We checked possible distinctions of the site, geographic area, and you may ethnicity having fun with t-testing and you may analysis from difference (ANOVA) into LIWC classification percentages. On two websites, half dozen of your several t-screening were significant from the after the categories: first-individual just one [t(3998) = ?5.61, p Supplementary Desk 2 getting setting, practical deviations, and contrasts between cultural organizations). Contrasts found extreme differences when considering Light as well as other ethnic teams for the five of the half dozen extreme ANOVAs. For this reason, i provided ethnicity because good dummy-coded covariate into the analyses (0 = Light, step one = Every other cultural teams).

Of your several ANOVA tests connected with geographic part, just two have been significant (family relations and you will confident feeling). Since the differences just weren’t commercially significant, i don’t envision geographical region from inside the further analyses.

Results

Regularity regarding keyword use goes without saying in the detailed statistics (select Dining table 1) and you will thru keyword-clouds. The word-cloud strategy portrays the essential popular terms and conditions across the entire decide to try along with all the http://kissbridesdate.com/belarus-women/ age groups. The expression-cloud system instantly excludes certain words, as well as stuff (a, and you may, the) and you will prepositions (to help you, which have, on). The remainder stuff words is scaled in size prior to the regularity, undertaking an user-friendly portrait of the most common posts conditions across the new try ( Wordle, 2014).

Shape step one shows the fresh new 20 most typical articles conditions utilized in the complete attempt. As well as get noticed, the quintessential commonly used terms had been love (looking in the 67% out-of users), instance (searching within the 62% out of profiles), lookin (searching in the 55% regarding pages), and you will some body (searching in fifty% out-of pages). Hence, the most common terms and conditions was basically comparable around the a long time.

Profile 2 suggests the next 30 typical articles terminology in the the fresh youngest and you will oldest age groups. By detatching the first 20 well-known articles conditions along the sample, we teach heterogeneity regarding the relationships profiles. In the next 31 terms into the youngest age bracket, high percentage conditions included rating (36% off users regarding youngest age bracket), go (33% from profiles throughout the youngest generation), and you can work (28% out-of users regarding the youngest age bracket). However, the latest oldest age bracket got high percent from conditions eg travel (31% out-of users in the eldest age group), great (24% out of profiles on oldest generation), and relationships (19% out-of profiles throughout the oldest age group).

2nd 30 typical words throughout the youngest and you can earliest ages organizations (once deducting the fresh new 20 most frequent terms and conditions away from Figure step one).

Theory Testing old Variations in Vocabulary from inside the Relationships Pages

To evaluate hypotheses, the newest part of terminology in the relationship character that fit for every single LIWC category supported since built variables in the regressions. I tested ages and you will gender given that independent details also modifying getting website and you can ethnicity.

Hypothesis step 1: More mature years might possibly be for the a top percentage of terms and conditions about following the categories: first-person plural pronouns, family relations, relatives, wellness, and positive emotion.

Conclusions mostly offered Theory step 1 (look for Desk 2). Five of five regressions shown a critical head impression to have ages, in a way that as the period of the fresh profile journalist improved, the fresh part of conditions from the group enhanced from the following the categories: first-people plural, family unit members, fitness, and self-confident feeling. We discover zero significant years perception with the proportion of conditions about nearest and dearest classification.

a Gender: 0 (female) and step 1 (male). b Webpages: The 2 websites were dictomously coded since the 1 and you will 0. c Ethnicity: 0 (White) and you can 1 (Cultural otherwise racial minority).

a good Gender: 0 (female) and you can 1 (male). b Web site: The two other sites have been dictomously coded just like the step one and you can 0. c Ethnicity: 0 (White) and you may step one (Cultural otherwise racial fraction).