Key numbers (at least for those of us who are numerically inclined) tend to stay in one’s mind. That’s the problem with pseudo-numbers. We may first see them with flashing warning lights (such as “X% at a 95% confidence level”), but the caveats fade and the numbers remain when the numbers are public and it's election season. Whether this number goes up or down (even if by less than 1%), the impact is felt by all of us. Yet, even these official numbers are often suspect.
Unemployment rates are a good example. It turns out that the published number each month is not a very reliable number (nor very valid). First of all, it is calculated by dividing the number of people who have jobs by that number plus those actively seeking employment. So, as we are sometimes reminded, those who have given up looking, but would like to work (I guess if a job magically appeared in front of them) are not included. Okay, so I guess that’s a defect. But, it’s a pretty small one given the fact that the entire calculation is suspect. Pointing at this defect is like the misdirection of a good magician. Seriously, do you really believe that looking at official W2s and/or those filing for unemployment benefit and/or a gigantic survey can really produce an accurate percentage (much less to the tenth of a percent) of those residents who are working full-time or even part-time in the U.S.?
Wouldn’t we be better off being precisely right (i.e., reporting the number seeking employment and/or the number of those receiving W2s each month) than reporting a precise number (the official unemployment rate) that we know is wrong?