Last night I had a dream that I was studying the happiness of 5 year olds at a birthday party. I had a sample size of 27, no power analysis, so no sense if that was big enough - so palpitations already... I had no well established tool for which to measure said happiness, so planned to use a visual analog scale, yuck. More palpitations... Then I ran a logarithmic regression that seemed off, so I ran the presence of a magician as a mediator variable, and ta-daaaaaaaaa... I had an interaction.

So my findings in my nighttime subconcious were that being at a birthday party did predict happiness in five year olds, but that that happiness was dependent on the presence of a magician.

Can you guess what I did on Thursday and Friday? (hint, one day involved log reg and mediator variables, and one day involved kids at a birthday party)

Amazingly, I DIDN'T dream about the articles on kangaroo care in newborns that I've been reading.

ETA: The real irony is that logarithmic regression was the wrong test to use if I actually had a visual analog scale - it would have been correct if I'd had only a yes/no outcome.

**sigh**