And the same question asked about hitting
Most parents who shout do not hit, and this is here anyway, because the two sit on the same road and because these are the two largest bodies of evidence on the subject. Both sets of authors state their own limits, and those are printed here beside the numbers rather than left out.
What the research says
Gershoff ET, Grogan-Kaylor A (2016). Spanking and child outcomes: old controversies and new meta-analyses. Journal of Family Psychology, 30(4), 453 to 469 doi
75 studies covering nearly 161,000 children. The overall association with worse outcomes was small to moderate, d 0.33 with a confidence interval of 0.29 to 0.38. Two entries in their table are worth more than the headline. The association with immediate defiance was not statistically significant, so it does not reliably even work in the moment. And the largest single association, d 0.64, was with having been physically abused, which is the escalation risk stated as a number. The authors are explicit that none of this establishes cause: children who are harder to manage draw more punishment, many designs cannot establish which came first, randomised experiments here are not ethical and never will be, and only one of their twenty adult outcome estimates came from a prospective study.
Heilmann A, Mehay A, Watt RG, Kelly Y, Durrant JE, van Turnhout J, Gershoff ET (2021). Physical punishment and child outcomes: a narrative review of prospective studies. The Lancet, 398(10297), 355 to 364 doi
69 studies that followed children forwards in time. Physical punishment consistently predicted increases in behaviour problems later, and the authors state that not one of the studies reported a beneficial effect on any child outcome. The more often it happened, the stronger the association. Their own stated limits: this is a narrative review rather than a meta analysis, so it does not weight studies by size, and 60 of the 68 articles came from the United States, which matters a great deal for anyone reading this outside it.