EnglishWe checked-out potential differences because of the webpages, geographic area, and ethnicity having fun with t-screening and you may research of variance (ANOVA) to your LIWC group proportions. Into a couple other sites, half a dozen of a dozen t-testing were extreme regarding after the groups: first-individual one [t(3998) = ?5.61, p Second Desk 2 getting mode, fundamental deviations, and you can contrasts ranging from cultural groups). Contrasts shown significant differences when considering Light and all of other ethnic teams inside four of six significant ANOVAs. Therefore, i included ethnicity while the a good dummy-coded covariate when you look at the analyses (0 = Light, step one = Almost every other cultural groups).
Of 12 ANOVA assessment associated with geographic part, only one or two was high (family and you may self-confident feeling). Due to the fact variations were not commercially important, i did not believe geographic part into the further analyses.
Performance
Frequency out-of word fool around with goes without saying during the descriptive statistics (discover Dining table 1) and you will thru keyword-clouds. The term-cloud technique depicts many popular terminology over the entire try and also in each one of the a long time. The phrase-affect system automatically excludes particular words, in addition to blogs (an effective, and you can, the) and you may prepositions (in order to, that have, on). The remainder blogs conditions are scaled in proportions relative to its frequency, performing an intuitive portrait sexy Plano, IA girl quite common articles conditions across the this new sample ( Wordle, 2014).
Shape 1 reveals this new 20 typical posts terms and conditions found in the whole take to. As well as get noticed, more frequently employed conditions was love (looking in the 67% out of pages), like (searching for the 62% of users), appearing (searching in 55% from pages), and you can some body (searching inside the 50% off users). For this reason, the best conditions had been comparable round the age groups.
Shape 2 suggests another 30 popular content words for the this new youngest and you may earliest age groups. By removing the original 20 well-known content terms along the attempt, we show heterogeneity on the relationships users. Next 31 conditions towards youngest age group, high level percentage words integrated score (36% of users regarding the youngest generation), go (33% away from profiles throughout the youngest age group), and you may performs (28% of profiles from the youngest age group). However, the new oldest generation had highest proportions regarding conditions such as for example traveling (31% regarding profiles on oldest generation), high (24% off profiles regarding earliest generation), and you may relationship (19% out-of pages throughout the eldest age group).
Second 30 common terminology on the youngest and you will earliest ages organizations (just after deducting this new 20 most commonly known terms and conditions out-of Shape step one).
Hypothesis Analysis of age Variations in Vocabulary into the Matchmaking Pages
To evaluate hypotheses, the newest portion of conditions on the relationship reputation that fit for each and every LIWC class supported just like the founded details during the regressions. We checked out age and you will gender as the separate variables also changing for web site and you may ethnicity.
Theory step one: Elderly decades might possibly be regarding the a high portion of terms on the following the groups: first-person plural pronouns, members of the family, family unit members, wellness, and you will confident feeling.
Results largely offered Theory step one (pick Desk 2). Four of one’s five regressions revealed a significant head impact to possess decades, such that once the period of the newest reputation publisher increased, the latest portion of conditions in the group improved in the following categories: first-people plural, nearest and dearest, health, and you will confident feelings. We located zero tall many years perception into proportion regarding terminology from the nearest and dearest category.
good Gender: 0 (female) and you may step one (male). b Site: The two other sites had been dictomously coded as the step one and you can 0. c Ethnicity: 0 (White) and you will step 1 (Cultural or racial fraction).
an effective Gender: 0 (female) and you will step one (male). b Web site: The two websites was indeed dictomously coded once the 1 and you may 0. c Ethnicity: 0 (White) and 1 (Cultural otherwise racial minority).