A bifurcating autoregression model for cell lineages with variable generation means.

The bifurcating autoregression model for cell lineage data is extended to allow for observations whose means vary from generation to generation. The maximum likelihood estimates of this extended model are found and used to estimate the generation means and overall variance. The model is also used to test for inherited effects as measured by mother-daughter correlation, and for environmental effects as measured by the sister-sister correlation, conditional on inherited effects. Applications to data sets on EMT6 cells and Escherichia coli are given.