On top of standard errors in data samples, researchers may add “sizeable” further “non-standard errors” through the way they handle data, research published by the Bank of England finds.

The “fincap” project brought together 164 teams to analyse millions of data points provided by Deutsche Börse over a two-month period.

The paper’s authors, Albert Menkveld et al, find “sizeable” dispersion in the results of hypothesis testing by different teams. This, the authors say, indicates that not only