Dancing by the Numbers: Analyzing 20 Years of Dancing With the Stars

Current logo for the ABC/Disney+ dance competition show.

Introduction/Background

Since the show first sashayed its way onto American television screens in June 2005, Dancing with the Stars (DWTS) has captivated audiences with its unique blend of glitz, glamour, and breathtaking performances. Originally pitched as the American version of the British dance competition show *Strictly Come Dancing, DWTS* has since gone on to produce over 500 episodes across 33 seasons. Now, on the eve of its 20th anniversary, I thought it would be interesting to take a look back at the show’s history and see how far it has come.

But before I get into that, I’d like to take a moment to briefly explain how the show works. For those who haven’t seen or heard of it before, the show pairs famous celebrities with professional dancers and each week they compete against each other for judges’ scores and viewer votes. At the end of each week, the couple who receive the lowest combined total of judges’ scores and viewer votes is usually eliminated. The pair of dancers that make it to end, are crowned that season’s champions and are awarded the coveted Len Goodman Mirrorball Trophy.

DWTS Len Goodman Mirrorball Trophy. Source: ABCNews.go.com

In this data story, I plan to explore the most up-to-date data available to answer the following questions posed in order of complexity:

1. How many celebrities have there been on the show? How does the number of couples compare to the number of episodes per season?
2. How many professional dancers have there been? Who has been on the show the longest? Who has won the most mirror ball trophies?
3. Who was the oldest/youngest celebrity to compete on the show? Who was the oldest/youngest celebrity to win? What was the overall age distribution?
4. Does the appearance of a guest judge significantly affect celebrities’ individual scores?
5. Has the overall viewership increased or decreased over time?

Now, without further ado, let’s dive into the data.

The Data

The first major hurdle that I had to overcome when choosing to pursue this project was the fact that, as of this writing, there was no officially licensed dataset to draw from. Instead, the best I had to work with was a series of user-supplied tables on a Wikipedia article. Obviously, this means that while I have done my best to remove any factual errors or inconsistencies, many of the findings presented in this data story should be taken with a grain of salt. Also, it is worth noting here that—unless otherwise indicated—season 15 of Dancing with the Stars was excluded from this analysis, as it was the first (and only) season to feature an “all-star” cast of returning celebrities.

Besides the manual copy-and-pasting of numerous rows of data from HTML tables into something useable, like a CSV or Excel file, there (thankfully) wasn’t too much pre-processing that needed to be done. When it came to calculating the contestants’ ages, I did have to do some computations based on their date of birth and the date of the season premiere when they first appeared on the show. These dates were not always easy to find—especially the birth dates—but by cross-referencing with other sources, I was able to make an educated guess where necessary.

How many celebrities have there been on the show? How does the number of couples compare to the number of episodes per season?

For my first topic of exploration, I wanted to start off by looking at the series as a whole and seeing how many contestants there have been. Since the show began in 2005, there have been a total of 394 celebrities / contestants who have competed on Dancing with the Stars.

Interestingly, when we categorize the contestants by their main occupation or profession (Figure 1 below), we can see that over half of them have been either actors/actresses or professional athletes (roughly 31.22% and 22.84%, respectively). The remaining contestants have been either TV, radio, & social media personalities (roughly 15.99%), musical artists (roughly 15.74%), or individuals in some other occupation (roughly 14.21%).

Figure 1. DWTS Contestants Grouped by Main Occupation

The pie chart above was created in Tableau using data curated from various sources. To group the celebrities by their occupation, I manually created the groups and assigned each individual to one of them. While Tableau’s grouping function made this process somewhat easier, there were still instances where celebrities could have easily been sorted into one or more groups. For example, some celebrities were both a famous actor/actress and a singer. Thus, for the sake of simplicity, I tried to focus on using only one main profession per celebrity.

When it came to comparing the number of couples to the number of episodes per season, the figure below shows that there was a major shift in recent years. In the first few seasons of Dancing with the Stars, the number of episodes were nearly double the number of couples. This then changed following season 16, where the number of episodes dropped to 15 or less on average (roughly 11.5 per season), while the number of couples remained virtually the same.

Figure 2. Number of Couples/Contestants vs. Number of Episodes per Season

According to official reports, the change in the number of episodes was due to the introduction of a new voting format, which saw the removal of weekly results shows in favour of a more streamlined approach. This new approach combined the live judging process and revealing of results into one weekly episode and would go on to become a staple of the series. The only other major departure from this format was in season 26, where a special four-episode season (the shortest in the show’s history) featured a cast entirely comprised of current and former athletes.

The bar chart above was created using Plotly Chart Studio and features a range slider on the x-axis to allow for user interaction and custom selection. Unfortunately, due to a recent update from the company, Plotly is no longer allowing the creation of new Chart Studio Cloud users, so I was unable to properly implement an interactive chart. Perhaps in a future update, I may be able to find an alternative solution.

How many professional dancers have there been? Who has been on the show the longest? Who has won the most mirror ball trophies?

For my second topic of exploration, I wanted to examine the past winners of Dancing with the Stars. While there has never been a celebrity who has won a mirror ball trophy more than once (at least not yet), there have been multiple professional dancers who have won two or three times. To count the number of seasons each professional dancer competed on the show, and the number of times they won alongside their celebrity partner, I created two formulas using the COUNTA and COUNTIF functions in Apple Numbers. The resulting table is shown in the screenshot below.

