The extent of sentiment on sexual health topics between moderated and non-moderated websites

Abstract

Sexual health information shared online is not always credible. Due to the nature of the internet, and how it allows anyone to create or spread content, often misinformation occurs, and being misinformed about (sexual) health can be dangerous. This sparks a research interest to create a model that can predict credibility, in this case, sexual health information. To create a model, we must identify which factors mediate and modulate credibility. In this study, the aim is to evaluate ’sentiment’ as a marker of credibility prediction. Using a moderated and a non-moderated source, we can compare if there is a significant statistical difference between the sentiment on sexual health information between the two sources. A statistical difference would indicate that sentiment is a promising candidate for credibility predictions, and the sentiment can tell sources apart in terms of credibility. Using a rule-based method (Pattern.nl) to compute sentiment and statistical analysis methods, it was concluded that there was no statistically sig- nificant difference between the credible and non-credible sources in terms of sentiment. However, some intruiging patterns surfaced such as the non-credible source scoring higher on high levels of sentiment, or a subtopic within sexual health information that did return statistically significant with a small effect size. Therefore, more research should be conducted to further analyze this marker.