What it took
One dataset, nine different routes through it, and nothing shared between them.
One dataset, nine analyses, and no two alike
A whole cohort in one department sat the same statistics course and was issued the same dataset. Nine of those analyses were mine to build, and every one had to be genuinely different from the other eight, because anything less would have been picked up. All nine passed, and passed well.
The situation
A cohort in the same department sat the same statistics course, and every one of them was issued the same dataset. Nine of those analyses came to me. The constraint was the whole problem: nine pieces of work out of one dataset, and no two could resemble each other.
What I did
I treated each one as a separate piece of research. Different questions put to the same data, different tests where the data allowed it, different structure, different write-up. Nothing was carried across from one to the next, because anything carried across would have been spotted.
What came out
- All nine passed, and passed well
- No two of the nine analyses resembled each other
- The same method has since carried several systematic literature reviews and related coursework
