I've been publishing the Renew / Cancel Index (and using it to predict the future of scripted broadcast shows) for the last two seasons.
It's been in basically the same that entire time, a show's season to average adults 18-49divided by its network's season to average adults 18-49 rating. In the past that ratio indicated quite well the future prospects for scripted shows (particularly one hour dramas, a bit less so for half hour comedies).
Two limitations have dogged it though:
- Unless shows on a network have a similar number of repeat episodes, the individual show ratings averages become apples to oranges comparisons solely because of repeat / new episode numbers, which doesn't have any influence on renewal decisions.
- When a network has unscripted programming (whether reality, sports or news) that either makes up a large part of its schedule or has ratings far different than the rest of its schedule ( , ), I ended up twisting my methods in knots, or just giving up, trying to remove those effects.
What I'm likely to begin doing next season, subject to testing after the end of this season, are make the following changes.
- Instead of the numerator of the Index being a show's overall 18-49 average, it will just be its new episode 18-49 average. That eliminates the problem comparing shows with very different numbers of repeat episodes. It does introduce another issue that some shows have higher repeat ratings than others, but I believe that issue is less influential than the number of repeats.
- Instead of the denominator of the Index being a network's overall 18-49 average, it will just be the scripted show new episode adults 18-49 average. That eliminates the problems associated with averages being influenced by sports, reality and news programming.
Again, I won't do this until next season, and then only after seeing how it turns out to have worked after looking at this season's final results, but here's how things would stack up right now, if this method was being used.
|Better Off Ted||0.40|
|Ugly Betty (Fri) (S), (F), (P)||0.43|
|The Deep End||0.50|
|Ugly Betty (Wed) (S), (P)||0.54|
|Brothers & Sisters (S), (P)||1.06|
ABC is not significantly changed by this new method because most of its episodes have similar repeat patterns and it's sports/reality/news programming has the less influence on its overall ratings average than NBC and Fox. Small shifts, but no big changes to the list compared with the way I'm doing it now.
|Three Rivers (P)||0.55|
|Numb3rs (F), (P)||0.56|
|Medium (F), (P)||0.62|
|New Adventures of Old Christine (S)||0.64|
|Ghost Whisperer (F), (S)||0.64|
|Cold Case (P)||0.64|
|Accidentally On Purpose (P)||0.96|
|: Los Angeles (P)||1.17|
|: Miami (P)||1.20|
CBS's list is similar to ABC in that little has changed vs. the way I am doing it now. Although mid-season shows with no repeats like will be more appropriately compared once they join the list.
|The Beautiful Life: TBL (P)||0.43|
|Melrose Place (P)||0.66|
For the CW,'s position is much more appropriately indicated vs. the current method. and 90210, which have already been renewed, are more appropriately ranked when the effect of their large number of repeats is removed.
|Til Death (S), (F)||0.24|
|Dollhouse (F), (P)||0.28|
|Til Death (Sunday)||0.32|
Fox's comparisons vs. other networks become possible after its average has unscripted shows removed from it. No more "Fox only" disclaimers about how their Index numbers are no longer comparable. and 24's results are a bit more appropriately compared because of their lack of repeats.
|Law & Order (F), (P)||0.62|
|Parks & Recreation (P)||0.88|
|Law & Order:(P)||1.11|
Without in the mix, NBC's Indexes become more comparable to other networks as well. Chuck's position (as well as in the future) is more appropriately indicated after factoring out the repeats in other show's averages.
Overall, at this point in the season, I like the results of the Experimental Index. I won't be publishing it on a weekly basis this season, but I promise to do it at least once more before all the renewal announcements.
I just thought of another reason my new method is *conceptually* better (although, again, whether it works or in practice is the real test).
Using a networks overall adults 18-49 average weights hour long show ratings 2x the value of half hour shows. But the overall average I calculate in this experimental version weights all the show ratings equally whether they are 1 hour or 30 minute shows.
In practice the difference may not be much, but my guess is that not time weighting the ratings average in the denominator makes more sense. When ranking shows by relative average, why should longer shows count more?