If I had to dream up a depressing narrative for how Americans react to economic data, it would go like this: When the reports are bad, they cling to granular and statistically insignificant numbers as proof we're heading into a new recession. When the reports are decent, they conclude the numbers are cooked by a government conspiracy to dupe voters.

Sadly, this is reality. Friday's employment report showed the economy added 114,000 new jobs in September, with the unemployment rate falling from 8.1% to 7.8%. Within minutes, former General Electric (GE 8.28%) CEO Jack Welch took to Twitter to suggest the numbers were a conspiricy to help the president: "Unbelievable jobs numbers...these Chicago guys will do anything...can't debate so change numbers." Dozens followed him, and by Friday afternoon most news stories covering the employment report were devoted to either the conspiracy line or its debunking. 

Of course, there is no evidence whatsoever that the numbers were manipulated. Full stop. Keith Hall, a former Bureau of Labor Statistics commissioner appointed during George W. Bush's presidency, said it would be virtually impossible to cook the numbers without getting caught. Former Department of Labor chief economist Betsey Stevenson wrote, "Anyone who thinks that political folks can manipulate the unemployment data are completely ignorant of how the BLS works and how the data are compiled."

Oddly, Friday's report wasn't even that good. It showed a slight bump in an otherwise muffled trend of jobs growth barely sufficient to keep up with population growth.

So why the conspiracy fuss? I think it's about two questions.

1. If only 114,000 jobs were created, how did the unemployment rate fall from 8.1% to 7.8%?
The reported number of jobs created and the unemployment rate come from two entirely different surveys. The first is from a survey of businesses. The second, from which the unemployment rate is derived, comes from a survey of households. The business survey is sent to roughly 140,000 businesses nationwide. The household survey only goes out to about 60,000 homes, making it much more volatile.

Even though September's headline of 114,000 jobs created was weak, the unemployment rate fell sharply because the household survey showed a gain of more than 800,000 workers. Month to month, it's entirely normal for the two to show conflicting trends.

2) OK, so you expect me to believe 800,000 jobs were created last month?
No, but you should never take employment numbers at face value. These aren't meant to be precise figures; they're broad estimates based on surveys with an admitted level of imprecision. The household survey has a margin of error of plus or minus 400,000. So when it's reported that 800,000 more people were working last month, what it really means is there's a 90% chance that the gain was somewhere between 400,000 and 1.2 million. But we really have no idea. And we never do.

Beyond the imprecision, wild month-to-month volatility is the norm with the household survey:

Source: Bureau of Labor Statistics.

To be incredulous of jobs numbers that are abnormally high, you also have to be skeptical of those that are abnormally low. Did the economy really add 532,000 jobs in Jan. 2010, as the household survey showed? Probably not. But it probably didn't lose 639,000 jobs the month before, as it also showed.

The same goes for last month. We almost certainly didn't add 800,000 jobs last month. But we also probably didn't lose 314,000 in July and August, 169,000 in April, or 423,000 last June.

What matters here are broader averages. The household survey's six-month average change is now 130,000 per month. For comparison, employment numbers compiled by ADP (a private company) show an average monthly gain of 161,000 over the same period. If the president is cooking the numbers for political gain, he's doing a terrible job of it.