Query: mother, father, mom, dad
Sample: Longitudinal sample

You are seeing the prevalence of American Twitter users whose profile bio contains each token for each year 2015-2020.

Prevalence

Think of this as the "popularity" of a word within American Twitter users' profile bios. Precisely, this is the value 10,000 * (count of profiles containing word / count of ALL profiles) rounded to the nearest integer. You should interpret prevalence as the number of users per 10,000 unique users who had a profile containing token. It is nice to work with prevalence, because people (even you) are bad at fractions and decimals. Most tokens are NOT used by most users, so all incidences are small compared to the total user count.

American Twitter users

Users are found in randomly sampled public tweets from the Twitter API. The users counted here are those who have tweeted, had their tweet sampled, and whose self-entered 'location' field was a US state, city, area code or other likely indicator of a US geographical location.

Profile Bio

Users are prompted at sign-up to 'Describe themselves in 160 characters or less.' Users can edit their bio at any time. This field is called the 'description' in the Twitter API's User object.

Token

Think of this as a word. Precisely, each token is a character string obtained by splitting up a bio. The tokenization is a simple string split on the regular expression \b, which is a word boundary. Many tokens are words like you would find in the dictionary, but not all. The JJJIT-V1 dataset provides counts for many tokens, so you will see abbreviations, numbers, and other things. Tokens whose prevalence is less than 1 in 10,000 are not included in the dataset.

Cross-sectional sample

All unique users observed as they are active. This is a good sample to use to study the population of active Twitter users at a time. You can answer the question: What are American users like on the Twitter platform?

Longitudinal sample

Only those users who were active and sampled in every year 2015-2020. This is a good sample to use to study how Twitter users change over time. You can answer the question: What words (on average) are Twitter users choosing to add, remove or keep within their bios?