Oscars Ratings Down 20% — Is That a Data Point Worth Reacting To?
Another year, another evening of the Academy Awards, a.k.a. The Oscars. I don't watch the show, but I do pay attention to the inevitable headlines about TV viewership or ratings being up or down — usually a two-data point comparison, which isn't very helpful.
When we think in terms of statistics and longer-term trends, having a data point that's “an all-time low” doesn't mean that data point is a “signal” or that it's statistically significant.
Even if the number of viewers was just fluctuating around a stable average over the last 20 years, there's going to be an “all-time low” data point there out of those 20, by definition. That doesn't mean the “all-time low” is worth explaining. The “lowest ever” doesn't mean anything has changed.
We shouldn't jump to root cause analysis, nor should we jump to countermeasures — unless that number on Sunday was a “signal” that the system (the system that generates a number of people watching) has changed.
Some headlines and articles did a two-data-point comparison that's mathematically correct but potentially unhelpful.
“… plummet almost 6 million viewers from 2019”
What does that tell us? What does it mean that viewership was:
“… a 20% drop-off…”
The 2019 ratings were a 12% increase from 2018.
We could continue the two-data-point comparison game (as many of our workplaces do):
- 2018 ratings were down 19% from 2017
- 2017 ratings were down 4% from 2016
- 2016 ratings were down 6% from 2015
- 2015 ratings were down 16% from 2014
Oh, so maybe there is a trend. Was the 2018 increase an exception to a recent rule?
As I wrote about in Measures of Success, the media often tries to explain every up and down in the number, which can be a fool's errand.
Ratings were up in 2019 — some speculated it was because it was the first year without an official host. But, they repeated the “no host” experiment and ratings were down. Ratings were lower than the last year with a host (2018). So, who knows, right?
What is helpful is visualizing more data points into even just a simple “run chart” (or what's called a “line chart” in Excel):
What do we see? Whoa! Instead of quibbling over or explaining the last three years and the up/down pattern, look at the big picture.
There's a pretty significant long-term downward trend… the numbers each decade get lower. The averages for each of the last three decades:
|2001-2010||39.2 million (down 15%)|
|2011-2020||34.4 million (down 12%)|
My advice: Stop explaining each year's numbers and look at longer-term causes. The downward trend in ratings is often attributed to the rise of streaming platforms… what the Academy actually do about that? They're never going to “fix” the Oscars if that means restoring past glory (and high viewership). It seems that there is a new reality.
Better than a “run chart” is a “Process Behavior Chart,” as shown below:
The PBC makes it a bit more clear that there's been a downward shift. The number of viewers was originally fluctuating around an average, but then we saw 8+ consecutive data points below the average (a shift in performance).
Since 2004, the number of viewers was fluctuating around a lower average… until the past few years when we see evidence of another shift.
What has changed the last three years to trigger two “signals” (data points below the Lower Limit of the PBC)? That's a better question than, “Why is the 2020 number a little bit below 2019's?”.
Here is a LinkedIn post that I shared the other day that triggered a little bit of discussion on the data and the chart: