TV ratings explained
The race made the news when the overnight Nielsen rating of its U.S. television audience dropped to 3.6 from 4.1 last year, and the local paper further reported that this constituted a 12-percent ratings drop for the 101st running of the race.
It’s easy to make a few quick conclusions with those numbers, but there’s a lot going on behind them that explains how they work. Television ratings are like the unit of calculus you never quite grasped. You know when a number seems way off, but you don’t really know why or how.
Before we crunch the numbers, let’s talk about what exactly we’re crunching.
How Nielsen Does Its Research
Nielsen Media Research does most of the U.S. television ratings and research you hear about in passing, and it’s the company that put together the ratings for the 2017 Indy 500. The company has been researching television ratings and habits since 1950, and Forbes reports that the company studied a sample size of about 41,000 households as of 2016.
Nielsen estimates that there are 118.4 million households in the country this year. That means the study sample Nielsen uses is just over 0.03 percent of the actual U.S. households with televisions. Nielsen intends for its sample study, like most valid research, to represent a diverse population in regards to gender, age, geography, cable status and the like, according to Forbes.
Chosen households that participate in the research are compensated by Nielsen, according to Forbes. Here’s a basic breakdown of what chosen households have to do, from Forbes:
Once they have their representative sample in place, Nielsen primarily measures viewing numbers with electronic meters that track what the televisions are tuned to. They also install black boxes – aka a computer and a modem – which deliver the viewing data from the TVs to Nielsen every night.
Nielsen keeps track of who and how many people are watching programs. Each member of the Nielsen household is given a special button that they turn on and off when they start and end a program. This viewer specific information is also transmitted to Nielsen through the black box.
But not everyone watches on a television, and Nielsen’s had to adjust to that in measurements on its studied sample. From Forbes:
With the rise of DVRs, OnDemand channels and instant watch mobile apps, people in the United States can watch shows whenever they want. …
Nielsen now keeps track of all viewing enabled for measurement across all platforms, including computers, tablets and smartphone devices, happening up to seven days after an original broadcast when calculating ratings.
But that only tells us how Nielsen does its studies, which isn’t by spying on your every move in your own living room. What it doesn’t tell us is how to read the company’s final ratings numbers, which is a whole other mystery in itself.
How To Determine The Real Numbers
Back to that Indy 500 ratings crisis. We’ll use that to do this weird, mysterious Nielsen math, and we’ll find out exactly how many households–estimated households, that is–tuned into the race.
The overnight Nielsen rating for this year’s Indy 500 was a 3.6, compared to a 4.1 from 2016. (The final rating that came out later and was a 3.4, but for these purposes, we’ll use that initial the 3.6 rating.)
It doesn’t take a lot of work to see that going from a 4.1 to a 3.6 isn’t great, but it’s still not clear what that means. The easy guess would be that maybe it’s in millions, and that a 4.1 rating means 4.1 million people watched it in 2016 and 3.6 million did in 2017.
That’s not how it works, although it’s oddly close to what the actual figures meant this year. Finding the real numbers takes a lot of effort and a little bit of math.
Individual people don’t factor into the ratings in the first place, households do. And if you want to know how many households watch an event based off of a Nielsen rating, you’re going to need to bust out a calculator every single time.
The numbers in a Nielsen rating act as a percentage. Nielsen estimates a total audience for a certain span of time, and ratings count as percentage points in that estimated audience. For the 2016 and 2017 television seasons in the U.S., Nielsen estimated 118.4 million households have televisions.
That marks a 1.6-percent increase in TV ownership from the previous year, which, like all of Nielsen’s numbers, looks straightforward and absolute. But it’s actually pretty uncertain, since “TV ownership" isn’t clear. Nielsen doesn’t say whether that number includes all televisions, or just televisions with the access to programs that it’s measuring. With the growth of modern streaming services, not everyone with a television has a subscription that includes the Indy 500.
Regardless, those are the numbers we have. There are 118.4 million households and each point on a Nielsen rating represents a percent of that total number. Someone familiar with Nielsen’s calculations confirmed to Jalopnik that this is how the company calculates its ratings, so let’s do the math:
- Nielsen estimated 118.4 million households in the U.S. for 2016 and 2017
- A point on the Nielsen scale is a percentage of that total
- The Indy 500 dropped from a 4.1 rating in 2016 to a 3.6 rating in 2017
If we look at the 2016 numbers, Nielsen’s 4.1 rating means 4.1 percent of its 118.4 million U.S. households tuned in for the Indy 500. That calculates out to 4.8 million households watching the race. In 2017, the 3.6 percent means that nearly 4.3 million households were included in the overnight rating. We’ll do the exact numbers and see what the real story was behind this 12 percent drop:
- A percent in the current Nielsen estimate represents 1,184,000 households
- In 2016, 4,854,400 households were included in the overnight rating
- In 2017, 4,262,400 households were included in the overnight rating
- Overall, there were 592,000 fewer U.S. households watching the Indy 500 in 2017
When you think about it, it’s hard to say how important that 592,000 figure is–even if it does constitute as “12 percent." The number is made up of estimated households rather than individual viewers, and it’s based off of a small sample size in the first place.
