Sunday, April 14, 2013


So I recently wanted to analyze the trends between how far Idol contestants made it relative to their singing ability and song freshness/creativity factor. I'll explain more in detail later, but for now, toy around with the complete set of data I made, exclude several seasons or placements, and see how the trends hold up. I'll elucidate in more detail later. The higher the placement value, song creativity factor, and the singing ability, the better.

Scroll down to play with the graphs.

What are some of the key realizations?

Part I: Contestants, In Specific; The Quadrants: Creativity Vs Singing

If you look at my last chart, I've organized the data into four quadrants--the top right illustrates contestants with higher creativity/greater singing ability, the top left illustrates contestants with higher creativity/less singing ability, the bottom right illustrates contestants with less creativity/greater singing ability, and the bottom left illustrates contestants with less creativity/less singing ability.Top right=best, bottom left=worst, essentially. We can isolate this graph by excluding the data based on seasons.

  • The bottom left are typically fodder contestants or contestants who know they are jokes, but nonetheless every year one of these contestants gets farther in the competition than their talent level would suggest. No winners out of this crew. (ie Scott Savol, Lil Rounds, Jasmine Trias, Kevin Covais, Aaron Kelly)
  • The bottom right is home of the shock elimination: contestants who can sing the playbook, usually, but choose the same old tired songs done on the Idol stage tons of times, or sing super old songs no one can relate to. One winner out of this crew--AI5's Taylor Hicks.  (ie Hollie Cavanagh, Pia Toscano, Joshua Ledet, Kree Harrison, LaKisha Jones) 
  • The top left is interesting: contestants who can't sing, but are higher level fodder contestants in that they help the mid-card go more seamlessly through introducing creative songs. 2 winners in recent years have been from here though--Scotty McCreery and Phil Phillips. (ie Jason Castro, Chris Richardson, Andrew Garcia, Kristy Lee Cook, Amanda Overmyer) 
  • The top right is where the TCOs are usually found in: they're creative and can sing well enough by WNTS standards. 8 of the 11 winners are from this crew. (ie Fantasia, Carrie Underwood, David Archuleta, Anoop Desai, Colton Dixon) 
Season By Season Analysis:

As mentioned, bottom left contestants typically play spoiler, and bottom right are home of the shocks.
AI3: TCO types (George Huff, Amy Adams, Diana DeGarmo, Jennifer Hudson, Fantasia, LaToya London); Top left (Matt Rogers, Jon Peter Lewis); Bottom right (none); Bottom left (Camile Velasco, John Stevens, Jasmine Trias, Leah LaBelle). As mentioned, John Stevens and Jasmine Trias really played spoiler and since there were so many TCOs, Amy Adams, LaToya London, Jennifer Hudson, George Huff and Amy Adams all became shocks. At the end, only the TCO types were at the top 2.

AI4: TCO types (Nikko Smith, Carrie Underwood, Nadia Turner, Bo Bice); Top left (Anthony Fedorov, Constantine Maroulis); Bottom right (Jessica Sierra, Anwar Robinson, Vonzell Solomon); Bottom left (Lindsay Cardinale, Scott Savol, Mikalah Gordon). Sierra and Robinson were shock eliminations at the expense of Scott Savol who advanced to the top five. Also, Nikko Smith and Nadia Turner were shocks from the TCO category. At the end, only the TCO types were at the top 2.

AI5: TCO types (Chris Daughtry, Mandisa, Paris Bennett); Top left (Lisa Tucker, Bucky Covington, Ace Young); Bottom right (Katharine McPhee, Taylor Hicks, Elliott Yamin); Bottom left (Kellie Pickler, Melissa McGhee). This was the year where great singing with zero creativity reigned supreme. The winner came from the bottom right, as did the runner up and third place winner--Hicks, McPhee and Yamin in that order. TCO types Daughtry and Paris Bennett were fourth and fifth as semi shocks, and Mandisa was a shock elimination while Kellie Pickler at the bottom left played spoiler.

