My first social experiment, a study in jury dynamics, was an utter failure. Although my college professor assisted me, I was still in control and being an educator she allowed me to learn from my mistakes. Thank you very much Barb, I did learn a lot.

You see, I included too many variables and the wrong type of measurement.

I separated the participants into groups of 12 and provided them real life jury trials from around the country. Cases included murder, burglary, assault, auto theft, and an auto accident.

My variables included age, ethnicity, number of children, education, type of job, marital status, suffered from a crime, know someone who suffered from a crime, and religion.

Then instead of “guilty” or “not guilty,” I had the participants indicate the level of guilt.

The two factors that caught me ill prepared were ethnicity and level of guilt.

I left ethnicity open to interpretation and came away with people identifying themselves as W.A.S.P., Latino, White, African-American, Cherokee and many others. I certainly learned why the government limits responses to Caucasian, Hispanic, Eskimo, Native American, African-American, and Asian.

I thought the level of guilt would provide an interesting insight into personal dynamics and points of view. Unfortunately, this turned into a mess. The responses were not simple black and white responses. I had results on a 0 to 10 scale of 3.12, 3.45, 6.789, and other wacky numbers.

From this fiasco I learned that scientific studies use tests with limited variables. Adding to many variables causes confusion and becomes too hard to track.

Now, I use the lessons I learned from that experiment to plan and implement conversion optimization studies. These studies require that I develop a hypothesis before we test which aids in limiting the number of variables we use in the conversion optimization testing process.