search

statnet / ergm.ego

Fit, Simulate and Diagnose Exponential-Family Random Graph Models to Egocentrically Sampled Network Data https://statnet.org
Other
14 stars 4 forks source link

Adjust p-values in gof.ergm.ego() for uncertainty in estimates of GoF statistics. #30

Open krivit opened 5 years ago

krivit commented 5 years ago

This is challenging, since the the statistics are assumed to be approximately normal under egocentric sampling, whereas the simulated statistic distribution is estimated empirically. It is not clear, in particular whether for a simulated (univariate) statistic vector sim.stat, and obs.stat.est and obs.stat.stderr egocentrically inferred statistics, the better "p-value" is

min(mean(pnorm(sim.stat, obs.stat.est, obs.stat.stderr)),
    mean(pnorm(sim.stat, obs.stat.est, obs.stat.stderr, lower.tail=FALSE)))

or

mean(pmin(pnorm(sim.stat, obs.stat.est, obs.stat.stderr),
          pnorm(sim.stat, obs.stat.est, obs.stat.stderr, lower.tail=FALSE)))

or if neither is valid.