For the last forty-two years, people all over the globe have been changing their televisions sets over to NBC as soon as the clocks in New York City stuck midnight. Here they hoped to watch their favorite celebrities, musicians, and budding comedians dress up in wigs and costumes and crack wise with one another. After that first night in October of 1975, Saturday Night Live has been a powerhouse of creating comedic canon. Whether it was the coneheads, two wild and crazy guys, or Christopher Walken's fever that could only be abated with a little more cowbell, Saturday Night Live has served as a great cauldron of laughs. Beyond its own immediate public appeal, the show has served as a jumping-off point for generations of actors. The collective entertainment garnered by the careers of Will Ferrel, Amy Poehler, Chevy Chase, Mike Myers, and hundreds of others whose careers began in studio 8H proves the show's massive influence in the world of entertainment media. Clearly, the show has had huge impacts on pop culture, but more recently, Saturday Night Live has been under a critical lens to determine the other aspects of contemporary life that it can affect.
The show has always relied on parody and satire of current events as bountiful sources for comedy. However, at what point does the comedic representation become a precision instrument for the expression and criticism of political thought? Can a comedy show be a serious player in shaping national perceptions of individuals and ideas? These are some of the questions that Kathryn Brownell investigated in her article “The Historical Presidency: Gerald Ford, Saturday Night Live, and the Development of the Entertainer in Chief.” She discusses how presidents have used the show in what she calls “performative politics,” using entertainment media to shape public sentiment and perceptions of political processes and offices. As soon as the second year of the show, people already recognized its potential as a tool of political expression and communication. The viewership and presentation of the show allowed for a more informal and therefore more trustworthy avenue for politicians to communicate with their constituents. Though this did benefit Gerald Ford, it had the unintended result of establishing Saturday Night Live as a force of political commentary.
This complicates the shows role in society, as occasionally it can become difficult for viewers to parse out when the show is being genuine about political news, when it is being strictly comedic, and when it is serving as a fusion of news and comedy. This has become especially apparent in more recent years as politicians have become even more visible, often serving in similar capacities as a celebrity. This has made them the target of satire and criticism, and it is not always clear to what degree they are being given a fair representation. One especially popular political impersonation on the show over the last few years in Sarah Palin. Her opinions, public statements, and visual similarity to show veteran Tina Fey provided much material for the writers of the show. Jason Peifer highlights the potential impacts that this murky political representation can have in her article “Palin, Saturday Night Live, and Framing: Examining the Dynamics of Political Parody.” She illustrates that these parodies play a strong role in “creating political realities.” The show has evolved from simply a reflection on current states of affairs into a dynamic force in shaping the political reality.
As previously mentioned, however, it is key to note that the realities that are shaped by the show are not always objective, representative, or equally degrading. While initially, Gerald Ford used it intentionally as a piece of media to shape perceptions about him, it now defines public personalities without their permission and often in a negative light. Nickie Wild in “Dumb vs. Fake: Representations of Bush and Palin on Saturday Night Live and Their Effects on the Journalistic Public Sphere” points out that the show can be a source of information for voters, despite the fact that it may never communicate with the people it is representing. Under the guise of comedy, Saturday Night Live is able to selectively and subjectively educate people about their politicians and political culture. With its rising power to influence this nation's political consciousness, it seems more important that ever to approach its producers and its consumers with a critical and analytical investigation.
Research Question and General Thesis¶
It was this potential for selective education that piqued my interest in studying the show. Clearly it has a lot of potential to impact the political climate. This has led to two distinct methods of exploratory analysis of show. The first method surrounds individual actors, given their ability to represent political figures, culture, and their role in shaping the show’s direction, I was curious who the most prolific and central individuals are who have been involved with Saturday Night Live. Has SNL’s culture stayed true to a few influential founders, or are more recent cast members more important in influencing the show as it evolves over time? The second is how the show as a whole has shifted in its reception over time. When has the show been received well, and when has it been received poorly? As we have seen, there has been an increase in political parody in more recent years, and has this increase been viewed positively or negatively by viewers, and what does this say about how people feel about representations of politics in recent years?
