Thoughts From the Big Chair: TV’s Late Night Talk Shows and Johnny Carson – A Quantitative Analysis Part II

Thoughts From the Big Chair
Comments on Television and Associated Media From a Lifelong Addict
TV’s Late Night Talk Shows and Johnny Carson – A Quantitative Analysis
By RALPH E. CHAPMAN (Twitter @RalphEChapman) ©2013, Ralph E. Chapman
Part II – Analyzing the data with Lots of Comments

It’s time to finish off the late night talk show discussion started in Last Week’s Column

But first a Newsflash!!

Before I get down to the analytical part, a few things have come up since last week’s column was submitted.

The Big News that’s Developing!

A few weeks ago, it was strongly rumored that Fallon would take over The Tonight Show from Leno sometime in 2014. This seemed natural to me and the grading and analyses I’ve done suggests he’s pretty much ready now to step into the time slot and start building his new and hipper audience.

Then last week, Fallon had a full week of special shows all featuring Justin Timberlake with lots of skits and new music by Timberlake, and lots of energy. To me, it seemed like Fallon, probably with NBC help, was making the case to become the next host very strongly.

They were great shows and I suspect Leno noticed the push. Monday, Leno not so subtly referred to the NBC execs as snakes within one of his jokes. On Tuesday (the 19th) he complained about NBC’s poor showings in the ratings and he has continued and stepped up the barrage all week.

So Leno seems to be feeling the heat, and I will resist the temptation to discuss the irony and karma of this situation relative to past events.

Tuesday on Fallon, comedian David Steinberg was the second guest. They talked a lot about the fact that Fallon’s studio was where Carson hosted The Tonight Show for its first ten years (while in NY) and how nostalgic it was to be there.

This may not sound like much, but Steinberg was the second-most featured guest on Carson’s old show – he was on it more than 150 times – and his whole career was linked to The Tonight Show. This really came across as a direct salvo aimed at Leno; it was Fallon placing his claim as the natural choice to be the next host of The Tonight Show, and soon.

The rumor mill suggested, and it now has been officially confirmed,  that The Tonight Show will be moving back to NYC in 2014 under Fallon, and going back to its original 90 minute length. So we’ll keep an eye open and see what happens; Leno cannot be a happy man as he is obviously being pushed out.

The schedule seems to be sometime in 2014 but I suspect the transition may happen earlier if the anger really boils over with Leno.

There has also apparently been a decision made that Seth Meyers will take over Fallon’s later time slot when Fallon moves to do The Tonight Show.

Back to the Analysis

So a quick review – the shows considered are (E = Early, typically 10:00 to 11:00 PM here in Mountain Time; L = Late, typically after 11:00; NY – recorded in New York; LA = recorded in L.A.) :

0. The Tonight Show Starring Johnny Carson(NBC, L, NY then LA).

1. The Late Show with David Letterman(CBS, L, NY).

2. The Tonight Show with Jay Leno(NBC, L, LA).

3. Conan O’Brien(TNT, L, LA).

4. Jimmy Kimmel Live!(ABC, L, LA).

5. Late Night with Jimmy Fallon(NBC, LL, NY).

6. The Late Late Show with Craig Ferguson(CBS, LL, LA).

These were graded for the following show segments/categories:

A. the Monologue: The introduction, the first segment, the welcome to the show.  

B. The Side Kick: The star’s support structure.

C. Music / Band: The musical support throughout the show, typically provided by the house band.

D.The B Slot: The 2nd show segment, typically used for more schtick.

E. The Main Interview: The top celebrity for the evening comes in and, often, does two time units with a commercial break in-between

F. The 2nd Interview: Typically a “lesser” guest spot for a single show time unit.

G. The Finish: The final segment. Typically a stand-up routine or a musical guest performing.

H. Audience Interaction: How the star interacts with the live audience.

I then graded each of the seven shows in the 8 categories and presented the following report card.

Some Quick Notes since last week:

This week, I’ve noticed that Fallon is using his announcer, Steve Higgins, as more of an active sidekick than I have ever seen. He must have seen my column suggesting his need for a stronger sidekick (OK, not likely, but an interesting development nonetheless).

I also watched some more of Craig Ferguson’s shows and I must say they are consistently great these days. His show from Monday (March 18,2013) was one of the funniest talk shows I have ever seen.

I also noticed that his show integrates the canned music better than both Leno and Kimmel integrate their live bands so I should have probably graded him a bit higher on this category.

Going Further In the Analysis:

So you have the basic report card I set up, but it is only the start of the analysis.

You can go further using the same analytical methods we use to analyze any complex data, such as environmental measurements, data used in climate modeling, data from medical or drug research, etc. These are powerful statistical methods and they can tell you some very useful things.

Below I redo the table with numbers; the grades are translated with  A=95, A+=97.5 and A-=92.5 etc. down to a lowest grade of B- = 82.5.

This yields the new table below (a matrix as we call it) that can be worked on statistically. You will note there are now two additional columns to the right and an additional three rows on the bottom that summarize the data.

The original data are highlighted in yellow. This looks like a lot of data but it is just the report card with the grades translated to numbers with a few summary columns and rows added. So relax.

The two new columns give the average score for each show based on the grades for each category, and the rank of that show, from 1 (highest) to 7 (lowest), based on that average. Both are a measure of quality, the better the number, the better the show (as compared with the Carson model).

You may think, given this, that Carson has to have the best score, but it is indeed possible for a show to have a higher average score than the icon; that they would be better at doing the Carson model than even Carson was. That isn’t the case here, but it is possible.

Back to the first new column, this is simply the average grade for each show across all the categories. It allows us to suggest what shows are the equivalent of an A student and which shows average to a different and lower grade.

So, you’ll notice, not surprisingly, that Carson’s show has the highest average grade with a 95.9; a little more than an A average. Letterman is the closest to Carson with an average grade right on 95, a pure A.

The second new column simply ranks the average scores from the first new column, you can see where each show rates from 1 (highest) to 7 (lowest). You will note that journeyman Kimmel is at the bottom (rank #7) with an average of 86.6, but he still is somewhere between a B and a B+, so it is still a very good show.

Carson and Letterman are #1 and #2, respectively. Conan is 3rd , Fallon 4th and Ferguson 5th. Leno completes the set at #6.

As I have said before, Ferguson is very difficult to grade and I personally believe he has the 3rd best show of all of these, but a lot of that just doesn’t show up in the grading.

So these new columns give you a good and simple indicator of the quality of the programs. You want quality from current shows, watch Letterman, Conan, Fallon and Ferguson.

The new rows in the table provide a very different outlook on the data. They tell you about the categories themselves.

The first of the three new rows provide the average grade for each category. This indicates just how well the shows as a group are doing in each category.

You might think that some aspects or parts of the show may just be more difficult to do than others. This would show up here.

Overall, the grades are relatively consistent, perhaps indicating that the shows tend to be of relatively uniform quality and that the shows tend to do a very good job in all aspects.

Surprisingly, the monologue, something your intuition probably tells you is really the toughest thing to do of all these categories, has one of the highest averages.

That’s not necessarily a contradiction, however, because the hosts have probably been selected primarily because of their ability to deliver a monologue over all other show components. So they have filtered out those potential hosts who could not do well here.

Even Fallon, who is tied for the lowest score in this category, is only there because they have structured his show to minimize the time allowed for a true monologue, not because he is not good when he does deliver one.

Now, just because the averages for the categories are similar does not mean they are necessarily all that similar in the original data. You can get an average by having a bunch of very similar grades around the average value, or you can get one by having higher grades and lower grades that just calculate to that average.