Even then, there’s a lot more that the figure could be hiding.
Ratings also aren’t in people, as explained earlier. They’re in households. You could be watching the Indy 500 all by your lonesome at home, whereas your neighbor could have 30 people over. Together, you and those 30 other people count as a total of two households. It’s a strange system. (Update: We received an email from a person who says his home is a Nielsen household, and that Nielsen asks for the genders and ages of each person watching. The person also said if he’s watching something in one room and his wife is watching something else in another, each will get household credit for however many households the home overall represents. We’ve reached out to Nielsen to check on this, and to see how it equates.)
It could also hold a lot more weight when we look at big events like the Indy 500. You’re much more likely to invite a bunch of people over for an annual event than Jeopardy! every night of the week, but all of those programs are judged by the same system–households.
And if a small percentage of the households that watched the race both years invited a couple more people over in 2017, it could take a huge chunk out of the estimated 592,000-household drop. We’ll never know, because the ratings don’t measure that.
Ratings also don’t measure audiences beyond the United States. The big news at the Indy 500 this year was Spanish Formula One star Fernando Alonso flying over to America to race. It’s safe to assume that viewership went up in his native Spain, possibly making up a significant part of Indy’s ratings drop. But none of that is included in Nielsen’s U.S.-only figure.
With all things considered, we can’t truly know how much worldwide Indy 500 viewership dropped by–if it did at all. That’s just part of the reason why ratings in the U.S. are so hard to go by.
How The Numbers Change Over Time
In the world of Nielsen ratings, there are only a few cases when you can truly, number for number, compare one set of television ratings to another. Nielsen’s estimates are more about percentages and market shares than actual viewership numbers, so numbers can change over time while ratings stay stagnant.
Nielsen usually comes out with a new estimate of households with televisions every two years, which has big effects on how the ratings look on the surface.
If we compare the Nielsen estimate from 2017 to that one from 2014, the same rating will produce different numbers with our equation. Nielsen estimated an audience of 115.6 million households for the 2013 and 2014 television seasons, which would make a percentage point–and therefore a full point on the rating scale–around 1.156 million households. A percentage point in the 2016 and 2017 Nielsen ratings is the equivalent to 1.184 million.
That’s another additional 28,000 households lumped into a single percentage point. If audiences for certain programs or events aren’t expanding faster than the audience is, it subsequently makes ratings look worse or stagnant as the years go on.
For example, the Indianapolis Business Journal’s breakdown of the ratings said this was the “lowest-rated since ABC began start-to-finish coverage of the race in 1986." But that doesn’t tell you about the actual estimated viewership–only the market share across all households with televisions.
The number of viewers tuned into the 1986 Indy 500 was a lot higher than this year, but you’d have to do the math to figure that out–just like for every other year. Sure, we can do the math as an aside, if you really feel like it:
- Nielsen estimated the number of households with televisions in 1986 as 85.9 million, according to the Chicago Tribune
- The Indy 500 had an 8.8 rating in 1986, according to the New York Times
- Dividing the 85.9-million audience by 100 and multiplying it by 8.8, you get nearly 7.6 million homes tuned in
- But if we looked at ratings alone, an 8.8 would be over 10.4 million viewers at current Nielsen household estimates
That’s a lot more people for the same rating. And sure, it would make sense to assume that audiences for programs would go up as the number of households with televisions does, but that isn’t the only important number that’s changing.
The more programs and streaming services there are, the more choices people have about what to watch–these days, cable subscriptions come with hundreds of channels. That wasn’t the case in 1986, meaning broadcasts like the Indy 500 were fighting in a much smaller arena than they are today.
It’s kind of like choosing a dinner spot in rural Kansas versus choosing one in NYC–the more things people have to pick from, the less likely it is that they’ll all end up in the same place.
Without knowing any of that, the entire landscape of television ratings can be a bit of a mystery beyond the rises and drops you see on the surface. So, don’t ignore the messages television ratings send entirely, but go in with a healthy amount of doubt–and maybe a calculator. Alanis King of Blackflag