AI6: TCO types (Blake Lewis, Chris Sligh, Stephanie Edwards, Jordin Sparks); Top left (Phil Stacey, Chris Richardson, Haley Scarnato); Bottom right (Gina Glocksen, Melinda Doolittle, LaKisha Jones); Bottom left (Sanjaya Malakar, Brandon Rogers). The season reverted back to AI3-AI4ish in the sense that two TCOs made the top two, and Sanjaya Malakar played spoiler by shock eliminating Stephanie Edwards and semi-shock eliminating Gina Glocksen. Chris Sligh committed Idol suicide by tanking even though he had the markers of TCO.

AI7: TCO types (David Cook, Michael Johns); Top left (Kristy Lee Cook, Amanda Overmyer, Jason Castro, Chikezie, Brooke White); Bottom right (Carly Smithson, David Hernandez, David Archuleta, Syesha Mercado); Bottom left (Ramiele Malubay). This was an interesting season with no one who really played spoiler--this was a very creative season, so many of the poor singers still proved themselves useful by introducing a ton of new songs. Michael Johns was a shock elimination, but everyone else was more or less useful. The winner, Cook, was a TCO type, and Archuleta came from the bottom right of good singing with poor creativity, as did 3rd place winner Syesha Mercado.

AI8: TCO types (Kris Allen, Adam Lambert, Anoop Desai); Top left (Scott MacIntyre, Jasmine Murray, Matt Giraud); Bottom right (Alexis Grace, Allison Iraheta); Bottom left (Megan Joy, Danny Gokey, Lil Rounds, Michael Sarver, Jorge Nunez ). AI8 was kind of a weird season in that people's perceptions were different than what the stats suggested. This was arguably the first season where sob story played a major role, particularly with Danny Gokey, who was a poor singer this season on top of being a completely un-creative song chooser. Unlike Scott Savol and Jasmine Trias, other bottom left types, he wasn't perceived as a poor singer who was advancing too far, even though he played spoiler to Alexis Grace and Anoop Desai, and semi-shocking Allison Iraheta as well. On the flip side, Kris Allen was perceived as fodder even though he compiled TCO type stats--like Desai, he was both creative and could sing well by WNTS standards, and at the end he was the winner. Again, the top two were both TCO types, even though this was largely an uncreative season as a w hole.

AI9: TCO types (Lee DeWyze, Casey James); Top left (Andrew Garcia, Lacey Brown); Bottom right (Didi Benami, Michael Lynche, Siobhan Magnus, Crystal Bowersox); Bottom left (Katie Stevens, Aaron Kelly, Tim Urban, Paige Miles). This was a poor singing season with average creativity, but DeWyze was the winner and a TCO type, while runner up was a good singer with somewhat poor creativity in Bowersox. Tim Urban and Aaron Kelly played spoilers in kicking out Didi Benami and to a slighter extent Siobhan Magnus.

AI10: TCO types (Casey Abrams, James Durbin, Lauren Alaina, Haley Reinhart); Top left (Scotty McCreery, Paul McDonald); Bottom right (Pia Toscano); Bottom left (Stefano Langone, Thia Megia, Jacob Lusk, Naima Adedapo, Ashthon Jones, Karen Rodriguez). This season sorely lacked creativity. Generally, however, no one in the bottom left really played spoiler to any of the TCO types, and after the save was used on TCO type Casey Abrams, he outstayed everyone at the bottom left. Pia Toscano, however, was proof that the bottom right is reserved for shock eliminations. The real surprise was that a top left type won, although to be fair, Scotty was like Phil Stacey--kind of borderline in singing ability between the top left and right, but slightly more to the top left. So at the end, he becomes our first top left winner, and the first non-TCO type winner since Taylor Hicks (bottom right). Lauren Alaina, a TCO type, was runner up.

AI11: TCO types (Colton Dixon, Skylar Laine, Elise Testone); Top left (Jeremy Rosado, Phillip Phillips, DeAndre Brackensick); Bottom right (Hollie Cavanagh, Jessica Sanchez, Joshua Ledet, Erika Van Pelt); Bottom left (Shannon Magrane, Heejun Han, Jermaine Jones). OK, so we're starting to notice a trend here. Like Scotty, Phillip is a top left type who won the show--not the greatest singer within his season, but emphasizing that as with AI10, it's creativity that the audience is looking for. Notice that the last un-creative good singing guy to win was in AI5. The runner up was a bottom right, as was third and fourth place, as they outlasted all three TCOs, so the top five was stocked with old fashioned good singers this season. The only real shock was with bottom left Heejun outlasting Erika Van Pelt, but not much else.

