Customs vary by field and also by how weak small the sample is and how unstable the findings are and whether you want to interpret that marginal result or ignore it. FWIW flagging p<.1 is pretty uncommon these days in the journals I read; Stata default is what I'm used to seeing across the board.
