Statistical Tests for Publication Bias in Safety Meta-analysis: A Case Study of Urban Traffic Calming Effects

Meta-analysis is a statistical systematic literature review which is common approach to summarize and obtain overall estimation of different studies. It has been increasingly used for evaluating the effects of various afety measures. Publication bias is an important error in meta-analysis, being cause of not publishing studies that have generally small size and non-significant impact. Funnel plot is a visual qualitative tool for detecting publication bias in meta-analysis. This paper introduces and explains two statistical tests, rank correlation method and linear regression method, based on funnel plot for detecting statistically significant publication bias in meta-analysis. These two statistical tests can be used as a quantitative tool for detecting publication bias. This is the first time that a quantitative method has been used for publication bias detections in safety metaanalysis.As a case study, the safety effects of urban traffic calming measures have been analyzed using meta-analysis. Safety effects of urban traffic calming have been gathered from 46 related studies and an overall effect for this safety treatment has been obtained. Two aforementioned tests are used to test publication bias. Meta-analysis shows that urban traffic calming measures approximately reduce the number of injury accidents, property damage accidents and total accidents by about 15, 17 and 14 percent, respectively. Statistical tests show that there is no evidence of publication bias in urban traffic calming studies