These two ways to get similar averages are very different, and we’ll be able to see that in the data. In order to get an indication of what trend is relevant, you have to look at the variability in the grades for each category and the bottom two new rows do just that.

The simplest way to view this variation is to calculate the range, the difference between the highest grade and the lowest one for the category. This is given for each category in the 2nd row from the bottom.

Here, the higher the number, the greater the variation. You can see that the three segments in the middle of the show – the B-slot and the interview segments – have the lowest range of 7.5. Next come the monologue and finish, with a range of 10 units.

The highest numbers come for the audience interaction, the sidekick and the music/band categories. If you read the evaluation of the report card from last week’s column, this would not surprise you; this is where there were big differences in the shows and this is where you expect to find the biggest effect on the differences in the quality rankings of the show

Leno is rated low (rank = #6) because his grades for the sidekick, the band and his audience interaction are low. All else is competitive with the higher-ranked shows.

Kimmel (rank = #7), the journeyman, has a low average for a different reason; he is very good but not excellent in almost all categories.

The range is a bit of a weak statistic, however, and most people who work with data tend to like something stronger.

The bottom row provides a better estimate of the variation in each column by providing a much stronger statistic; the standard deviation.  Again, the higher the number, the greater the variation.

You can, not unexpectedly, see it gives very similar results to that of the range. But it really stresses the importance of the sidekick and music/band numbers, and a bit less so, the audience interaction, as main indicators of the differences among the shows.

So, with this relatively simple exercise, you can now understand what shows are of the highest quality, as defined by the Carson model, and which shows are simply good to very good.

You also understand, in some way, how these shows are different ,with some de-emphasizing the band and music, or doing less with the sidekick, or being relatively consistently good to very good, but not great.

With this you can decide on what you’re looking for in a late night talk show: look at the grades and patterns, and find the one for you.

Bonus Round!

Now, when it comes to analyzing data, we are really just getting started and I will go through one more step. If you’re up for it, continue reading this section. If not, just skip to the final conclusions.

The data we are using is one for seven shows graded for eight categories. That actually means the data are eight dimensional (I can argue they are really seven dimensional but there is a limit to my esoteric nature as well).

I don’t know about you, but I can see in three dimensions, have a good grasp of the fourth (time in many circumstances), but that’s about it. So how do we make sense of eight dimensions?

Well, in my more typical work, I am often dealing with hundreds or, even, thousands of dimensions. The only way to make sense of it all is to be able to analyze the data in such a way as to get the patterns reduced back down to two or three (or four) dimensions but still have it represent a great deal of the patterns or information that exists in those many dimensions.

Luckily, we have various ways to do just this, and I’ll demonstrate one of them called an ordination. Here, I’ll use a very simple version of an ordination, one called Principal Components Analysis, that nicely works on these data.

Ordinations are how we measure a few hundred environmental parameters and extract patterns that show what environments are most similar to others. In this way we sometimes find oil or other resources.

Or it’s how we figure out what climate parameters contribute to climate change. Or what combination of factors may contribute to greater cancer rates.

The list can go on and on.

So here, I did the ordination and generated a two-dimensional map that reflects most of the information contained in our eight-dimensional report card (about 80% of it, actually). The results allow us to view where our shows are in this map, and why they are where they are.

The resulting graph is shown below – each show is where a labeled star is shown. The closer the shows, the more similar they are.

The further apart, the more different they are for some reason.

The horizontal axis is really showing the distribution of the shows based on all the grades. To the right are the better shows, to the left, the less great ones.

In actuality, we go from very good to great; there are no bad or even average shows as they would not last. You certainly can see, however, the rankings discussed above nicely illustrated here by the position of shows from the right (better) to the left (less good).

The vertical axis mostly separates Fallon from Ferguson. The former has a great band, reduced sidekick,  and a reduced monologue; Ferguson has a great monologue and sidekick, but no band.

Two very different shows and they are apart for that reason. The two are of comparable quality, however, although, as I have stated, I think Ferguson probably should be rated better and, consequently, should be much more to the right.

