METHOD比较Bayesian Hierarchical / Multilevel Modeling贝叶斯层级模型对多层级、部分汇聚(partial pooling)与参数不确定性进行概率建模,在组间共享信息的同时保留层级差异。mth_bayesian_hierarchical · v2 · 来源已关联
METHOD比较Difference-in-Differences双重差分Estimates treatment effects from changes over time between treated and comparison groups under design-specific identifying assumptions.mth_did · v2 · 来源已关联
METHOD比较E-value Sensitivity AnalysisE值敏感性分析量化一个未测混杂因素需要与处理和结局达到多强的关联,才可能在特定风险比尺度条件下解释掉观察到的关联。mth_evalue · v2 · 来源已关联
METHOD比较Fuzzy-set Qualitative Comparative Analysis模糊集定性比较分析Analyzes necessity/sufficiency and configurations of conditions using calibrated set memberships and truth-table logic.mth_fsqca · v2 · 来源已关联
METHOD比较Group-Time Difference-in-Differences针对多时期与不同首次处理时间的组别时期效应设计,要求核对条件平行趋势与比较组。mth_did_group_time · v1 · 来源已关联
METHOD比较Instrumental Variables工具变量Uses an instrument that shifts treatment/exposure to identify a causal effect under relevance and exclusion-type assumptions.mth_iv · v2 · 来源已关联
METHOD比较Interrupted Time Series Regression中断时间序列Estimates level/trend changes after an intervention time using repeated outcome observations; causal interpretation depends on the counterfactual trend and competing events.mth_its · v2 · 来源已关联
METHOD比较Meta-analysis元分析Combines comparable effect estimates across studies while accounting for sampling uncertainty and, where modeled, heterogeneity.mth_meta · v2 · 来源已关联
METHOD比较Multiple Imputation多重插补通过生成多个合理的缺失值补全数据集,并结合各补全数据集分析结果,将缺失值不确定性传递到统计推断。mth_multiple_imputation · v2 · 来源已关联
METHOD比较Negative Control Design/Analysis负对照设计与分析在观察性研究中使用理论上不应存在目标因果效应的暴露或结局作为负对照,以探测残余混杂、选择或其他系统性偏差。mth_negative_control · v2 · 来源已关联
METHOD比较Network Analysis网络分析Studies relational structure among nodes using graph representations and measures of degree, centrality, communities and paths.mth_network · v2 · 来源已关联
METHOD比较Parametric g-formula / g-computation参数化g公式在明确因果模型和协变量结构下,通过建模条件结局(纵向场景还涉及时间变化过程)并标准化来估计策略下的平均潜在结局。mth_gformula · v2 · 来源已关联
METHOD比较Propensity Score Adjustment倾向得分调整Uses treatment propensity conditional on observed covariates to balance measured confounders via matching, weighting, stratification or adjustment.mth_psm · v2 · 来源已关联
METHOD比较Quantile Regression分位数回归估计条件结果分布的指定分位数与协变量之间的关系,而不是只建模条件均值。mth_quantile_regression · v2 · 来源已关联
METHOD比较Randomized Controlled Trial随机对照试验Uses randomized assignment to support causal comparison between intervention conditions; interpretation still depends on implementation, adherence, missingness and estimand.mth_rct · v2 · 来源已关联
METHOD比较Regression Discontinuity Design断点回归Identifies a local effect at a cutoff when assignment changes discontinuously with a running variable and potential outcomes are suitably smooth absent treatment.mth_rdd · v2 · 来源已关联
METHOD比较Structural Equation Modeling结构方程模型Represents relationships among observed and latent variables through measurement and structural models; causal interpretation requires substantive identification beyond fit.mth_sem · v2 · 来源已关联
METHOD比较Synthetic Control合成控制Constructs a weighted donor combination approximating the treated unit's pre-treatment path and evaluates the post-treatment gap.mth_synth · v2 · 来源已关联
METHOD比较Target Trial Emulation目标试验模拟用观察性数据明确模拟一个假想随机试验的方案要素,以减少因果问题、时间零点、入组与随访定义不清带来的设计偏差。mth_target_trial · v2 · 来源已关联
METHOD比较Targeted Maximum Likelihood Estimation (TMLE)目标最大似然估计针对预先定义的目标参数,将灵活的数据自适应学习与针对目标参数的更新步骤结合,以获得有针对性的估计和统计推断。mth_tmle · v2 · 来源已关联
METHOD比较Topic Modeling with Latent Dirichlet AllocationLDA主题模型Models documents as mixtures of latent topics and topics as distributions over terms; topics require human interpretation.mth_nlp_lda · v2 · 来源已关联