As a result of my investigation to the scholarly work, as well as my own experience watching the show, I theorized that Saturday Night Live is the most relevant today as it has ever been. With the increasing role that comedy and satire have played in the political process, I believe that ratings will reflect this in recent years, especially among young people, who have grown up in an age of digital accessibility that makes SNL skits easy to find, and political dissatisfaction. Given the abundance of episodes and skits, however, I was unsure if the actors who have done political impressions would be the most influential to the show. I think that analysis of the actors will demonstrate that the more recent ones will be the most relevant, interacting with current cast members, and older ones who are coming back as hosts.
Loading Packages and Data¶
Going in, I was unsure of exactly what data packages I was going to be using. I knew that I was most probably going to need ggplot2, a package used for producing more nuanced graphics, as well as igraph, a package used for creating networks from data. It never hurts to have more functionality and more options for visualization.
library(HistData)
library(reshape)
library(RColorBrewer)
library(ggplot2)
library(rgdal)
library(sp)
library(rgeos)
library(maptools)
library(igraph)
library(rgexf)
library(gridExtra)
I downloaded the data from a website called Kaggle, a site that runs competitions for digital models, where users can post and download any number of different data sets. The original data came in six different files and so each one had to be read in individually, hence the six different commands and the six different data frames.
actor_type <- read.csv(file="snl_actor_title.csv", as.is=TRUE)
actor <- read.csv(file="snl_actor.csv", as.is=TRUE)
episode <- read.csv(file="snl_episode.csv", as.is=TRUE)
season <- read.csv(file="snl_season.csv", as.is=TRUE)
rating <- read.csv(file="snl_rating.csv", as.is=TRUE)
title <- read.csv(file="snl_title.csv", as.is=TRUE)
head(actor_type)
head(actor)
head(season)
head(episode)
head(rating)
head(title)
The next steps appear a little redundant, as I needed to merge the separate data frames into one. This became a problem, because the different elements of the data frames would either be needed or an obstacle to the ultimate visualizations and argumentation I used. For my network analysis, I wanted the names of every actor in each episode, causing repeats for actors who appeared in multiple episodes. While necessary for the network, I needed to use different iterations of my data further on when I stepped back to just look at episodes and seasons.
Luckily, however, the data that I downloaded came pre cleaned, free from missing values and NAs, so there was no need to do any sort of NA removal or data cleaning of my own.
The first thing that I wanted to do with the data was a network exploration of all of the different actors involved with Saturday Night Live. To do this, I needed to combine the "episode", "actor type", and "rating" data frames. I wanted a data frame where each row represented an actor in a particular episode, so that I could create a network of people that had starred in episodes together. As a part of the visualization, I wanted to potentially include the rating of specific episodes, as well what kinds of actors (hosts, cast members, musical guests) were more important to the show. Each new "total" is the introduction of an additional data frame, merging them by season and episode, common columns in most of the data frames.
Finally, I decided to include the "actor" data frame, because while "actor_type" had the majority of the information I wanted, such as episodes and seasons, it represented the actors by abbreviations of their names, less helpful for visualizing. I then subsetted the data to have it ordered chronologically by season and then episode.
rating <- rating[order(rating$sid, rating$eid), ]
total <- merge(episode, rating, by=c("sid", "eid"))
total <- merge(total, actor_type, by=c("sid", "eid"))
total <- merge(total, actor, by= "aid")
total <- total[order(total$sid, total$eid), ]
head(total)
unique(total$actorType)
Immediately we can see some repetitions in the data frame, there are two rows of Al Franken for the first episode of the first season. This highlighted a problem that I was going to experience in using the "total" data frame for the nodes of the network. Not only would small errors like the one that produced the double Franken create errors, but I hadn't previously realized that I would have multitudes of repeat nodes. I ran the function unique to find out all of the different "actorType" options there were: crew, cast, guest, music, filmed, host, cameo, and unknown. Problems of repeat performances immediately become clear. Justin Timberlake has been both a host and musical guest. Some cast members have been cast, crew, guest, and host. Because of these overlaps, which would cause problems in having a node attribute of actor type, because some people have multiple, I decided that for the network, I was only going to look at specific actors’ names, hence the creation below of a new character vector that only has each person's name once.