AI12: TCO types (Angie Miller, Devin Velez); Top left (Burnell Taylor, Janelle Arthur); Bottom right (Kree Harrison, Amber Holcomb, Candice Glover); Bottom left (Paul Jolley, Curtis Finch Jr, Lazaro Arbos). The most un-creative season ever, even though the season has been above the norm, believe it or not. Angie and Devin, in the AI aggregate, are typically bottom right contestants, although Angie might be a bit borderline, but this season is so old-fashioned/overdone that it makes them look super current by comparison. This season is somewhat of a continuation of AI11 in that the top four consists of the girls from the bottom right. But considering a bottom right hasn't won since AI5, it really looks like it's Angie's to lose. In terms of shocks, bottom left Lazaro Arbos outstayed his welcome by kicking out TCO markers type Devin Velez.

What Quadrants Are The Contestants Who Sell Found In?

Consulting my other article for this, we find the major youtube views to be:

Kelly Clarkson (217mil)
Carrie Underwood (151.3mil)
Jordin Sparks (129.1mil)
Daughtry (92.8mil)
Adam Lambert (82.3mil)
David Archuleta (53.4mil)
Jennifer Hudson (41.6mil)
Kellie Pickler (32.2mil)
Fantasia (22mil)
David Cook (19.4mil)
Elliott Yamin (16.9mil)
Scotty McCreery (15.8mil)
Katharine McPhee (13.2mil)
Chris Richardson* (10.7mil)

Since we can't rate Kelly, but we can rate the rest, let's do that: Carrie, Jordin, Daughtry, Adam Lambert, Jennifer Hudson, Fantasia, and David Cook were all TCO types during their season. David Archuleta, Elliott Yamin and Katharine McPhee are the bottom right representatives, and Kellie Pickler, Scotty McCreery, and Chris Richardson are the top left representatives. To be fair, in the whole AI aggregate, Archuleta, Yamin and McCreery actually rank as TCO types, so it might be somewhat unfair to lump them in their original categories. By and large, top left can only be successful off the show if you're in the country genre, and bottom right can only be somewhat successful if as a whole, you rank as a TCO candidate.

Among AI12's crop, the most TCO-like by far is Angie Miller, but even she doesn't rank as a TCO candidate in the aggregate, because the season is too old fashioned (but she's close). Still, out of this crop, she might have the greatest chance to have slight commercial success. Candice, Amber, and Kree are too old fashioned, and considering none are even close to TCO like in the aggregate, there's no precedent for any selling ability here.

Part II: Contestants, In General; Based on Race, Gender, and Region, with A Dash of the Audience and TPTB

So let's zoom out our scope a little bit and focus not on the specific contestants and whether their brand of creativity or singing ability is conducive to winning or selling, which is what we did in part I; instead, let's focus on the audience, and the types of contestants they prefer, based on race, gender, and region. In doing so, we'll also find out the sort of contestants TPTB like to promote, and whether there's a correlation between the number of contestants of a certain subset and how far they advance in the competition.

So let's find out the audience first: is an incredibly useful resource for finding out the sort of audience  Idol attracts. Going to key sites reveals that Idol's audience is heavily female dominated, and in particular attracting many 45 year old types and older. This jibes with the idea that the average age of the Idol viewing audience is in their mid-40s. There are some 18-24 year old teens voting, but using the aggregate of the two sites suggests a heavy dose of the 45-54s females at work here.

