We checked-out prospective distinctions of the web site, geographical part, and you will ethnicity having fun with t-tests and you will investigation of difference (ANOVA) on the LIWC class percentages. With the two other sites, half a dozen of the several t-evaluating had been extreme on pursuing the kinds: first-person just one [t(3998) = ?5.61, p Additional Desk dos to own mode, important deviations, and you will contrasts ranging from cultural organizations). Contrasts revealed extreme differences when considering White and all of almost every other ethnic groups in five of one’s six high ANOVAs. For this reason, i included ethnicity because the an effective dummy-coded covariate during the analyses (0 = Light, 1 = Almost every other ethnic communities).
Of your own several ANOVA examination associated with geographic part, only a couple of have been high (relatives and self-confident feelings). Due to the fact differences were not theoretically meaningful, we didn’t believe geographic area within the further analyses.
Abilities
Regularity of keyword fool around with is obvious inside detailed analytics (get a hold of Dining table 1) and via word-clouds. The phrase-cloud techniques depicts one particular commonly used conditions along the whole test and in each of the age groups. The phrase-cloud system instantly excludes specific words, and additionally content (a great, and you Bakersfield, CA beautiful women may, the) and you may prepositions (to help you, which have, on). The rest articles terms and conditions is actually scaled sizes according to its volume, carrying out an intuitive portrait of the most extremely common content terms and conditions across the the newest try ( Wordle, 2014).
Figure step one reveals the 20 most typical posts words used in the complete shot. As can rise above the crowd, many commonly used words was in fact love (looking within the 67% away from profiles), particularly (searching within the 62% regarding profiles), searching (lookin in the 55% off users), and you may anyone (looking in the fifty% regarding users). For this reason, typically the most popular conditions had been equivalent around the a long time.
Shape 2 shows the following 31 most frequent blogs terms when you look at the the youngest and oldest age groups. By eliminating the first 20 preferred content terms along the attempt, we train heterogeneity regarding the relationships profiles. Next 29 words towards the youngest generation, high percentage conditions included score (36% regarding users about youngest age group), wade (33% from pages regarding the youngest age group), and you can works (28% out of profiles on youngest generation). Alternatively, brand new eldest generation got higher rates of terms and conditions including take a trip (31% off profiles in the oldest generation), higher (24% out-of profiles in the earliest generation), and you can matchmaking (19% out-of users in the oldest age bracket).
2nd 31 popular terms and conditions regarding youngest and you may eldest many years communities (after subtracting the newest 20 most commonly known terms out of Figure step one).
Theory Evaluation of age Differences in Words within the Matchmaking Users
To check on hypotheses, the latest percentage of terms throughout the relationships reputation that suit for every LIWC class offered once the oriented details in regressions. I examined decades and gender since the separate details and additionally adjusting for web site and ethnicity.
Theory 1: More mature age might be on the a higher portion of terms and conditions about after the categories: first-individual plural pronouns, family relations, household members, wellness, and you may self-confident feeling.
Conclusions mainly offered Theory step 1 (find Table 2). Four of the four regressions found a life threatening main impact to have decades, in a manner that since period of the newest reputation creator increased, the latest portion of conditions in the class improved on the following the categories: first-individual plural, friends, wellness, and you can positive emotion. We located zero high age impact to your ratio of conditions throughout the nearest and dearest category.
a great Gender: 0 (female) and you can step 1 (male). b Web site: Both other sites was indeed dictomously coded as the 1 and you will 0. c Ethnicity: 0 (White) and you can step 1 (Ethnic or racial fraction).
a great Gender: 0 (female) and you will 1 (male). b Website: The 2 other sites was basically dictomously coded as step 1 and you can 0. c Ethnicity: 0 (White) and you will 1 (Ethnic or racial fraction).