Table 2.

Variables for modeling age of Reddit users in the training set derived from the comment, post, and user data collected for users whose ages were confirmed by manual labeling.

Variable group	Metadata used	Example	Variables (N=1523), n
Summary statistics	All	Median post score	189
Subreddit frequencies	Posts and comments	Frequency user posted to “Teenagers” subreddit	624
Emoji frequencies	Comments	Frequency of “” out of emojis used by user	101
Literary characteristics	Posts and comments	Average Flesch Reading Ease score	28
Patterns in Posting	Posts and comments	Percentage of user’s posts that were videos	42
Term usage	Comments	TF-IDFa score for the term “school”	539b

^aTF-IDF: term frequency–inverse document frequency.

^bThe number of TF-IDF terms varied across the cross-validation folds based on the comments and submissions vocabulary present in the training portion of each fold. The value presented here is the number of TF-IDF features when calculated on text from the full training set.