Many papers on soil gas fluxes use parametric tests, after log transformation, to assess the significance of temporally discrete data sets (see Giammanco et al., 2010). My question is: “why is there a preference for parametric tests when non-parametric tests (e.g. Mann Whitney U for example) may be used on the raw (nontransformed) data?” Is this a matter of preference or is there a strong statistical reasons for favouring parametric tests as opposed to non-parametric tests (which are perhaps better suited to the naturally non-Gaussian distribution of soil gas data)?

More Clinton Rissmann's questions See All
Similar questions and discussions