Google search data for social scientists: A tutorial and best practices

Google search data has been described as the most important dataset on human nature ever assembled, giving nearly instant access to datasets that can provide insights to questions about various topics, including disease, racism, religiosity, well-being, and mental health. These data are customizable—researchers can compare search volume across most of the world or zoom into specific geographic regions; access hourly data within the last week or look at monthly data since 2004. However, they have important limitations. We provide a comprehensive overview and tutorial, covering (a) how Google Trends data are calculated, how reliable they are, and why some results yield low-quality data, (b) how to create custom datasets beyond what Google Trends provides by default (creating long trends of daily data, creating datasets with many cross-sectional comparisons, creating panel data), (c) a list of potential promises and pitfalls of Google Trends data, and (d) recommendations to ensure data quality and sound interpretation.