Three different statistics for stochastic background searches are currently under investigation: (i) the standard cross-correlation statistic, (ii) the maximum-likelihood statistic, and (iii) a robust version of the cross-correlation statistic for non-gaussian detector noise. Monte Carlo simulations have shown that for simple models the robust statistic performs better than the standard cross-correlation statistic; while the maximum-likelihood statistic performs either better or worse than the standard cross-correlation statistic, depending on how well one can estimate the auto-correlated detector noise in advance. There are a number of important unresolved problems facing stochastic background searches. These include: (i) how to best estimate the auto-correlated detector noise, (ii) how to set a threshold for signal detection, (iii) how to deal with non-gaussian and/or non-stationary detector noise, (iv) how to distinguish cross-correlated environmental noise from a cross-correlated stochastic gravity-wave signal, and (v) whether or not one can claim detection (or only set an upper limit) given a two detector cross-correlation.

pia astone 2001-08-23