We checked-out prospective distinctions because of the website, geographic area, and ethnicity playing with t-tests and you may data out of variance (ANOVA) towards LIWC class proportions. Toward several other sites, half a dozen of the a dozen t-examination was in fact extreme about following the kinds: first-individual one [t(3998) = ?5.61, p Additional Desk dos having means, basic deviations, and you can contrasts ranging from ethnic communities). Contrasts revealed significant differences when considering White and all of other ethnic communities during the five of the six tall ANOVAs. Thus, i provided ethnicity given that a dummy-coded covariate into the analyses (0 = White, 1 = Some other cultural groups).
Of your own 12 ANOVA testing regarding geographical part, only several was in fact high (household members and you can positive feelings). Since the differences were not technically significant, i don’t imagine geographical part inside the subsequent analyses.
Overall performance
Regularity out of keyword use goes without saying into the descriptive statistics (find Desk step 1) and you can via phrase-clouds. The definition of-affect strategy portrays probably the most commonly used terms and conditions along the whole decide to try and also in all the age groups. The word-cloud program instantly excludes specific conditions, as well as articles (an effective, and, the) and you can prepositions (to help you, which have, on). The remainder articles words was scaled in dimensions in line with its volume, performing an user-friendly portrait of the very most prevalent posts terms round the the fresh shot ( Wordle, 2014).
Shape step 1 shows the fresh 20 most common articles conditions included in the complete decide to try. As can get noticed, the quintessential frequently employed words were like (lookin during the 67% out-of users), like (lookin during the 62% of profiles), appearing (looking in 55% out of profiles), and you will some body (looking into the fifty% regarding profiles). Hence, the most popular terms and conditions was indeed similar around the age range.
Contour dos reveals the following 29 most frequent articles terms during the the new youngest and you may oldest age range. By eliminating the first 20 preferred articles words along side shot, we teach heterogeneity on dating profiles. Within the next 29 terminology to Laostian naiset vs. amerikkalaiset naiset your youngest age bracket, raised percentage terms integrated score (36% of pages regarding youngest generation), wade (33% regarding pages from the youngest age group), and you can works (28% regarding users in the youngest age bracket). However, the latest eldest age group had large proportions regarding terms and conditions such as travel (31% off users throughout the earliest age group), higher (24% away from profiles from the oldest age bracket), and you will dating (19% from users from the earliest age group).
Next 29 popular conditions on youngest and you can earliest age teams (just after deducting new 20 most commonly known terminology out-of Contour step 1).
Hypothesis Assessment of age Variations in Words during the Relationships Pages
To evaluate hypotheses, the brand new portion of conditions from the matchmaking character that fit for every LIWC category offered since the dependent details into the regressions. We tested age and gender due to the fact separate parameters plus modifying to possess site and you can ethnicity.
Theory step one: Earlier many years would-be of this a top percentage of terms and conditions regarding the after the groups: first-individual plural pronouns, family members, relatives, fitness, and you may confident feeling.
Conclusions mostly supported Hypothesis step 1 (get a hold of Desk dos). Five of the four regressions revealed a serious chief impact to own age, in a manner that as age the brand new character writer enhanced, the fresh new portion of conditions regarding the class improved regarding adopting the categories: first-person plural, nearest and dearest, wellness, and you may confident emotion. We discovered no high decades feeling into the proportion out of terms throughout the friends classification.
an excellent Gender: 0 (female) and step one (male). b Web site: The two other sites was in fact dictomously coded given that 1 and you can 0. c Ethnicity: 0 (White) and you can step one (Ethnic or racial fraction).
an excellent Gender: 0 (female) and you will 1 (male). b Site: Both websites was in fact dictomously coded once the step one and 0. c Ethnicity: 0 (White) and you can 1 (Ethnic otherwise racial minority).
