typical stuff words included in the complete sample. As well as rise above the crowd, the quintessential commonly used terms were like (lookin in 67% out of users), like (lookin during the 62% out-of users), appearing (appearing from inside the 55% of users), and anybody (looking within the fifty% out of profiles). Thus, widely known terminology had been comparable round the age groups.
Figure 2 reveals next 30 most typical posts conditions for the brand new youngest and earliest age range. By removing the original 20 common stuff terms along the try, we train heterogeneity on relationship pages. In the next 29 conditions into youngest age bracket, significant number words free dominican dating sites incorporated get (36% from users regarding youngest generation), go (33% away from profiles regarding youngest age bracket), and you can work (28% out-of pages throughout the youngest age bracket). In contrast, the eldest generation had highest percent from conditions such as for example travel (31% off pages on oldest age group), higher (24% out of users on the eldest generation), and relationship (19% regarding pages in the earliest age group).
Second 31 most frequent terms and conditions in the youngest and you can earliest many years organizations (shortly after subtracting brand new 20 most typical terms out of Figure 1).
Theory Comparison old Variations in Code from inside the Relationship Pages
To test hypotheses, the percentage of conditions in the matchmaking profile that suit for every single LIWC class supported given that situated variables in regressions. We examined years and you will gender since independent variables including changing having site and you can ethnicity.
Hypothesis step one: Earlier years will be in the a high percentage of words regarding the pursuing the categories: first-person plural pronouns, members of the family, family, wellness, and you may confident feelings.
Conclusions largely served Hypothesis step 1 (look for Desk 2). Four of your five regressions shown a life threatening fundamental impact for many years, such that since the age the character publisher increased, the fresh part of terms and conditions from the category enhanced about following the categories: first-individual plural, nearest and dearest, health, and confident feeling. We discover zero high decades impact to the ratio out-of words from the family relations group.
an effective Gender: 0 (female) and you will step one (male). b Site: Both other sites was in fact dictomously coded as step one and 0. c Ethnicity: 0 (White) and you can 1 (Cultural or racial minority).
a Gender: 0 (female) and 1 (male). b Webpages: The two websites was indeed dictomously coded as the step 1 and 0. c Ethnicity: 0 (White) and step 1 (Ethnic or racial fraction).