This post has been updated with new Nielsen numbers.
When it became clear that Donald Trump planned to skip Thursday night's Republican debate, we tried to figure out how ratings for Fox News, host of the event, might suffer.
There were two possible scenarios, it seemed. In 2012, ratings increased as the Iowa caucuses (which are Monday) approached. But overall, ratings have been trending down in 2016's debates. The 2016 debates have aired on two tiers of cable networks: four on Fox News and CNN, and three on Fox Business and CNBC. Which gave us a picture that looked something like this.
If the trend of that top red line continued, we figured something like 16 million people might tune in to Thursday night's debate, all other things being equal. Trump dropping out, of course, made things unequal.
On Friday morning, CNN's Brian Stelter reported early numbers from Nielsen: The debate got an 8.4 household rating, meaning that 8.4 percent of Nielsen viewers watching TV during the debate were watching Fox News. That's a slightly different metric than the millions-of-viewers we'd been using, but we can still take a look at what it suggests. (Official numbers from Nielsen will be out later on Friday.) In Stelter's estimation, the audience will perhaps hit 13 million people.
If the trend for the upper-tier networks continued at the same downward rate, we'd have expected the final household rating to be 9.7. (We're comparing apples-and-oranges a bit. Stelter's number is the "metered markets" household rating -- but that has tended to be higher than the final household rating.)
Update: We now have final numbers from Nielsen, in which the final household rating is at 7.2, lower than Stelter's rough figure. Here are the ratings by household percent and raw viewership.
Our estimate is a very rough calculation, based on not-very-many points of data -- and on a linear trend downward. Maybe the trend would naturally have bent more quickly downward than moving in a straight line.
That's the thing about it: We can never know. We can't run a controlled experiment with two debates, one including Trump and one not. Which some observers might think Trump would use to his advantage, tweeting about how Fox News's ratings sank due to his absence. We can't run an experiment on whether or not he'll do that, either, but I think we all know what the data would show.