Saturday, June 21, 2008

The minimum and the maximum are converging

Sylph asked me to generate this plot, and it is very interesting:



As in, the maximum is trending to the minimum. The minimum, which used to be the threshold of cancellation, has stayed pretty flat for 50+ years. And now the maximum is trending toward the minimum, suggesting the best rated of the genre is trending toward the threshold of cancellation.

Yup, I'd say it was interesting.

Thanks for proposing it.

I should also add, because I am very interested in variability, that I find it interesting that the lines have SMOOTHED over the recent years. That implies several possibilities to me:

First, the network emphasis on reducing pre-emptions post-OJ has caused greater stability of viewership.

Second, we're down to a core loyal audience that does tend to stay tuned no matter what sh*t makes it to air :-).

I do wonder what the HELL happened in 1965 to spike the minimum AND maximum so much :-). Soaps were apparently ON FIRE that year. The years of minimum spikes coincide with the introduction of major new shows (e.g., Y&R in 1972-73 season), so one gets a sense that there were times when new shows actually GREW THE BASE for soaps overall. I'll bet there was actually promotion back then!

Because the variability can make it hard to see the real TREND, here are the smoothed linear trends for the minima and maxima. Same story, but a little clear to see without the vagaries of seasonal variation.

It actually looks a little better this way, but I caution it is misleading. As you can see, the maximum line over-estimates the true maximum in recent years? Why? Because the REAL trajectory of the maximum rating is -- brace yourself -- quadratic or accelerated. In recent seasons, the true maximum is declining at a faster rate than this line implies.

So, below, I now show the quadratic trends. These are a little misleading too (remember...they are model-based estimates, not actual data), but they tell stories I didn't expect. The model implies that (a) the minimum has bottomed out! We're there! and (cool.gif the maximum is trending toward the minimum, at an accelerated rate...but not as fast as I subjectively felt. Small mercies :-)

Linear model-based estimates of minimum and maximum ratings decline:



No comments: