According to Lindley's paradox, most point null hypotheses will be rejected when the sample size is too large. in this paper, a two-stage block testing procedure is proposed for massive data regression analysis. New variables selection criteria incorporat
