When people see more positive messages online, they tend to post more positive messages themselves; the same pattern appears with negative messages. This relationship is correlation: two variables vary together without proving that one causes the other. section 7.1
Correlation alone cannot show why the pattern exists. People may genuinely feel different, imitate the style they encounter, or be affected by another factor. The study measured sentiment through words, so negations such as “not joyful” could be misclassified. section 7.1
A careful interpretation keeps several limits in view:
- Treat observed exposure and later behavior as a relationship, not proof that exposure caused the behavior. section 7.1
- Check effect size: reducing negative posts lowered negative-word usage by only B=−0.07%, possibly a negligible amount. section 7.1
- Question how variables were measured, especially when sentiment analysis ignores negation. section 7.1





















