NSI Research: Within-Person and Group Comparisons

Many rows do not necessarily mean many independent people

A journal contains repeated entries from several participants. It is tempting to combine every row into a single chart and ask whether higher reported pressure accompanies more rule violations. The chart might be informative, but the structure of the observations matters. Bland and Altman’s BMJ statistics note warns that combining repeated observations from several people and calculating a correlation as though they formed a simple sample can be misleading.^[1] This is a methodological caution, not evidence that NSI predicts behaviour. Before interpreting a pattern, ask who contributed the rows and what comparison the analysis is intended to represent.

Separate two questions that sound similar

Imagine two fictional participants. One usually reports higher pressure than the other, yet each participant’s own score hardly changes from session to session. A difference between their average scores does not, by itself, show what happens when pressure increases within either person. Conversely, a person’s changes over time may be interesting even if their average resembles someone else’s. These are different questions, and this example reports no actual NSI result. Write the intended question in ordinary language first: “Do people differ?” or “Does this person’s behaviour change when their reported state changes?” A pooled headline should not silently answer both.

Keep participant and session structure in the record

Use consistent anonymous participant identifiers, session dates and a clear definition of each recorded outcome. Preserve missing entries as missing rather than filling them with a convenient assumption. The Observations That Never Reach the Journal develops that problem. An analysis plan should explain how repeated entries will be handled; ask a qualified statistician to select an appropriate method rather than treating this article as an analysis recipe. Write the Test Before Seeing the Outcome connects the data structure to the original research question. More observations do not remove the need to specify what those observations represent.

Make the limit visible in the conclusion

A responsible report states whether a result concerns differences between participants, changes within participants or both, and explains the uncertainty and comparison used. Do not transform a group description into a personal alert threshold without separate evaluation. How Large Is the Difference? adds the magnitude question. NSI.GOLD and The Gold Nervous System Indicator introduce a proposed framework for investigating the human trading decision. They are not evidence of a validated physiological instrument, a clinical assessment or a reliable trading signal. Clarifying whose change is being described is one step toward a testable question, not a shortcut around validation.