Is it feasible to teach query programming in three different languages in a single session? A study on a pattern-oriented tutorial and cheat sheets

Abstract

Undergraduates and postgraduates in science subjects are increasingly expected to conduct their data analyses using R, SQL and Python. This requires of instructors to develop resources that get students up and running quickly. This study presents and evaluates a learning design that (1) uses a pattern-oriented tutorial to teach language-independent key operations for implementing data analytic queries, and (2) uses cheat sheets to show how these operations map onto language-specific syntax. The evaluation study (N=21) concludes that using this approach, two thirds of the data science novices sampled could implement simple to moderately complex queries in all the aforementioned languages within two hours. A permutation test moreover produced a significant main effect of language, with SQL ranking the highest in accuracy. The results form part of a general discussion on the merits and language-dependent feasibility of pattern-oriented aids for accelerated data science instruction.