We looked at prospective distinctions by the webpages, geographic region, and you will ethnicity using t-evaluation and you may research out-of variance (ANOVA) with the LIWC classification percentages. To your two websites, half a dozen of several t-tests were extreme regarding the after the categories: first-person just one [t(3998) = ?5.61, p Supplementary Dining table 2 to have form, simple deviations, and you can contrasts ranging from cultural communities). Contrasts shown significant differences between Light and all sorts of almost every other cultural communities within the four of half dozen extreme ANOVAs. Therefore, we integrated ethnicity because the a beneficial dummy-coded covariate from inside the analyses (0 = Light, 1 = Some other cultural organizations).
Of one’s a dozen ANOVA testing associated with geographical region, only a few was basically extreme (family relations and you may positive feeling). As differences were not officially meaningful, i failed to consider geographic part inside the further analyses.
Efficiency
Frequency out of keyword play with is obvious into the detailed statistics (come across Dining table 1) and you can thru keyword-clouds. The definition of-cloud techniques illustrates probably the most popular conditions across the entire test along with each of the age range. The expression-affect system instantly excludes certain terms, also blogs (a great, and, the) and prepositions (so you’re able to, having, on). The remaining stuff words is actually scaled in proportions prior to its regularity, starting an intuitive portrait of the very commonplace posts words across brand new test ( Wordle, 2014).
Profile step one reveals the latest 20 most common stuff terms used in the complete take to.