vertices <- unique(total[ ,c("name")])
head(vertices)
class(vertices)
After creating the data frame for the nodes, I now needed to create one for the edges of my network. The following code was the first step, taken from the code that Professor Hall used to create the costarring network for actors in movies that had been nominated for best picture. Initially it seemed like it would be impossible, because unlike the movies, episodes do not have unique names. Luckily and somewhat unexpectedly the column "aired" which represents the date that each episode aired has a unique character representation for each episode in the form of a day, a month, and a year. Using this, and the unique names of actors, I changed the for loop to create a hit whenever an actor had appeared in a new aired episode, creating an edge between every actor that had costarred in an episode together.
trueEdges <- list()
actors <- unique(total$name)
episodes <- unique(total$aired)
for (i in 1:length(episodes)){
hits <- which(total$aired == episodes[i])
subsetnodes <- unique(total$name[hits])
rating <- unique(total$US.users_avg[hits])
newedges <- data.frame()
if (length(hits) > 1){
for (x in 1:length(subsetnodes)){
if (x != length(subsetnodes)){
miniset <- data.frame(S = subsetnodes[x], Target=subsetnodes[-c(1:x)], Episode=episodes[i], Rating=rating,
stringsAsFactors = FALSE)
newedges <- rbind(newedges, miniset)
}
}}
else{
newedges <- data.frame(S=total$name[hits], Target="", Episode=episodes[i], Rating=rating, stringsAsFactors = FALSE)
}
trueEdges[[i]] <- as.data.frame(newedges)
}
library(plyr)
edge_df <- rbind.fill(trueEdges)
Once the data frames had been created for both the nodes and the edges, I used the igraph package to create a network of all the SNL actors. In the next section of code, I create the network, set a layout for the nodes, and create a visual representation of Saturday Night Live contributors of the last forty-two years.
g=graph_from_data_frame(edge_df, vertices, directed=FALSE)
g
set.seed(1)
lout <- layout.fruchterman.reingold(g)
plot.igraph(g, layout = lout, asp = 0, vertex.size = 5, vertex.label.cex = 0.5)
Several things about this step presented as problems after running the above code. The first was the numbers involved with the network, there are 2,135 nodes connected by 165,259 edges, a massive amount of information to plot. As a result of this, igraph and R took about twenty minutes within the notebook to plot the network. This created several problems. Visually, the network is impossible to read, it is clustered and looks meaningless. However, the process of cleaning the visualization involves trial and error for different values and subsetting for spacing, label and node size, edge thickness, etc. I couldn't keep on waiting twenty minutes every time I wanted to adjust a visual aspect of the igraph. Given this, with the counsel and assistance of Professor Hall, I decided to convert the network to a .gephi file, so that I could manipulate it in software that is built for handling networks. Although the following code was not actually used to accomplish the exportation of the data into an object readable by Gephi, this is what the following code would accomplish.
g_gephi <- igraph.to.gexf(g, position=NULL)
f <- file("SNL.gexf")
writeLines(g_gephi$graph, con = f)
close(f)
Gephi was a far more effective method of handling a network of this size, but I found that the data file needed a few new attributes so that I could create specific visualizations. In order to demonstrate whether the old guard of popular SNL classics was more or less influential than newer members that have connected with older ones, the data needed to be adjusted. I needed to add an additional edge attribute based on when the edges(episodes) took place. Additionally, Gephi's software was having trouble coloring the edges by their rating, and a column needed to be added for color so that Gephi could read them. Using the following code, provided by Professor Hall, I was able to explore visualization using these new columns.
