Popularity and Audience Appeal of The Office

Introduction:

The Office is one of the most popular and recognizable TV (television) shows of its time. However, the success of a TV show can be measured and assessed in different ways. If the episode attracts a lot of attention, retains the audience thought time, and receives positive feedback or ratings, does not necessarily mean the same type of TV show success.

This report examines the episode-level data from all nine seasons The Office TV show, with a purpose of investigating the true popularity of the show, its appeal, and viewer engagement. In this report, the viewers are the primary derivatives of popularity, measured in millions who watched each episode during the original airing of the show. Audience appeal will be measured using the imdb_rating, with the average rating given by the IMDb users. Lastly, the total_votes will measure the total number of IMDb ratings that were submitted for each episode and will be used as the measure of the viewer engagement.

The main question guiding this report is: How did the popularity and the audience appeal of The Office change throughout the series and how can unusual (outlying) episodes help explain the relationship between the TV viewership, long-term engagement, and audience ratings?

Throughout this report, we will use the tidyverse package to create visualizations and the DT package to display the data set.

library(tidyverse)
library(DT)

The Data:

The data used throughout this report is called ‘office_ratings’. It contains information about each individual episode of TV show The Office across all nine seasons. Each episode is an observation, and the data contains 186 observations and 7 variables. The variables are described as follows:

Variable	Variable Type	Description
season	Categorical	Season during which the episode aired
episode	Categorical	Episode number within the season
title	Categorical	Title of the episode
viewers	Continuous	Number of viewers in millions on the original air date
imdb_rating	Continuous	Average IMDb rating from 1 to 10
total_votes	Continuous	Number of IMDb ratings submitted
air_date	Date	Date the episode originally aired

Distributions of the Main Variables:

Before analyzing the relationship between popularity, audience appeal, and the viewer engagement, it is important to understand how each continuous variable in the dataset are distributed, which include viewers, imdb_rating, and total_votes.

Each shape of distribution will help identify where most episodes fall, whether the distributions are skewed, and whether any outliers may be present.

Distribution of Viewership

ggplot(data = office_ratings)+ 
  geom_histogram( mapping = aes(x = viewers), fill = "steelblue", binwidth = 1, color = "white" )+ 
  labs( title = "Distribution of Original Television Viewership", 
      x = "Viewers (millions)", 
      y = "Number of Episodes"
  )

The overall distribution of the viewership is mostly concentrated between 4 and 10 million viewers with the highest concentration of episodes at around 7 to 9 million viewers. The distribution is also right skewed with most episodes being grouped within this range, and with a small number of episodes extending towards much higher viewership. There is also an evident extreme observation with approximately 23 million viewers. This observation suggests that that episode attracted much more viewership compared to the typical The Office episode. Since this observation is different from normal, it will be investigated later in this project.

Distribution of IMDb ratings

ggplot(data = office_ratings)+ 
  geom_histogram( mapping = aes(x = imdb_rating),fill = "steelblue", binwidth = 0.25, color = "white" )+ 
  labs( title = "Distribution of IMDb Ratings", 
      x = "IMDb Rating", 
      y = "Number of Episodes"
  )

The overall distribution of IMDb rating seems somewhat symmetric, however there are small number of episodes that received noticeably lower ratings. Most episodes received ratings ranging from 7.8 and 8.9 with the highest concentration at around 8.1 to 8.9. This suggests that viewers generally rated The Office episodes positively and most of the episodes are valued at a fairly consistent range.

Distribution of IMDb Votes

ggplot(data = office_ratings)+ 
  geom_histogram( mapping = aes(x = total_votes),fill = "steelblue", binwidth = 250, color = "white" )+ 
  labs( title = "Distribution of IMDb Votes", 
      x = "Number of IMDb Votes", 
      y = "Number of Episodes"
  )

The overall distribution of IMDb votes is strongly right-skewed with most episodes receiving approximately 1,300 and 2,800 IMDb votes. The highest concentration seems to be around 1,400 and 2,200 votes, which in my opinion is a lot of votes. The long right tail shows that there are a few episodes that attracted much more online votes and viewer participation. My guess is that these outlying episodes gave memorable or important content which triggered much larger number of ratings compared to normal The Office episodes.