Singing, and Creativity

Before we talk about who the 45-54 females are voting for, let's talk about the trends of the contestants. Keep in mind because there are 121 contestants, there are about 121 values, so the average singing ability is about 60.5. In terms of pure WNTS singing ability, Idol's best singers tend to be black girls. Generally, black girls of any age, because out of 121 contestants, with 121st being the best, they rate about 75-79. White girls up to age 23 are also above average, rating about 67, and white guys ages 20 and beyond also tend to be slightly above average with a dash over 60. Black guys ages 20-23 are also a dash above average. So largely, any black girls and white girls up to age 23 are the lifeblood of great singing. Notice it's mostly girls up top. Also, non blacks/non whites of any age tend to be subpar singers, as well over 24+ black guys and 0-19 aged white guys.

Now let's focus on creativity. As usual, the average is 60.5. It's a very small sample of only David Hernandez and Andrew Garcia, but aged 24+ male "others" (non blacks/non whites) are the most creative contestants. White guys and white girls aged 24 and older are very creative at about 82 as well, so we can already see that a lot of the older contestants are the more creative. Black guys ages 20-23 are also very effective, and black girls up to age 23 are also up there (although the younger ones are more creative). Aged 20-23 white guys and girls are a dash above average in creativity as well. The less creative ones include white girls up to aged 19, black girls between ages 20-23, black guys between 0-19 and 24+ in age, all non black/non white girls, and non black/non white guys between ages 20-23. The white and "other" girls are affectionately called pageantbots in the Idolsphere, and the black guys mentioned are considered old soul singers, more or less.

We can also separate this by region, also shown in another graph below. Again, average singing is 60.5. Interestingly, the best female singers tend to be from the northeast, midwest, and then the south. And then it's the guys, where only southern guys are a dash above average in singing--all other guys are below the average. Western girls are also are inferior to their girl counterparts in singing.

Let's look at the other side of the coin: creativity, average 60.5. Southern guys are the most creative, but midwestern guys and western guys are also up there. For the girls, only the midwestern girls are very creative. Also other girls are subpar, but in particular northeastern girls. Only the northeastern guys are subpar in creativity.

So what does this tell us? Midwestern girls have the best combination of singing and creativity, ranking 2nd and 2nd (both out of 8) in singing and creativity within their region/gender subset. Southern guys are next ranking 4th and 1st (out of 8) respectively, then it's a three-way tie between northeastern girls, southern girls and midwestern guys (of which the midwestern guys are the most creative, northeastern the least). Western guys and girls, and Northeastern guys are at the bottom of the combination pack.

Also, black girls ages 0-19 (on the basis of their singing) and white guys aged 24+ (on the basis of their creativity) have the best combination of singing and creativity. Black and white girls aged 24 and older also rank highly, as do black and white guys and girls aged 20-23. As mentioned, many of the non blacks/non whites rank poorly, as do black guys that are not between 20-23 and white guys up to age 19.

What does this mean for TPTB, and how do they cast their contestants?

So what does this mean for TPTB, for the optimal combination scale? Perhaps this mean they should cast more black girls up to age 19 from the Midwest, or age 24 and over white guys from the South. These hit all the major markers. TPTB probably see Western and Northeastern non blacks/whites, teenage whites and very young/very old black guys are fodder, because they just struggle to produce in terms of singing and creativity. All others are pretty competitive.

But instead, if you look at how TPTB has cast, from AI3 to AI12, they have cast 20 24+ years and older white guys, which as mentioned above is good--they are very competent, and probably the second best subset Idol has to offer, particularly with their creativity. Great move.

They have also stacked the deck with white girls, between ages 20-23 but also those between ages 0-19. Now, these are pretty average to decent in their singing/creativity combinations overall, but they tend to be better singers than creative types. White guys between ages 20-23 are also a common appearance, and again they're competent at both ends, as are over 24+ black and white girls, which tend to be very competent. Decent moves.

But given their #1 ranking in singing/creativity, 0-19 aged black girls are actually underrepresented. They are only the seventh most common archetype in the Idolsphere, with only 7 out of the 121 contestants filling that role. Black guys aged 20-23 have been decent, ranking 4th in their combinations but 13th in commonality (only 4 of their 121 contestants) so they are also underrepresented. In general, it looks like TPTB are trying to make younger up to aged 23 white girls happen, because they're very common despite middling numbers, at the expense of more competent black singers. Poor moves.

Also, TPTB likes to throw in 0-19 aged non black/non white girls and 20-23 non black/non white guys for diversity purposes, even though in general they have the worst combinations of singing and creativity.

