Experiments

If this is in fact an experiment, it requires a hypothesis.  The most elementary format for a hypothesis is an "if...then" statement, such as "If I drink three glasses of wine, then I will get a raging headache."  Or "If my daughter is still sleeping, then she stayed out later than she promised."  As you (and my eighth grade students) can clearly see, imbedded in these statements are the independent and dependent variables that define the structure for experimental design.  Whether I suffer another raging headache depends on my independent choice of how much wine I consume.  Whether my daughter lied again depends on her independent choice to sleep in...again.  We test the truth of the hypothesis by playing with the independent variable (let's call that the IV) to see what happens to the dependent variable (DV, see?).  Thus, tonight I might swill two glasses of a 2008 Gnarled Vine Chardonnay (grade B) with no observed change in the DV (my throbbing temples), and tomorrow night I'll down three glasses and spend the day after tomorrow hugging pillows and popping Imitrex.  A good experiment would test at least three levels of the IV, but my prescription is running low, and I'd hate to waste a decent Chard on anything other than cooking.

As I continue to chase this particular rabbit, I see a problem (a.k.a., another rabbit hole) that I've encountered before:  My second example hypothesis is, well, a mess.  This is when I give my students a patient look and say "Not all if-then statements are hypotheses, and vice versa."  Observe:  While I might be able to design an experiment to test the relationship between my daughter's somnolence and her nocturnal behavior, it would require me to record both her sleep habits and her nighttime prowling, and THAT would require a level of truthiness and/or direct interference that would simultaneously destroy both the experiment and the carefully negotiated peace that currently prevails in our house.  In other words, dear students, some hypotheses are NOT worth the funding offered to pursue them.

Let me climb out of that last hole by saying that a hypothesis is much easier to test if it's focused on measurable (preferably numeric) phenomena or entities.  Behavior is measurable, but so flaky, IMHO.  Oh, and guess what.  Here's another hole.  If this blog is an experiment, it's focused on MY behavior.  What now?

• Quantitative:  "If I post to this blog once per week, then my writing skills will improve by 74%."
• Qualitative:  "If I post to this blog once per week, then I'll develop followers who will give me constructive feedback and help me improve my writing skills."
• Human:  "If I publicly explore mental rabbit holes in writing on a regular basis, then my fragile ego will be ripped to shreds by a hundred faceless voices."

Maybe I'll save that last one for the null-hypothesis test.

Fondly,

Sam