Relationship between Viewership and Ratings

ggplot(data = office_ratings) +
  geom_point(mapping = aes(x = viewers, y = imdb_rating))+
  geom_smooth(mapping = aes(x = viewers, y = imdb_rating), method = "lm", se = FALSE)+
  labs(
    title = "IMDb Rating vs. Viewership",
    x = "Viewers (millions)",
    y = "IMDb Rating"
  )

The overall relationship between viewership and IMDb ratings is positive. Episodes with larger audiences tend to have somewhat higher IMDb ratings. However, I believe the relationship is not strong because of the overall spread of points on the plot. There are many episodes with smaller viewership that received noticeably different ratings, and there are several highly rated episodes that had smaller or moderate viewership. There is also a noticeable extreme where the episode had more than 22 million viewers. This episode alone is strongly affecting the observation making it more positive. There is also another unusual episode that had 11 million viewers with a lower IMDb rating below 8. This episode suggests that generally the audience size alone does not determine how liked the episode is.

Viewership vs. IMDb Votes

After comparing the viewership with audience ratings, the next question is whether episodes with larger audiences also generated more IMDB engagement. If more people watched the episode, can we expect larger number of IMDb votes

ggplot(data = office_ratings) +
  geom_point(mapping = aes(x = viewers, y = total_votes))+
  geom_smooth(mapping = aes(x = viewers,y = total_votes),method = "lm",se = FALSE)+
  labs(
    title = "IMDb Votes vs. Viewership",
    x = "Viewers (millions)",
    y = "Number of IMDb Votes"
  )

The scatter plot shows a clear positive relationship between viewership and number of IMDb votes. This concludes that episodes with higher viewership tend to receive more IMDb ratings. However, the relationship is not perfect, again. Most of the episodes are in 4 and 10 million viewers range at about 1,500 and 3,000 IMDb votes. There are several episodes that stand far beyond that range. One episode received 6 million viewers and received almost 8,000 IMDb votes, while another episode received 8 million viewers and received close to 5,800 votes. These episodes definitely generated more long-term engagement than original audiences would suggest. Lastly, we have an episode with approximately 23 million viewers which received large number of IMDb votes. This episode follows the overall positive trend of the scatter plot, but it’s interesting since it’s located far away from any other episode. My conclusion is that this episode must have been very memorable to viewers.

Popularity Change Over Time

ggplot(data = office_ratings) +
geom_line(mapping = aes( x = air_date, y = viewers))+
  geom_point(mapping = aes(x = air_date, y = viewers))+
  geom_smooth(mapping = aes(x = air_date,y = viewers),se = FALSE)+
  labs(
    title = "Viewership of The Office Over Time",
    x = "Air Date",
    y = "Viewers (millions)"
  )

The graph shows that the popularity of the show changed noticeably over time. Viewership increased during the early years of the show and remained high during the middle of the show’s running. After reaching its strongest period, the overall trend began to steadily decline with later episodes in the show having smaller audience. By the final years of the series, many episodes fell to under 5 million viewers compared to more popular period of the show with approximately 8 to 10 million viewers. The most noticeable episode that had more than 22 million viewers happened in 2009, which stands above any other episode. This value alone is unusual which probably resulted from special circumstances, since the show did not follow that spike. Lastly, the last episode of the show has a noticeable increase in viewership compared to last year of the show. My conclusion is that the show grew in its early years and reached its peak during the middle part of its airing and then experienced gradual decline in viewership.

Audience Appeal Over Time

ggplot(data = office_ratings) +
  geom_line(mapping = aes(x = air_date,y = imdb_rating))+
  geom_point(mapping = aes(x = air_date,y = imdb_rating))+
  geom_smooth(mapping = aes(x = air_date,y = imdb_rating),se = FALSE)+
  labs(
    title = "IMDb Ratings of The Office Over Time",
    x = "Air Date",
    y = "IMDb Rating",
  )

The IMDb ratings show that the audience appeal changed throughout the series and compared the change in viewership it’s much different. Ratings increased during the first years of the series with the overall trend reaching the highest point around 2007 and 2008. After this period, audience ratings gradually declined with some of the last episodes of the show having ratings below 8. Despite this decline, the graph also shows several episodes that received high ratings even during the declining period. Some episodes received ratings over 9. Lastly, the last three episodes of the series show a sharp increase in ratings. This suggests that the show was still capable of delivering highly rated episodes even though its overall appeal declined. In conclusion, the pattern suggests that the audience appeal did not simply decrease at consistent rate, some episodes delivered beyond expectations and continued to have strong reactions.