Let's look at the region for a clearer picture here. TPTB sure love their southerners--southern girls and southern guys make up nearly half of the finalists (30 and 27 each respectively). Western girls and guys are next, then the midwesterns, and finally the Northeasterns are the least represented.

Good moves--casting Southern guys, as they rate second highest in the combinations. Decent moves--casting southern girls, they're decent, but since they're cast the most they're at the expense of... Poor moves--the Midwestern girls. They have the highest combination of singing and creativity, yet they are severely underrepresented. Even more underrepresented than the Western girl and guy, who are cast more yet have way worse combination parameters.

So if we put two and two together, TPTB have been overcasting Southern white girls up to age 23, at the expense of Midwestern girls and 0-19 aged black girls. By and large 24+ year old southern white guys are cast a ton but they pull their weight. Also, TPTB love to cast the Western non black/white contestant despite poor combinations, likely for diversity purposes. Northeastern types have poor combinations and TPTB know that as they are rarely represented.

Based on casting commonalities for these subsets, if we're to extrapolate this to a 12-finalist season, roughly 2 of them would be 24+ year old white guys, possibly 2 20-23 year old white girl(s), 1 20-23 white guy, 1 0-19 year old white girl, a 24+ black girl, a 24+ white girl, a 0-19 aged black girl, a 0-19 other girl, a 20-23 other guy, and and a 20-23 year old black girl. In other words, we'd have 4 24+ year olds, 5 20-23 year olds, and 3 0-19 year olds, with 4 white girls, 3 white guys, 3 black girls, 1 other girl, and 1 other guy. Of those, 3 Southern guys, 3 Southern girls, 2 Western girls, 2 Western guys, 1 Midwestern girl and 1 Northeastern guy are the common attributes. We don't even see a black guy in a season extrapolating backwards (although the non 20-23 year olds are generally incompetent), and only 1 Midwestern girl, and this adds to the idea that TPTB are trying to make Southern white girls and Western non black/white guys/girls happen with their casting. does the audience vote, given this TPTB framework?

Excellent question. One of the things that us about jork vs tptb have been suspecting ever since the WGWG era is that the audience loves creativity more than actual singing. It's the reason we've introduced that creativity marker in all of our graphs and tables below, where that creativity is based on song age/song freshness within the Idolsphere. Pia Toscano ranked incredibly low in creativity, and we know what happened to her. And if you look at the regional graph in particular, this is virtually a direct correlation between creativity and placement: Southern guys are the most creative, then the Midwesterns, then the Westerns, then Southern girls, then the Northeasterns--and the placement order tends to play out the same. This pretty much proves that the audience is looking for creativity rather than singing, in a vacuum.

The problem is the casting. As mentioned, Southern guys are the second most commonly cast by TPTB, meaning that there is more of them so they have a greater likelihood of winning. In fact, if you've noticed, many of the guy winners (Taylor Hicks, Kris Allen, Scotty McCreery, Phillip Phillips) are from the South. As mentioned, the midwestern girl is one of the least commonly cast archetypes (6th out of 8 types) so even though the audience is very likely to vote for them, because there are so few of them that decreases the chance of them winning. Midwestern guys are so rarely cast as well, but they get votes, as explained below. Southern girls are cast heavily but they are not a favorite of the audience, which will also be explained below--which is why they never really win at the end. Northeastern types are essentially kicked out early in the show and as mentioned are not creative in general.  

This is where I created the "audience rating" algorithm which is the very first column of both tables: this is a number that accounts for the archetype's singing ability, creativity, and placement, and spits out a number that shows how favorable that sort of contestant is to the audience. The higher the number, the more favorable they are to the audience. For example, Midwestern guys don't really have a good combination, in particular lacking in WNTS singing ability, but they have the highest audience rating because the audience seems to ignore that and still elevate them to a high ranking. Southern guys, Midwestern girls and Western girls seem to get the same love from the audience.

