AUTHORS' RESPONSE
VDTs and Miscarriage
We appreciate the chance to respond to Dr. Robinson's .;tter [ 1989 We agree that workplace regulations should be based on sound scientific research and believe that our study contributes responsibly to the ongoing debate on visual display terminals (VDTs) and reproductive health.
Dr. Robinson points out some "problems" with our study [Goldhaber et al., 19881. To call them problems is a little misleading. They represent results of our data analysis that were fully identified by us in our article. For example, while the finding that miscarriage risk had different associations with VDT use in different occupational categories provides some argument against a VDT effect, such an effect still cannot be ruled out. Numbers of VDT users in occupational categories other than clerical workers were too small to be conclusive.
With regard to bias, we can never be completely certain that no biases exist. Obvious possibilities, however, should be avoided, and/or their likelihood in our data should be investigated. That is why we took precautions in our study to select representative samples of pregnant women, to maximize survey response rates, to implement quality control measures during data collection, and to control for important covariates in our analyses. Dr. Robinson suggests three possible sources of covariate bias that he feels could have affected our findings.