Forecast scores and bias.

Forecast were scored as the loglikelihood of observing the 24 community-level prevalence of ocular chlamydial infection at 36 months, with a higher (less negative) loglikelihood indicating a better forecast. Positive bias indicates that the expectations for the 24 communities were on average higher than the observed prevalence.
Forecast scores and bias.

预测结果以36个月时24个社区的眼部衣原体感染（ocular chlamydial infection）社区水平患病率的对数似然（log-likelihood）作为评分依据，对数似然值越高（负值越小），代表预测效果越好。
正偏差则意味着，针对这24个社区的预测期望值平均高于实际观测到的患病率。
预测评分与偏差情况。