Regression analysis in translation studies

This paper provides an overview of how to compute simple binary logistic regressions and linear regressions with the open source programming language R on the basis of data from the INTERSECT corpus of English texts and their French and German translations. First, we show how one of the key statistics of logistic regressions is conceptually similar to the chi-square test of frequency tables. Second, we exemplify different applications of logistic regressions – with a binary predictor, with an interval/ratio-scaled predictor, and with a combination of both. Finally, we briefly exemplify a linear regression. In all cases, we discuss significance tests and provide examples for effective visualizations.

• Affiliations: 1: University of California, Santa Barbara; 2: University of North Texas, Denton