I looked at prospective differences from the site, geographic area, and you will ethnicity playing with t-evaluating and research from variance (ANOVA) for the LIWC group rates. To the two websites, half a dozen of a dozen t-evaluation had been extreme about pursuing the groups: first-people one [t(3998) = ?5.61, p Second Dining table 2 getting form, fundamental deviations, and you can contrasts between ethnic communities). Contrasts found tall differences between White and all of other ethnic groups when you look at the four of the half a dozen high ANOVAs. Hence, we integrated ethnicity while the a great dummy-coded covariate inside the analyses (0 = White, step 1 = Any kind of ethnic organizations).
Of your a dozen ANOVA screening linked to geographic area, simply one or two were extreme (family and you will self-confident feelings). Because variations just weren’t theoretically significant, i don’t consider geographical area within the further analyses.
Show
Volume out of keyword use is evident inside detailed statistics (pick Table 1) and you may via keyword-clouds. The expression-affect method depicts one particular widely used words along side entire test plus each one of the a long time. The definition of-cloud system instantly excludes particular conditions, and additionally blogs (good, and you will, the) and prepositions (to, having, on). The remaining stuff terms and conditions try scaled in size in line with its regularity, starting an user-friendly portrait of the most extremely common content terms round the brand new test ( Wordle, 2014).
Figure step one shows the latest 20 most common content terms found in the whole test. As can get noticed, by far the most commonly used words was in fact like (searching during the 67% out-of profiles), for example (searching inside the 62% away from users), appearing (lookin when you look at the 55% regarding users), and you can some one (looking when you look at the 50% regarding profiles). Therefore, the most used terms and conditions was indeed equivalent across a long time.
Profile 2 suggests the following 31 popular content terms when you look at the the new youngest and you will eldest age range. By removing the initial 20 prominent posts conditions over the take to, i show heterogeneity on the relationships pages. In the next 29 terms for the youngest age bracket, significant number conditions integrated rating (36% out of pages regarding the youngest generation), wade (33% out of pages on the youngest generation), and you can performs (28% of users on the youngest generation). Alternatively, this new earliest age bracket got highest proportions off terminology particularly travelling (31% of users about eldest age group), great (24% from pages in amolatina dating the earliest age group), and you will relationship (19% of profiles about earliest generation).
Next 30 most frequent words from the youngest and you may earliest years groups (immediately after subtracting the brand new 20 most typical terminology out of Shape step one).
Hypothesis Research old Variations in Code inside Matchmaking Pages
To check on hypotheses, the new portion of terms and conditions throughout the relationship profile that fit each LIWC group offered since founded parameters inside regressions. I tested decades and sex once the independent details along with adjusting getting webpages and ethnicity.
Theory step 1: Older ages could be associated with the increased part of terms in the following classes: first-people plural pronouns, household members, loved ones, fitness, and you may self-confident feeling.
Findings mainly offered Hypothesis step one (see Desk 2). Five of one’s four regressions found a life threatening head perception getting ages, such that as ages of the reputation journalist improved, the fresh portion of words regarding class enhanced about pursuing the categories: first-individual plural, relatives, wellness, and you can confident emotion. We discover no significant ages effect on the proportion off terms throughout the nearest and dearest classification.
an intercourse: 0 (female) and you can 1 (male). b Website: The 2 other sites were dictomously coded as step 1 and you can 0. c Ethnicity: 0 (White) and you can 1 (Ethnic or racial minority).
a sex: 0 (female) and you will step 1 (male). b Webpages: The 2 other sites were dictomously coded since the 1 and you may 0. c Ethnicity: 0 (White) and you may step 1 (Ethnic otherwise racial minority).