After reading in the edges data frame that had been exported as a .csv, the code creates distinct bins (ranges of values) for different ratings, creating a new column in the data frame called "binID" and placing each row in a bin according to its rating value. It then creates a vector of colors, in the reverse of R's "heat.color" pallet with a color for every single bin. A new column, "color" is added to the data frame where every single row is given a color based on its binId. The code then looks at the first ten rows of the two named columns in the data frame to verify that it has matched up. This section of the code will make it easier for Gephi to assign colors to edges, by already having run the code to associate color and rating, rather than relying on Gephi to do so.
The next section of the code is a for loop, where a new attribute is created and defined: weight. In this case, weight is being assigned based on the date of the episode, with higher weights representing more recent episodes. A date vector is created using every unique episode air date from the data frame in the "Episode" column. The as.Date function is an attribute of R where it can read in character vectors as dates, and in this loop are subsequently assigned a weight value. The dashes are removed, the new weight column is explored, and then adjusted, in this case divided by ten million to make it more manageable.
Finally, the code specifies that the edges and therefore the network is undirected, and the new data frame, with both the color column and the weight column are exported in a .csv to be used by Gephi.
snlEdges <- read.csv("SNLEdges.csv", header = TRUE, stringsAsFactors = FALSE)
bins <- unique(quantile(snlEdges$Rating, seq(0,1, length.out = 10)))
snlEdges$binId <- findInterval(snlEdges$Rating, bins)
colSet <- rev(heat.colors(length(bins)))
colSet
snlEdges$color <- colSet[snlEdges$binId]
snlEdges[1:10, c("Rating", "color")]
write.csv("SNLEdgesColor.csv")
colnames(snlEdges)
dates <- unique(snlEdges$Episode)
snlEdges$weight <- 0
for (i in 1:length(dates)){
hits <- snlEdges$Episode == dates[i]
date_no <- as.Date(dates[i], "%d-%b-%y")
snlEdges$weight[hits] <- as.character(date_no)
}
snlEdges$weight[1:10]
snlEdges$weight <- gsub("-", "", snlEdges$weight)
snlEdges$weight[1:10]
class(snlEdges$weight[1])
snlEdges$weight <- as.numeric(snlEdges$weight)/10000000
snlEdges$weight[1:10]
snlEdges$Type <- "Undirected"
write.csv(snlEdges, "SNLEdgesColorWeight.csv", row.names = FALSE)
Attached is a pdf file that includes the network visualization generated in Gephi, and a caption that details the attributes of the visual, and the manipulations within Gephi done to achieve them. Below I include a few more steps of code and the overall interpreations of the code and the network, in terms of who are the most important actors in the show.
The code below calculates the eigenvector centrality, betweenness centrality, and degree of the network, and correlates them, to demonstrate if the network visual, displaying eigenvector and degree, is missing a different story of centrality and therefore importance that would have been communicated with betweeness centrality.
eigenCent <- evcent(g)$vector
sort(eigenCent, decreasing = TRUE)[1:10]
sort(eigenCent, decreasing = FALSE)[1:10]
betweenCent <- betweenness(g)
sort(betweenCent, decreasing=TRUE)[1:10]
sort(betweenCent, decreasing=FALSE)[1:10]
deg <- degree(g, mode="all")
sort(deg, decreasing=TRUE)[1:10]
sort(deg, decreasing=FALSE)[1:10]
(cor(eigenCent, betweenCent))^2
(cor(deg, betweenCent))^2
(cor(eigenCent, deg))^2
As we can see, there is a moderate correlation between degree and betweenness centrality, indicative that the network would look quite similar if the nodes were sized by betweenness. There as an even stronger correlation between eigenvector and degree, indicating that whoever the most central nodes in the network are, they are both prolifically connected, and connected to many other well connected actors. These nodes are massively influential in the show, and the subsequent political realities that the show takes part in creation. The last question to ask is: Who are these nodes?
