Peter Hajas

Notes

Excel Forensics and Soft Science Research

I love stories about Microsoft Excel, especially when they involve scientific misconduct. From Data Colada (via Bruce Schneier):

Two different people independently faked data for two different studies in a paper about dishonesty.

They found evidence in the Excel document:

A little known fact about Excel files is that they are literal zip files, bundles of smaller files that Excel combines to produce a single spreadsheet … CalcChain tells Excel in which order to carry out the calculations in the spreadsheet … Analyses of calcChain for the 5 other out-of-sequence observations similarly support the hypothesis that an analyst (manually) moved observations from one condition to the other. Click the links below to see them

I am suspicious of modern published research in “soft” fields like these. A social psychologist and 269 co-authors repeated studies where they were unable to replicate more than half of the results from 98 papers found in psychology journals. The rates of hypothesis acceptance in “softer” fields is unsurprisingly higher than in other disciplines.