Using machine learning to identify important predictors of COVID-19 infection prevention behaviors during the early phase of the pandemic

Van Lissa C. J.;Stroebe W.;vanDellen M. R.;Leander N. P.;Agostini M.;Draws T.;Grygoryshyn A.;Gutzgow B.;Kreienkamp J.;Vetter C. S.;Abakoumkin G.;Abdul Khaiyom J. H.;Ahmedi V.;Akkas H.;Almenara C. A.;Atta M.;Bagci S. C.;Basel S.;Kida E. B.;Bernardo A. B. I.;Buttrick N. R.;Chobthamkit P.;Choi H. -S.;Cristea M.;Csaba S.;Damnjanovic K.;Danyliuk I.;Dash A.;Di Santo D.;Douglas K. M.;Enea V.;Faller D. G.;Fitzsimons G. J.;Gheorghiu A.;Gomez A.;Hamaidia A.;Han Q.;Helmy M.;Hudiyana J.;Jeronimus B. F.;Jiang D. -Y.;Jovanovic V.;Kamenov;Kende A.;Keng S. -L.;Thanh Kieu T. T.;Koc Y.;Kovyazina K.;Kozytska I.;Krause J.;Kruglanksi A. W.;Kurapov A.;Kutlaca M.;Lantos N. A.;Lemay E. P.;Jaya Lesmana C. B.;Louis W. R.;Lueders A.;Malik N. I.;Martinez A. P.;McCabe K. O.;Mehulic J.;Milla M. N.;Mohammed I.;Molinario E.;Moyano M.;Muhammad H.;Mula S.;Muluk H.;Myroniuk S.;Najafi R.;Nisa C. F.;Nyul B.;O'Keefe P. A.;Olivas Osuna J. J.;Osin E. N.;Park J.;Pica G.;Pierro A.;Rees J. H.;Reitsema A. M.;Resta E.;Rullo M.;Ryan M. K.;Samekin A.;Santtila P.;Sasin E. M.;Schumpe B. M.;Selim H. A.;Stanton M. V.;Sultana S.;Sutton R. M.;Tseliou E.;Utsugi A.;Anne van Breen J.;Van Veen K.;Vazquez A.;Wollast R.;Wai-Lan Yeung V.;Zand S.;Zezelj I. L.;Zheng B.;Zick A.;Zuniga C.;Belanger J. J.

2022-01-01

---

Abstract

Before vaccines for coronavirus disease 2019 (COVID-19) became available, a set of infection-prevention behaviors constituted the primary means to mitigate the virus spread. Our study aimed to identify important predictors of this set of behaviors. Whereas social and health psychological theories suggest a limited set of predictors, machine-learning analyses can identify correlates from a larger pool of candidate predictors. We used random forests to rank 115 candidate correlates of infection-prevention behavior in 56,072 participants across 28 countries, administered in March to May 2020. The machine-learning model predicted 52% of the variance in infection-prevention behavior in a separate test sample—exceeding the performance of psychological models of health behavior. Results indicated the two most important predictors related to individual-level injunctive norms. Illustrating how data-driven methods can complement theory, some of the most important predictors were not derived from theories of health behavior—and some theoretically derived predictors were relatively unimportant.