The blog post fails to mention the sample size, which according to the downloaded data, is <i>164 respondants</i>. Additionally, only 34 respondents (20%) actually said heart (the most popular word in the dataset), so there's a lot of variation in interpretation.