Contained in this for every single webpages, we gathered 1,one hundred thousand profiles of heterosexual guys and you can step one,000 profiles of heterosexual girls. Profiles seek out users through geographical place, years, and you may intercourse filter systems.
To make sure a geographical dispersion from users, we selected equal quantities of users away from five major towns and cities along with metropolitan, residential district, and you will rural portion: Los angeles, Denver, il, Atlanta, and New york. I randomly picked zip rules out of each one of the four section to find profiles. Inside per zip code, each intercourse, we following randomly chosen profiles certainly one of five age groups: early young adulthood (old 18–31 ages; n = step 1,000; Meters = , SD = step 3.17), late more youthful adulthood/very early midlife (old 30–forty-two age; n = 1,000; Yards = , SD = 5.61), later midlife (old fifty–64 years; letter = step one,000; M = , SD = step 3.99), and you will the elderly (aged more than 65 age; letter = step one,000; Yards = , SD = 4.29). I made use of this type of stratifications in order to guarantee a complete age range of relationship profiles inside the sampling. Just like the older adults group you may utilize to three decades, i managed decades since the a continuing variable as opposed to given that an excellent collection varying in the analyses.
Out-of for every reputation, i removed: gender, ages, ethnicity, therefore the “In the Me personally” otherwise “Inside my Conditions” free response area. In order to guarantee anonymity out-of reputation publishers, i did not receive a lot more group advice (age.g., degree, spiritual choice, income) which will serve as distinguishing pointers. New sampling system is illustrated in Additional Appendix A great .
The latest decide to try varied into the years off 18 to help you 95 ages. Another t-take to found zero difference between imply years for females (Yards = , SD = ) and you will boys (M = , SD = ). The latest breakdown of ethnicity from the try try 70% White/Caucasian, 11% Black/Dark colored, 7% Hispanic/Latino, 2% Western, and you may 10% combined race/other.
Character posts
We used the LIWC application to research the content of your users. This program calculates new frequency and you will proportions of specific types of conditions within this a text document. The fresh LIWC program measures up for each and every word-of a text file that have an inside dictionary of greater than cuatro,five-hundred words assigned to phrase kinds. This research drew on eleven mainly based LIWC categories: first-individual one pronouns, first-individual plural pronouns, family relations, members of the family, really works, conclusion, currency, health, sex, self-confident feelings, and you can bad emotion. Dining table 1 include example terms for the each of the LIWC category (for more information regarding this type of rules, come across LIWC, 2007).
We along with created a sounding terms and conditions for attractiveness not available from inside the situated LIWC categories. I observed methods to own design regarding LIWC classes ( Tausczik & Pennebaker, 2010) because of the creating an extensive list of words out of dictionaries, thesauruses, forms regarding past look, and you can terminology produced by the research people. Next, i chose twenty five terms and conditions most user away from appeal predicated on physical appearance 
into the thesauruses and you will participant solutions (elizabeth.g., precious, good-looking, good-looking, hot). The fresh appeal classification is nearly entirely different from this new intimate class, with just one to overlapping term (sexy). Advice with the appeal classification are also included in Desk step one; into done listing of terms and conditions from the appeal class, look for Additional Dining table step one .
Analytic Means
I earliest tested detailed statistics on proportions of terms some one of different ages found in its relationships pages. I also generated illustrative rates portraying widely known terminology.
I then turned to hypothesis investigations using normal minimum squares regression. The outcomes details within data have been brand new ratio regarding terminology fitting all the 12 kinds on the LIWC analyses. This new LIWC groups were every absolutely skewed considering the count out of no values (we.elizabeth., participant didn’t explore people terms in the category). I went for every single research which have a square-means conversion (used to address non-normality from inside the past degree using the LIWC; DeAndrea, Shaw & Levine, 2010; Hirsh & Peterson, 2009). The development out-of conclusions is comparable once applying the changes. For ease of interpretation, results was showed utilising the untransformed LIWC group analysis. The brand new separate variable was years, managed just like the a continuous varying. We also provided gender.