You can see that you have a nice cluster of Conan, Letterman and Carson, defining the best late night talk shows. Ferguson just has to ramp up his interviewing and his general schtick a bit and he would be great – he would move far to the right.

Fallon pretty much needs to ramp up his sidekick presence and make his monologue much better – mostly by expanding it – and he would move to the right as well.

Shows close to the mid line are the most typical of the Carson format. Those away from that line are the more variable in format.

I also superimposed aspects that are not measured by the data – hipness, wackiness, and being a “safe” show – onto the map, something we do all the time in analytical circles.

Hipness is mostly exemplified by four of the shows – Letterman, Carson, Fallon and, to some degree Conan (a nerdier hipness). You can see they are concentrated to the right and up on the map.

Wackiness is more of a Ferguson, Conan and Letterman aspect (and sometimes Carson) and this defines the right and lower part of the map. It’s not odd that some shows are both hip and wacky.

The left and a bit up is the refuge of the safer shows. Not particularly hip or wacky, and really not excelling in very much but good basic talk show stuff.

Oddly, this probably also is where the higher ratings will always be found as hip and wacky people have never comprised that high a percentage in our population. However, the more hip or wacky the audience, typically the more loyal they tend to be.

So the choice may be between a smaller but ardent fan base, or a less motivated but bigger one.

The map tells you what you might want to know in choosing a show. If you liked Carson, watch Letterman and/or Conan.  You probably will not go as crazy over Leno or Kimmel.

If you like hip, watch Letterman, Conan, and Fallon or Carson reruns. If you really like Wacky, then watch Ferguson and, much less so, Conan and Letterman. Letterman was more wacky early on in his history.

Want a more basic or conventional program, go for Leno and/or Kimmel.

If Kimmel ever sees this, he will go crazy that he is linked closest with Leno, and not his idol Letterman. However, even his schtick tends to converge on Leno and not Letterman; for example, with more man-on-the-street stuff.

Hey Kimmel! You want to move to the right and away from Leno towards Letterman, better get a great sidekick and work that band into the show a lot more. It would help to step up the whole show, for that matter.

Final Conclusion:

Most analyses confirm what you already suspect because you tend to see trends intuitively and just need help to elucidate what’s there.

We knew Carson was the best before doing anything. He’s the Canonical Exemplar, for gosh sakes.

Not surprisingly, and agreeing with most critics, it shows Conan and Letterman have the best shows right now and Leno and Kimmel the least exceptional. Both of the latter are very watchable but just are safe; perhaps too safe.

Ferguson is pretty much on his own and would have to change everything to get to the highest levels. Probably not worth it as he has a very loyal and significant audience and the show is cheap to produce and a real money maker.

Leno is the same way, he would have to stop being Leno to move and he has a big general audience, so why change?

Fallon is the wild card. He has a very good show but it is still in its infancy. Like Conan when he started, lots of things don’t work wonderfully yet on the show.

Many of his bits are very inconsistent and lots just don’t work well. However, most of that probably can be fixed with a larger writing staff.

But the potential is humongous for that show. If he really wants to take the show to its top – and perhaps to an historic level of quality, he needs do a few things.

He needs to strengthen the monologue – less the average quality of the jokes, which is high but could get better – but he needs to get it 3 or 4 times longer than it is now.

He also needs to find that very hip and very funny sidekick that can add the counterpoint that will take him over the top. If Justin Timberlake ever wants to quit singing and acting, he would do the job nicely; he is a good model for the desired attitude and quality of the sidekick he needs.

If he does improve things, watch out everyone, he could challenge Carson.


Editor’s note: Ralph Chapman is a paleontologist, technologist and statistician who enjoys living in Los Alamos, New Mexico – a place with great history; whether it be geological, ancient, old, or recent.


LOS ALAMOS website support locally by OviNuppi Systems