The topic was about sentiment, not usage. And the Pew Article has broad sample sizes ranging from 5,000 to 28,000.