If we look at the gender/age chart, it's harder to see those direct creativity/placement correlations, but we can see there really isn't a direct correlation. In fact, white girls over age 24 and black guys between the ages of 20-23 tend to be very creative within the Idolsphere, with marks in the 80s, yet they have audience ratings in the 9s and 10s and get voted out very fast. In fact, we know that the Idol audience, as evidenced by the WGWG era of AI7 and beyond, really, really favor white guys--up to age 23, they have super high audience ratings, and the audience also favors teenage black and non black/white girls as well. So with the gender/age chart, there might be a race component as well, given that high creativities within older white females and middle-aged blacks aren't rewarded, while white guys are--in fact, the younger white guys tend to be incredibly incompetent in WNTS singing ability but their audience rating is so high that they tend to have very high placings--they tend to be "the overachievers" that the audience wonders "why are they still here?". Or we can call it the Bieber effect, obviously. Yes, it really exists on Idol.

From the chart, any non white male over 24 year old really, really struggles to get votes, as do any black guy over the age of 20, or any non-teenage female "other" contestant. The contestants that get kicked out super easily fit these profiles and are Northeastern.

The problem again resides in casting. Just like how Southern guys are cast heavily and get votes leading to wins, white guys aged 20 and beyond are cast super heavily, and these types have high audience rating factors and have the 1st and 3rd highest combinations of singing/creativity--but all these factors together, you have an easy win for many of them. If they're Southern guys and are white and aged 20 and above, that's an incredibly lethal recipe for winning. As we keep mentioning TPTB keep trying to make younger and middle aged white girls happen, but they don't really have a high audience factor and as a result don't really win. Young black girls are very competent but incredibly under-cast, as they do get votes and many have the potential to win. With TPTB's under-casting though, they rarely get the opportunity. Teenage "other" girls have high audience ratings but woeful combinations so they don't get the chance to win, not to mention TPTB doesn't cast them much. TPTB cast a decent amount of over 24+ black and white girls, but they don't get votes and are always perceived as fodder. Overall, the end result is a white guy winning from all of this.

So...Let's Talk About How TPTB Have Reacted This Season

It's essentially like TPTB have looked at all the jork vs TPTB graphs and saw no way around it...particularly looking at the audience ratings numbers in my graphs. For many years they have been casting the same way, allowing a bunch of Southern girls and Southern white males into the competition and for many years see them advance far while others fall by the wayside. Now, for the top 10, they've engineered a process where the only sole white male (Paul Jolley) had such horrible creativity and singing numbers that even though he fit within the major inset of getting votes (Southern white male aged 20-23) that wasn't enough to help him, and he ended his Idol run early. Burnell was a black 0-19 guy, but historically they struggle to place well on the show. Devin Velez was a 0-19 "other" non black/non white and those also don't do well, as with Lazaro Arbos's incredibly poor numbers in the 20-23 "other" category, and CJ Finch fell within the black 24+ subset which almost never gets any votes and finishes nearly last all the time. In a nutshell, the guys had no precedent for finishing well on the show, and of course they didn't.

That made it a cakewalk for the girls provided they hit the markers. Amber Holcomb severely lacks creativity for her type but is a far better singer than the typical 0-19 black girl, but as a Southern girl that's not that creative she's struggling to get votes (as we noted there's a direct correlation between region and placement), even though 0-19 black girls tend to get votes on the show. Candice Glover is a 20-23 black Southern girl who's like Holcomb--can sing, little creativity, and she's fighting against the Southern girl factor not getting votes, and 20-23 year old black girls historically get fewer votes than 0-19 year old types. Angie's breaking a ton of barriers about 0-19 year old white girls not getting votes as well as being from the Northeast, but she's by far the most creative and that could lead to wins. Kree's a 20-23 year old Southern white girl, and both the age and the region are lukewarm for getting votes in this competition, and she's also old fashioned, but she can sing. But as we know, singing isn't a real correlation for getting votes.

Of the four, only Amber has a real positive marker for winning, and possibly Angie, but all of them have offsetting markers from past seasons such that if they had cast any competent guy with voting markers, they could have been ousted. It was more wrecking the guys' side than really stacking the girls' side with voting markers. It was blatant, but probably had to be done as the graphs showed below.




No comments:

Post a Comment