Unsurprisingly, there are many nodes in the network that have low eigenvector centrality and degree, represented by the sea of small, yellow dots. These are the one time hosts and musical guests that have appeared on the show a single time, and are relatively unimportant. The most crucial story that Figure 1 demonstrates relates to the dates around which the most influential cast members are starring in episodes. Given the concentration of larger and darker nodes on the left side, connected by dark green nodes, this network tells us that the most influential actors are those that have been appearing on the show in more recent years. This aligns with the notion that SNL is the most relevant it has ever been, and the actors participating in the more contemporary, more politically satirical version of the show, are more central to its functioning. They have interacted with the most and the most well connected other actors, former cast members coming back as hosts, as they all participate in the blossoming age of political parodies. As more recent actors on the show become more and more powerful in shaping the culture, one that is increasingly based around political representations, it is crucial to move from a look at the show itself, and into public reception of the show. Are people reacting positively to the more recent years and more recent actors? Or, despite their gaining power within the show, is SNL itself fading as a relevant media object in the eyes of its viewers? Conversely, is SNL and its new political approaches giving the comedic impressions of more recent cast members even more power in shaping political realities?
A Look at SNL Ratings Through the Years¶
After exploring the network attributes of who has been the most influential, and when they have been on the show, I began looking at the more numbers based reception to Saturday Night Live. I wanted to look at how rating's have changed over the years, whether today's climate for political satire has made the show more popular, and if there are any demographic differences in who is reacting well to Saturday Night Live. The creator of the "ratings" data frame stated that he used IMDb statistics to generate the ratings data. The IMDb website has a measured amount of transparency about how they calculate their ratings (http://www.imdb.com/help/show_leaf?votestopfaq). Basically, only registered users can cast votes, each account can only have one vote per item, votes can be recast, but they overwrite the previous score, so one account can't have thirty separately considered ratings. However, it is important to note that the ratings are calculated using a weighted average, but to "avoid leaving the scheme open to abuse" they do not tell you exactly how the weighting is done. Before using this data, I wanted to acknowledge that this is just a single method of determining reception of SNL episodes, and there are several biases to note. Because it is volunteer-based, the ratings do not reflect the opinions of every person that has watched the episode. Additionally, there is no way to prevent a single user from registering multiple accounts, or from voting on items they have not actually seen. I have no reason to expect that these are serious issues with the data, but before using it I thought it was crucial to acknowledge how the data could be flawed, and therefore the conclusions drawn from it are framed by whatever biases existed in data collection.
Below is a table coming from IMDb's website (http://getsatisfaction.com/imdb/topics/percentage-of-registered-users-by-gender) that gives, as of two years ago, the age and break down of registered users. Clearly there is a pretty large skew towards male users, and those aged 18-45, important considerations in moving forward. When looking at overall data it is important to remember that it is not taken from a sample of users representative of the population, and there are large percentages of users whose demographics are unknown.
| All | Male | Female | Unknown | |
|---|---|---|---|---|
| Age | ||||
| All | 1,477,645 (100%) | 1,025,298 (69%) | 208,914 (14%) | 243,433 (16%) |
| Under 18 | 15,833 (1%) | 12,988 (1%) | 2,845 (0%) | 0 (0%) |
| 18-29 | 632,346 (43%) | 524,807 (36%) | 107,539 (7%) | 0 (0%) |
| 30-44 | 434,688 (29%) | 370,132 (25%) | 64,556 (4%) | 0 (0%) |
| 45+ | 88,809 (6%) | 72,190 (5%) | 16,619 (1%) | 0 (0%) |
| Unknown | 305,969 (21%) | 45,181 (3%) | 17,355 (1%) | 243,433 (16%) |
I began by looking at a simple breakdown of episode and season ratings for the entire life of the show. This required once again bringing in the rating data, because as mentioned above, when I merged the data for the network, I ended up with repeats for episodes for as many actors were involved, and I was now just curious at stepping back and looking at episodes and seasons, rather than individual actors.
