Clarity of Research Methods – Trimming and Cooking Data

Sample Paper Excerpt

An investigator is completing research on a study she is hoping to publish. The study involves a relatively small set of observations (140) on the magnetic properties of a certain compound. She is in the ideal position of testing two competing mathematical models predicting different strengths and orientations of the magnetic fields of the compound, 1) one proposed by a senior colleague/co-author, and the other, 2) by one of the senior colleague's research competitors.

As the investigator performed the analysis, she noticed that a group of the observations (49 of them) appeared to be poor measures, not consistent with either theory. The other 91 observations conformed very nicely to the predictions generated by her advisor's model. She analyzed all observations, and included in a footnote that,

"...49 of the observations seemed 'problematic.'"

The senior colleague recommended redoing the analysis section using only the 91 observations that behaved according to expectations, and to mention in a footnote that,

"...49 observations were dropped (trimmed) due to measurement error."

How should the investigator report the analysis?

Analyze and report all the data (140 cases).
Analyze and report the 91 cases verifying the theory.
Report neither.
Analyze and report both analysis, full and trimmed data set.

Proceed to the

Finished?

Please add this site to your Trusted Sites, or temporarily disable any Popup blocker.