Effects of Mobile Phone Use on Accident Risk Problems of Meta-Analysis When Studies Are Few and Bad

This paper provides a meta-analysis of studies that have evaluated the effects of the use of a mobile phone while a person is driving on the risk of accident. The paper illustrates the difficulties of doing meta-analyses when studies are few and bad. Two main groups of studies were formed: one consisted of six studies that reported the actual use of mobile phones at the time of an accident; the other consisted of seven studies with less-precise information about the use of mobile phones at the time of accident. One study belonged to both groups. Study findings were heterogeneous even within these small groups. The trim-and-fill technique suggested the presence of publication bias, but its effects on summary estimates of risk varied according to whether a fixed-effects or random effects model of meta-analysis was used. Evidence from the epidemiological studies was not consistent with the findings of effects on driver behavior from driving simulator studies.