Out of sight but not out of mind: Why failure to account for left truncation biases research on failure rates

We note at least three major issues in entrepreneurship theory that can be clarified by studying the survival chances of new ventures: the extent to which entrepreneurs are so constrained by initial founding conditions that they are unable to learn; the degree to which heterogeneity and innovative capabilities are lost due to the failure of new ventures; and the imprinting effects of new ventures' early days on their subsequent development. However, previous research on these issues has been inconclusive because of problems in research design and data analysis. In this paper, we shed light on new venture failure rates by assessing the validity and generalizability of previous findings. We argue that research using registration data to study new ventures is very likely to generate biased results and that research attempting to track new ventures from a very early stage can still suffer from selection bias due to left truncation. Using a sample of new ventures from the Panel Study of Entrepreneurial Dynamics II, we provide evidence for the extent of such biases. We offer a statistical solution to left truncation that can be easily applied in widely used statistical programs.

► We note issues that can be clarified by studying organizational emergence.
► We shed light on venture failure by assessing the validity of previous findings.
► We argue that selective sampling threatens the validity of empirical findings.
► We provide evidence for such biases and offer a solution to left-truncation.