Home > Publications . Search All . Browse All . Country . Browse PSC Pubs . PSC Report Series

PSC In The News

RSS Feed icon

Geronimus says black-white differences in mortality "help silence black voices in the electorate"

Do universities need more conservative thinkers?

Starr critical of risk assessment scores for sentencing

Highlights

Frey's new report explores how the changing US electorate could shape the next 5 presidential elections, 2016 to 2032

U-M's Data Science Initiative offers expanded consulting services via CSCAR

Elizabeth Bruch promoted to Associate Professor

Susan Murphy elected to the National Academy of Sciences

Next Brown Bag

PSC Brown Bags
will resume fall 2016

Competing Hazards with Shared Unmeasured Risk Factors

Publication Abstract

Hill, Daniel H., William Axinn, and Arland Thornton. 1993. "Competing Hazards with Shared Unmeasured Risk Factors." Sociological Methodology, 23: 245-77.

Most competing hazards models are based on the rather strong assumption that alternative destinations are stochastically independent. Individual- specific unmeasured risk factors that are shared by two or more alternatives are, as a result, ruled out. The present paper develops a generalization of the standard discrete-time competing hazards model that allows for the types of stochastic dependencies resulting from shared unmeasured risk factors. An empirical example is provided using the process by which young women form their first conjugal residential union, with married and unmarried cohabitation representing the competing alternatives. The results suggest considerable and significant similarity of the alternatives in terms of the unmeasurables. It is also shown that, as a result, the independence assumption leads to substantially biased estimates of the net marriage and net cohabitation survival functions. While the model does require a temporal independence assumption, Monte Carlo simulations indicate that the biases introduced by violations of this assumption are confined primarily to the estimates for time-varying covariates. Estimates for other covariates and the cross-destination correlation coefficient, itself, are relatively robust.

Licensed Access Link

Browse | Search : All Pubs | Next