rating2 <- read.csv(file="snl_rating.csv", as.is=TRUE)
head(rating2)
In deciding what rating category I wanted to use for looking at overall ratings overtime, I decided to look at the average for US users. As is displayed by the summaries of each rating system below, Us.users has a more precise range, although both the raw user ratings and the average were pretty centralized, with not much spread in the interquartile range. I decided it would be easier to approach the data with a range of 6.3, rather than 166. In order to assist with visualization of episodes that have been reviewed as relatively better or worse instead of objectively according to IMDb's ratings across all media, I cut the ratings into ten distinct bins, with bin "10" being home to the highest rated episodes, and "1" being the lowest rated episodes.
summary(rating2$US.users)
summary(rating2$US.users_avg)
allbins <- unique(quantile(rating2$US.users, seq(0,1, length.out= 11)))
allvals <- cut(rating2$US.users, allbins, labels=FALSE, include.lowest=TRUE)
blues <- RColorBrewer :: brewer.pal(10, "BrBG")
ggplot(rating2, aes(factor(sid))) + geom_bar(aes(fill = factor(allvals)), position = "fill") +
#scale_y_continuous("", breaks= NULL) +
scale_fill_manual(values=blues) +
labs(x = "Season", y ="Percentage of Episodes", color = "Rating") + theme(axis.text.x = element_text(angle = 90, hjust = 1))
This graphic presents the ratings of every single season of Saturday Night Live. Each bar represents a single season, colored by ratings. Because there is variability, good and bad episodes in every season, I decided that it would be more telling to look at which seasons have had the most of their episodes reviewed either positively or negatively. As mentioned above, the color bins are so that we can see how SNL episodes and seasons have been rated relative to one another, rather than all seeming less objectively popular, because they haven’t been rated as highly as incredibly popular moves, for example. Each bar (season) is colored by the percentage of episodes it had that fell into each rating bin. The coloring of the bins is labeled on the legend to left, with darker teals being the best, and darker browns being the worst.
All the way on the left side of the graph, we see that viewers have responded well to the very first season, with all episodes being rated relatively highly. This is followed by a progressive fall in ratings until season 13, where ratings remained generally unfavorable, with a few seasons being reviewed more positively. Ratings had temporary bumps in seasons 17, 19, and 26. Around season 30 however, we see the negative reviews begin steadily decreasing, replaced with more positive ones. This coincides with scholarly reflections of SNL as adopting the role of political parodists especially in more recent years. Ratings began rising around ten years ago, in 2007-2008, right when Sarah Palin, the thus far unknown eccentric governor of Alaska was selected by Republican frontrunner John McCain as a Vice Presidential candidate for the 2008 presidential election. The highest rated seasons in years, it is no wonder that the show began to make political skits a regular factor of the show. The highest rating years are those that directly precede or follow a presidential election, as the show is infused with new character impressions to do. The highest rated season ever, the most recent, boasted a huge selection of political commentaries following the 2016 election. Alec Baldwin's Donald Trump, Melissa McCarthy's Sean Spicer, Kate McKinnon's Betsy DeVos, Jeff Sessions, and Kellyanne Conway have made it so that, in season 42, over three fourths of the episodes were reviewed in the top decile of ratings for episodes. This timeline supports the idea that recent political parodies have made the show more and more popular, giving it more power to create political impressions of members of the government.
Looking at only the overall data can dilute other more interesting trends, however. In the next sections of code I break down the ratings by gender to see what, if any, connection there is between these variables and if certain groups are more predisposed to like the newer, more politically active Saturday Night Live.