Figure 3. Screenshot of DWTS Professional Dancers Summary Table

From the results shown in the screenshot above, we can see that of the 51 professional dancers who have been on the show, only 19 have won at least once. Of those who have won, Derek Hough is the only professional dancer to have won a mirror ball trophy six times. In terms of who has been on the show the longest, Tony Dovolani had the most consecutive seasons (21), while Cheryl Burke had the highest total number of seasons (25).

Who was the oldest/youngest celebrity to compete on the show? Who was the oldest/youngest celebrity to win? What was the overall age distribution?

As mentioned in the intro of this data story, for my third topic of exploration, the calculation of contestants’ ages ended up being the second major hurdle that I had to overcome in pursuing this project. Luckily, by doing some research online, I was able to come up with a solution. Without getting too deep into the weeds, the solution mainly came in the form of the DATEDIF function. Despite its known issues, which I won’t get into here, the function was able to produce fairly accurate results. Any errors or inconsistencies that did come up, I was thankfully able to mitigate or remove by way of cross-referencing with other sources.

Once I was able to calculate the age of each contestant, based on their date of birth and the date the season aired when they first appeared on the show, I was then able to connect the data to Tableau and generate the following results.

Figure 4. Age pyramid of DWTS Contestants (Males vs. Females)

Based on the age pyramid chart above, we can see that the range of values for males and females was virtually the same (i.e., roughly 15 to 80+ years old). For females, the oldest contestant to appear on the show was comedian/actress Cloris Leachman (82 years old, season 7), while the youngest contestant to appear on the show was actress Willow Shields (14 years old, season 20). For males, the oldest contestant to appear on the show was former astronaut and engineer Buzz Aldrin (80 years old, season 10), while the youngest contestant to appear on the show was internet personality Hayes Grier (15 years old, season 21).

Figure 5. Box-and-whisker Plots for Age Distribution of DWTS Contestants by Season

Overall, the average age of contestants was approximately 38.9 years old, as indicated by the dashed line in the figure above. For females, this was slightly younger at approximately 37.8 years old, while for males, the average age was slightly older at 40.1 years old. When comparing the ages of past winners, the oldest contestant to win was singer and actor Donny Osmond (51 years old, season 9), while the youngest contestant to win was artistic gymnast Laurie Hernandez (16 years old, season 23).

Does the appearance of a guest judge significantly affect celebrities’ individual scores?

From the very first episode of Dancing with the Stars, the show has always had a panel of (at least) three judges who share their ballroom expertise with viewers and evaluate the celebrities’ based on their dance performance and technique. After each dance, the judges award each couple a score out of 30. A score of 30 out of 30 is considered a perfect score.

Beginning in the third week of season 18, however, a celebrity guest judge was invited to join the judges panel and (for the first time in the show’s history) couples could potentially receive a score of 40 (out of 40) for their dances. This has continued in nearly every season since, with various guest judges appearing at different times throughout each season.

For my fourth topic of exploration, I wanted to explore how significantly the addition of a guest judge affected celebrities’ total individual scores. I considered this topic to be one of the more complex, because to answer the question thoroughly, I felt that I needed to conduct some form of regression analysis. Unfortunately, due to time constraints (and the limitations of the data) I wasn’t able to properly conduct this kind of analysis, but perhaps I will revisit it sometime in the future. For now, I simply chose to pick one season from before and after the introduction of guest judges and compare the results using Tableau.

Figure 6. Total Individual Scores by Celebrity (Season 2, DWTS)

In the first stepped line chart (Figure 6 above), we can see that the total individual scores for each celebrity never went above 30. This is because, in season 2, there were never any guest judges. Jumping ahead in time to season 21, the second stepped line chart (Figure 7 below) shows that in weeks 3, 5, and 6, some of the scores were clearly above 30. As you may have guessed, these bumps in individual scores corresponded directly with the appearance of a guest judge.

Figure 7. Total Individual Scores by Celebrity (Season 21, DWTS)

Thus, anecdotally, the appearance of a guest judge does appear to affect the celebrities’ total individual scores. To what extent, though, I can’t definitively say without further analysis.

Has the overall viewership increased or decreased over time?

For my fifth and final topic of exploration, I wanted to study the overall viewership of the past 33 seasons of Dancing with the Stars. When choosing to write this post, I considered this to be the most complicated topic to explore, as the reporting of viewership numbers can often be quite difficult to measure and notoriously unreliable. Depending on who you ask or what sources you use, the results can vary greatly. Some studios and networks don’t even like to share their numbers publicly, which can create further distrust amongst fans and critics alike. Despite all of this, I believe I was able to find a reasonable set of data from which to run my analysis.

Figure 8. Average U.S. Prime-time Viewers per DWTS Season (Premiere vs. Finale)

According to the bar chart above, the average viewers per season has been on a steady decline. At its peak, Dancing with Stars saw its highest average number of viewers during the season finales of season 2 and season 3 at roughly 27.2 and 27.5 million, respectively. In season 32, the show saw its lowest average number of viewers with only roughly 4.78 and 5.5 million viewers tuning in to watch the season’s premiere and finale.

While this may look depressing, it is worth noting here, that these numbers only reflect those viewers watching on a prime-time TV network. In April 2022, it was announced that Dancing with the Stars would be relocating from its regular spot on ABC to Disney+ for season 31. Following that, the series was streamed live, both on ABC and Disney+, for season 32 and was then officially renewed by ABC for season 33. So, perhaps with this new setup there’s hope that the show’s numbers can make a comeback. Until then, I guess only time will tell.

If you liked this post, consider giving it a like and sharing it with your friends and family. What was your favourite DWTS moment from the past 20 years? Let me know in the comments below. And as always, thanks for reading.