As I've said, most of my work this summer has gone into working on a competition adjustment that would give a fair shake to low-viewed, low-competition periods like weekends and holidays. I'm saving that portion for tomorrow and just focusing for today on the adjustment based on the actual competition level.
Once again, I'll start with the thing I did not change: and once again, it's the "impact" measurement. Last year, a relatively formal study of big competition changes came up with a 3.75% change in the show's rating for each 1.0 rating change in competition. I thought I might have to change that since I was redefining competition and everything, but ultimately it seems like that's still a pretty good number. I tried tinkering with it a bit, but generally that seems pretty close to where the deviations are lowest. Again, maybe next year I'll re-evaluate this more formally.
Definition of Competition
Much like the "gap" thing I put into the lead-in adjustment last season, my definition of competition was another thing that seemed to overvalue big shows and undervalue smaller shows. Essentially I counted "competition" as "all broadcast shows in the timeslot" rather than "all broadcast shows in the timeslot EXCEPT itself," meaning a huge show would get full credit for "facing itself."
On the other hand, a pure "all shows EXCEPT itself" measure also seemed a bit unfair to the big shows, because some shows are big enough that they actually depress the ratings of opposing shows. For each bit that your sheer hugeness bites out of your competitors, you get penalized.
My compromise to this has been to apply a competition adjustment... to the competition adjustment. In other words, try to bring a program's competition up/down a bit based on the size of the program itself. So to the competition calculation I add 3.75% times the difference between the show's rating and a normal primetime program (5.5% of the PUT, also used as the normal lead-in). This is not exactly a huge number, but it seems to paint a better picture than either "all programs" or "all other programs" on their own.
An undeniably big problem with doing a timeslot metric is the fact that we almost always don't have numbers for Fox and (much less importantly) the CW at 10:00. There are a lot of ways to go about this, all of which are fairly uninformed. What I did last year (add the same number every time) wasn't completely awful, but it turned out to be something of a burden in weekend/holiday times, when the 10:00 hour appeared to get too much credit. To mitigate that effect, this time I'm just adding a percentage of the viewing levels.
What I did was look at the broadcast percentage of the PUT in each half-hour and try to normalize it at whatever the average is from 8:00 to 10:00.
So I add those percentages of the PUT under "diff" to each competition total, putting each half-hour (counting all shows) at about 29% on average. I'm not really sure making every half-hour "normal" is a good idea, because maybe some half-hours are just more competitive than others, but since 8:00 through 10:00 are all about the same (before the adjustments), it doesn't seem like it'll be too egregious a mistake.
The Competition Adjustment So Far
OK, so that's pretty much everything I can cover without having to go into the "holiday number," though there are a couple things beyond the holiday number that I'll need to hit in tomorrow's post as well. Here's what we have for competition so far:
Competition Impact = 3.75%
Competition = Sum of ratings for all OTHER broadcast shows + 10:00 Hour Adjustment + Impact * (A18-49 - 5.5% * TPUT)
Competition Adjustment = Impact * (Competition + Holiday Number - Expected Competition)
We'll do the stuff in blue tomorrow!
True2 Week: Viewing | Lead-ins | Competition pt1 | Competition pt2 | Wrap-Up