Reactions of Men and Women¶
In order to get another story of the shift in ratings, I am looking at is the difference between the show's reception among men and women. I expect the two progressions of ratings to be pretty strongly correlated, because men and women in the country are living in the same political climate, and I would expect them to react similarly positively to an increase in political parody. As the show has tied more directly into spheres other than pure comedy, I expect both men and women to be finding it more relevant and therefore rating it more positively.
men <- qplot(sid, Males_avg, data = rating2, main="Male", xlab="Season", ylab="Rating", ylim=c(2,11), color = I("darkblue")) + geom_smooth()
women <- qplot(sid, Females_avg, data = rating2, main="Female", xlab="Season", ylab="Rating", ylim=c(2,11), color = I("darkred")) + geom_smooth()
grid.arrange(men, women, ncol=2)
summary(rating2$Males_avg)
summary(rating2$Females_avg)
r <- cor(rating2$Males_avg, rating2$Females_avg)
r^2
In this graphic I explore the different ways that male and female users have rated episodes of SNL throughout the years. Every single dot represents an individual episode; they are arranged vertically according to which season they fall in. Above is the five number summaries that accompany the male and female rating plots respectively. These are meant to highlight both trends and more comparable stats. In both graphs it is apparent that seasons have potential to have pretty large variability. The geom_smooth lines of both present the most interesting trends. Both genders seem to have initially thought decently highly of the show, but men appear to have lost interest a little more slowly at first, compared to women whose ratings drop a little quicker over the first ten seasons. Men continued rating it lower and lower until season twenty-seven, the lowest rated season for men. Men then began steadily rating it higher, with a slight increase in the rate of improvement around the same time that was shown in overall ratings, around season thirty-four, when political parody became more common. Female ratings have had slightly more fluctuation, hitting their lowest around season fifteen, and peaking right around season thirty-two or thirty-three, beginning to decrease over more recent, more politically charged years of the show. It is important to note, however, that women have traditionally rated the show higher than men, the five number summary is higher on all counts for women, and that it is only in the last few seasons that men have been reviewing the show more highly.
What does this say about the show, and/or the system used to rate it in this data set? Initially, it is important to draw attention to the fact that the pattern of reception for men is more similar in trend to the overall data, reflecting the gender makeup of rating users displayed in the table above. Beyond this mirroring in the male plot, it is interesting that there is such a small correlation (r-squared value of 0.10) between the way men and women have rated episodes. The fact that men have been more receptive to SNL since, as scholars have noted, they began doing political parody, is perhaps indicative of the general male humor, and its response to these potentially fraudulent and offensive depictions. The low correlation, and the dropping approval amongst women demonstrates that the more recent structure and content of Saturday Night Live is doing well with men, but is less appealing to women.
Conclusion: Gendered Humor Amongst the Rise of Political Criticism¶
Although initially I had wanted to do more analysis of age differences, I decided that it would be too inconclusive. Age as a variable in this particular data set is a little too murky, as age on the date of viewing and age on the date of rating may be different, even falling into different age categories. Given the fluctuations that individuals can experience over a lifetime of viewership, it would have been difficult to draw any conclusions from a variable that is not representative of the same thing for every data point.
From my investigations both of the network of actors and of the changes in ratings, it is clear that contemporary Saturday Night Live is the most influential it ever has been. Going into this project with my own subjective viewership, confirmed by research, I was expecting that in an era abundant with sometimes ridiculous political celebrities, an impression-based show like Saturday Night Live would be doing better than ever.
My network analysis confirmed that the more recent cast members of SNL have more power to shape the legacy of SNL, and therefore increased power to sculpt the political realities that the show now deals in. Similarly, looking at overall trends in male viewers, there is a distinct rise in the show's ratings since they have started to more heavily utilize political parody. The most unexpected, and potentially most crucial finding was the falling female opinion of the new political era of the show.
One potential explanation of this is the ways in which women in power tend to be politically criticized. In a New York Times article, Susan Chira reported on the superficial and sexist tones that often define criticism of female political figures. SNL parody often accentuates personal flaws in its subjects, and it is possible that men are more insulated against the sinister nature of the far-reaching effects of this kind of criticism. Men may see the recent, politically charged years of SNL as simply populated with more humorous impressions of people they have heard of and see on the news, whereas female viewers have begun to lose interest in the show given the traditional criticisms of women in power as reductive and dangerously generalized.
