I tested potential distinctions of the web site, geographical region, and you will ethnicity having fun with t-examination and you may investigation off difference (ANOVA) for the LIWC group rates. On a couple other sites, six of a dozen t-evaluating was basically extreme regarding the following the categories: first-people just one [t(3998) = ?5.61, p Secondary Desk 2 having form, practical deviations, and you can contrasts ranging from cultural organizations). Contrasts revealed tall differences when considering Light and all other ethnic groups inside five of your half a dozen extreme ANOVAs. For this reason, we included ethnicity while the a great dummy-coded covariate during the analyses (0 = White, 1 = Any cultural communities).

Of the several ANOVA evaluating about geographic region, merely a couple have been tall (nearest and dearest the most beautiful Tianjin girls and you will positive emotion). Because differences weren’t technically important, i did not envision geographical region in the then analyses.

Results

Frequency away from term have fun with goes without saying when you look at the detailed analytics (come across Dining table step one) and you may via word-clouds. The expression-affect technique portrays by far the most popular terms over the whole try plus in each one of the age range. The word-affect system immediately excludes certain terms and conditions, together with articles (an effective, and you can, the) and you will prepositions (to help you, which have, on). The rest posts terms try scaled in proportions relative to their regularity, undertaking an intuitive portrait of the most extremely commonplace blogs terms and conditions round the the fresh try ( Wordle, 2014).

Profile step one reveals new 20 common content terms used in the whole shot. As well as rise above the crowd, the most frequently employed terminology was like (lookin when you look at the 67% off pages), particularly (appearing in the 62% out-of pages), appearing (appearing inside 55% from pages), and someone (lookin for the 50% of pages). Therefore, the most common terms was equivalent across age groups.

Figure dos suggests the second 31 most commonly known posts conditions into the the fresh youngest and eldest a long time. By eliminating the initial 20 prominent articles terminology across the decide to try, we instruct heterogeneity from the dating pages. Next 30 terminology to your youngest age group, significant number terminology integrated rating (36% from profiles from the youngest generation), wade (33% from pages regarding the youngest age bracket), and you will functions (28% out-of pages about youngest age group). Having said that, the eldest age group had large proportions from terms eg travel (31% regarding users in the oldest age bracket), higher (24% from users regarding oldest age group), and you will relationship (19% from pages regarding the earliest generation).

Next 31 most typical words on the youngest and you may earliest ages groups (shortly after subtracting the fresh new 20 most typical words out-of Profile step one).

Theory Evaluation old Differences in Language into the Relationship Users

To check on hypotheses, the newest part of words in the dating reputation that fit for every LIWC class supported once the built variables in regressions. We examined years and you will gender as independent parameters in addition to modifying to have website and you will ethnicity.

Hypothesis step one: Elderly decades might possibly be with the increased percentage of words in the after the categories: first-individual plural pronouns, relatives, family, fitness, and self-confident emotion.

Conclusions largely supported Hypothesis step 1 (select Table 2). Five of your four regressions found a significant main perception for ages, such that given that age the brand new profile journalist improved, the fresh new portion of terms about category enhanced from the following categories: first-individual plural, nearest and dearest, fitness, and you can positive emotion. I receive zero significant ages impact towards ratio out of terminology on members of the family class.

a beneficial Gender: 0 (female) and you can step 1 (male). b Site: The two websites was dictomously coded just like the step 1 and 0. c Ethnicity: 0 (White) and you will step one (Ethnic otherwise racial minority).

an excellent Gender: 0 (female) and you will step one (male). b Webpages: The 2 websites have been dictomously coded while the 1 and 0. c Ethnicity: 0 (White) and you can 1 (Cultural or racial fraction).