Comparing Popularity and Audience Appeal

By comparing the two previous graphs we can conclude that the popularity and audience appeal of The Office TV show did not change in exactly same way throughout the series. The viewership increased during the early years of the show and remained fairly strong thought the middle seasons, while having a noticeable decline during its last seasons. IMDb ratings followed a somewhat similar pattern, but the decline was not consistent. Even when the show attracted fewer viewers, it still managed to receive very high audience ratings.

This difference suggests that the lower viewership does not mean that the remaining audience dislikes the show. The popularity measures the number of people that watched the show during the original broadcast, while the appeal measures how positively viewers evaluated the episode. Consequently, some later episodes had a smaller audience but a very high rating. This shows that two measures are related but not identical.

Viewership by Seasons

ggplot(data = office_ratings)+
  geom_boxplot(mapping = aes(x = season,y = viewers))+
  labs(
    title = "Viewership by Season",
    x = "Season",
    y = "Viewers (millions)"
  )

The boxplot shows a clear change in viewership over the nine seasons of The Office series. First, eighth and ninth seasons of the series show a relatively low viewership, while seasons two through 5 attracted much larger audience levels overall. Seasons six and seven show a clear sign of the decline of viewership. Season five shows a very large outlier, which matches the previous findings of an unusual episode with more than 22 million viewers. Overall, the season comparison shows that the earlier series grew in popularity, maintained and peaked viewership during the middle of the show, and finally declined during its last seasons.

IMDb Rating by Season

ggplot(data = office_ratings) +
  geom_boxplot(mapping = aes(x = season,y = imdb_rating))+
  labs(
    title = "IMDb Ratings by Season",
    x = "Season",
    y = "IMDb Rating"
  )

The IMDb rating boxplot shows that the audience changed across seasons and compared to the viewership it’s much different. Seasons 3 and 4 had the highest rated episodes with most of them receiving ratings over 8. Season 5 was also strong, but it shows more variations in episodes ratings. The real decline in the ratings can be attributed to season 8 and 9 where the typical rating received a lower rating compared to the stronger seasons of the show. However, these seasons also received several highly rated episodes which can be especially seen in season 9 where three episodes received much higher ratings. We can also see that season 8 delivered the worst rated episode of the show with the outlier of lower than 7 IMDb rating. Overall, this shows that series season became weaker over time.

Viewership Within Individual Seasons

ggplot(data = office_ratings, mapping = aes(x = episode, y = viewers, group = 1))+
  geom_line() +
  geom_point() +
  facet_grid(rows = vars(season)) +
  labs(
    title = "Viewership Within Each Season",
    x = "Episode Number",
    y = "Viewers (millions)"
  )

The facet graph shows that there is no clear pattern of viewership in each season. Most of the episodes in each season fluctuate, and we can see that there’s a small gradual decline in later seasons. First season began with unusually large audience and then it quickly stabilized. Seasons 2 through 4 remained relatively consistent, while season 5 delivered a huge spike with one episode receiving more than 22 million views. From season 6 to 8 we can see somewhat consistent viewership with slight decline. Last couple of episodes of the series show an increase in viewership. Overall, the changes within season were gradually smaller than long-term decline, with a few major noticeable changes and temporary spikes.

Conclusion

The analysis shows that the popularity and audience appeal of The Office series changed over time, but they did not change in the same way. Viewership increased during the early seasons of the show, while audience peaked during the very low viewership. Another conclusion is that popularity, viewer engagement, and audience appeal are related but do not measure the same success. Overall, with several unusual episodes, a very long run, and overall strong IMDb ratings, The Office is definitely on my list of series to watch next.