Google Books Ngrams and Political Science: Two Validity Tests for a Novel Data Source
By Sean Richey, Georgia State University and J. Benjamin Taylor, Kennesaw State University
Google Books Ngrams data are freely available and contain billions of words used in tens of millions of digitized books, which begin in the 1500s for some languages. We explore the benefits and pitfalls of these data by showing examples from comparative and American politics. Specifically, we show how usage of the phrase “political corruption” in Italian, French, German, and Hebrew books strongly correlates with Transparency International’s well-cited Corruption Index for France, Italy, German, and Israel. We also use Ngrams to show that the explosive growth in usage of the phrases “Asian American,” “Latino,” and “Hispanic” correlates with real-world changes in these populations after the Immigration and Nationality Act of 1965. These applications show that Ngram data correlate strongly with similar data from well-respected sources. This suggests that Ngrams has content validity and can be used as a proxy measure for previously difficult-to-research phenomena and questions.