This has been corroborated by scholarly investigations into the most recent season, the one that was collectively rated by women as the lowest in seventeen years despite being the most watched. In their article "Homophobic masculinity and vulnerable femininity: SNL’s portrayals of Trump and Clinton," Weinhold and Bodkin explore the differentials in male and female political criticisms. Both genders are reduced in their portrayals, but the stereotypical male has desired qualities hyperbolized for comedy, rather than the negative ones of female impressions. Although fake-tanned and poorly-tailored, Alec Baldwin's bumbling Trump is dynamic, social, and over-confident. The portrayals of women, however, draw on gendered stereotypes of femininity as being synonymous with vulnerability and sometimes even submissiveness, observable in Kate McKinnon’s portrayals of a powerless Kellyanne Conway as well as Hillary Clinton. Based on their presentation and mannerisms, these criticisms reduce female politicians to these superficial and often more negative attributes and ignore any discussion of merit or belief. The representations of genders, although equally comedic, can still be understood as more degrading to women than to men. This may have little to nothing to do with the difference in male and female receptions to the show, but it is a possible factor. Beyond that it is an interesting example of systems of media and information presentation that may privilege a male experience.
As noted, this gendered difference is not completely isolated within the sphere of entertainment media, and shows like Saturday Night Live are shaping political perceptions of its viewers. The show has been proven to alter perception in what Jody Baumgartner, Morris, and Walth call the "Fey Effect." Exposure to Tina Fey's impression of Sarah Palin, presented in the gendered terms that Weinhold and Bodkin discuss, was found to have impacts on voters' attitudes towards Palin and her role in politics. If Saturday Night Live can shape perceptions of individual politicians, then it seems reasonable that it is also capable of demonstrating a link in between female politicians and submissiveness, that even the most powerful women in the country are still second in traditional qualities of leadership to their male counterparts.
This is a much more tangential analysis of the increasing popularity of the show, one to think about in terms of future studies into political humor and entertainment/news media fusions. Especially if they continue to be rated more highly in the future, particularly just by one gender. But what I found in my analysis supports current scholarship, that political criticism is on the rise, and that it has real life consequences in affecting political and gender attitudes in viewers. The empowering of SNL to be a source of political fact by "Entertainers in Chief" such as Geralf Ford, carved out channels through which gendered rhetoric has begun to flow more readily. This is incredibly important to think about in the broader context of the gendered attitudes of society and politics, attitudes that shows like Saturday Night Live may be helping to reproduce and strengthen.
Works Cited¶
Baumgartner, Jody C., Jonathan S. Morris, and Natasha L. Walth. 2012. "The Fey Effect." Public Opinion Quarterly 76, no. 1: 95-104. Academic Search Complete, EBSCOhost.
Brownell, Kathryn Cramer. "The Historical Presidency: Gerald Ford, Saturday Night Live, and the Development of the Entertainer in Chief." Presidential Studies Quarterly 46, no. 4 (December 2016): 925-942. Academic Search Complete, EBSCOhost.
Chira, Susan. "Sexist Political Criticism Finds a New Target: Kellyanne Conway." The New York Times, March 5, 2017. http://www.nytimes.com/2017/03/05/us/kellyanne-conway-sexist-political-criticism.html?_r=0.
Michaud Wild, Nickie. 2015. "Dumb vs. Fake: Representations of Bush and Palin on Saturday Night Live and Their Effects on the Journalistic Public Sphere." Journal Of Broadcasting & Electronic Media 59, no. 3: 494-508. Academic Search Complete, EBSCOhost.
Peifer, Jason T. 2013. "Palin, Saturday Night Live , and Framing: Examining the Dynamics of Political Parody." Communication Review 16, no. 3: 155-177. Academic Search Complete, EBSCOhost.
Weinhold, Wendy M., and Alison Fisher Bodkin. 2017. "Homophobic masculinity and vulnerable femininity: SNL’s portrayals of Trump and Clinton." Feminist Media Studies 17, no. 3: 520-523. Academic Search Complete, EBSCOhost.