Subbing in for Matt Yglesias, Sean Vitka looks at a new report from the National Federation of Independent Business, the trade association for small businesses, and says that “No, Obamacare Hasn't Led Small Businesses to Cut Employees' Health Care”:

Just a reminder: all that hype about companies cutting employees' hours to under 30 per week turned out to be wrong too.

For the record, cogent criticism of law is good. And in today’s climate, we should all be happy that NFIB has the integrity to release surveys that don’t back up their own case, and honestly answer questions that undermine their political leanings -- even if their PR department hasn’t gotten the memo. Hopefully this report will help inform everyone’s opinions.

Here’s the full report. They hid the good news on page 26: "If small employers follow those plans, the net proportion of them offering would rise, breaking a decade-old trend."

Good news! But here’s a surprising thing: The numbers in that report seem to be at odds with its words.

Here’s the report's table on employers who plan to drop health insurance versus employers who plan to add it:

If you look at the percentages, then Vitka’s point is correct. Thirteen percent of employers who do not offer insurance plan to add it, while only 7 percent of those who offer insurance plan to drop it. Unfortunately, percentage is not the correct way to look at it.

To see why not, let’s look at some other numbers in the table. Out of a total sample of 921 employers, 668 said they offer insurance, and 253 said they do not. So, 72 percent of businesses surveyed offer insurance, and 28 percent don’t.

Seven percent of 668 employers, or 46, are thinking about dropping insurance. Meanwhile, 13 percent of 253, or 32 employers, might add insurance. That is a net decrease of 14 employers, or about 1.5 percent.

Either the table is incorrect, or the NFIB (and Sean Vitka) seems to have misread its own table. Or, I guess I could be missing something. But I’ve been staring at the table for 20 minutes, and I’m unable to see what.

Meanwhile, if you look at another column, it looks even worse: The NFIB has 48 percent of employers saying they will “probably” or “definitely” offer insurance, while 48 percent say they will “probably” or “definitely” not. If you assume that this is what will actually happen, this would mean that the number of employers offering insurance will fall from 72 percent to 48 percent.

The numbers in that column also don’t comport with the numbers in the other columns: 90 percent of employers who offer insurance say they will “probably” or “definitely” offer insurance next year, which is 601 employers, or about 65 percent of the total. I have no idea how they got 48 percent in the total column: one or the other must be off.

We may never know whether employers are dropping insurance in response to the Affordable Care Act; the number is inherently hard to measure, because employers add or drop insurance for their own reasons all the time. But this report doesn’t seem to tell us one way or the other.