I’ve just finished reading a great book:
Hubbard, R. (2015). Corrupt research. Sage.
I’ve just given it a five-star review on Amazon.
In brief, Hubbard is highly–as in extremely highly–critical of the conventional ‘significant difference’ paradigm, centred on finding p < .05. He advocates what he calls the ‘significant sameness’ strategy, which requires sustained problem solving and building of cumulative understanding over a series of close and not-so-close replications. Yes, it’s definitely a much better approach, and a big part of what Open Science requires.
I don’t like the term ‘significant sameness’, because ‘significant’ is such a deeply ambiguous and misleading word. But the strategy is spot on. It focusses on estimation, replication, and meta-analysis, so it fits perfectly with the new statistics. Hubbard’s book provides a scholarly, wide-ranging, and philosophically sound framework for current best research practice, with Open Science practices and the new statistics both to the fore.
It’s definitely worth a squiz. (If you are not Australian or Kiwi, you can guess what that means, or consult Dr Google. A handy little word…)