Download PDFContents
Methodology
Sampling
The process for collecting a sample was random. The first task was to compile a list of Netflix’s American-produced original television programs. Specifically, I drew from Netflix’s live-action comedy and drama programs, excluding reality programs, nonfiction programs, and children’s programs, as they use different storytelling mechanisms and narrative beats.2 While an argument can be made to focus on only television or film programs, streaming platforms are blurring the lines between these two mediums. The flexibility and consequent freedom of streaming have made the biggest difference now between film and television an individual program’s length.
There were 337 items in the list, meaning there were 337 programs to choose from. The website www.randomnumbergenerator.org was then utilized. For the first round, 20 numbers were drawn. The goal was to get 30 different programs so a second coder and I could code as many different programs as possible within one semester. Duplicates were automatically discarded. I then went to the IMDb pages of each program to see if they had any Asian/American3 actors in their cast, looking at headshots and names. Through this method, I was able to find all regular, recurring, and guest characters. I was not able to find information on extras. Only programs containing at least one Asian/American character were kept in the sample. All other programs were noted for their lack of roles but coded no further.
Once programs were determined, the website was utilized again to determine which random episode I would watch, as well as which would be viewed by both my second coder and me. Only one episode from television programs was watched to widen the study’s scope. In instances where a program had more than one Asian/American character, we coded the four top-listed characters on IMDb. Consequently, the study would code for more than one episode if that second or even third episode contained one of the four top-listed characters.
Coding Scheme
The coding scheme consists of five levels of variables, with 55 variables total. For guidance on how to structure my coding scheme, I looked at Scharrer’s (2012) prior content analysis.
The first level has one variable: the number of Asian/American characters in the program. If a program chosen in the sample has no Asian/American characters, my second coder or I would log a “0” and move on. If there were any Asian/American characters, we would record the number and move onto the next level of the coding scheme. The second level consists of variables on program demographics: the name of the program, the name of the episode (if applicable), the program airdate, genre, and format. The third was variables related to demographics: their name, their billed versus narrative role4 , ethnicity, gender, and occupation. Variables such as immigrant status, accent, and traditional dress were included to see how often Asian/American characters were depicted as more foreign. All variables were determined by the character verbally declaring these aspects of their identity or exhibiting visual cues.
Level four held the most potential for ambiguity, as it consisted of variables related to the character’s personality. Variables were determined by drawing upon historical representations of Asian/Americans in American media to the following characteristics: prowess/success (an ability to achieve success in the following categories: education, business, STEM, humanities, and social skills), duplicity (lies, gives a false impression), ambition (desires to rise above their station), selfishness (considers only their own wants or needs, extroverted (goes up to people and chats enthusiastically, befriends others easily, is open about their personal life), spirituality (believes specifically in “chi” and other spiritual forms associated with Eastern religions and philosophies), formal religion (strictly follows official religious practices), and displays emotions (displays the following emotions: happiness, sadness/grief, anger, fear, empathy/sympathy, guilt/remorse). To code these variables, the coders write yes, no, or unknown.
Level five variables related to the character’s relationships with others. Many were drawn from Dillman Carpentier et al. (2017). We chose to focus on this aspect because as Deo et al. (2008) wrote, many Asian/American characters are not written to hold relationships in a way that allows them to build families and/or fold themselves into the fabric of American society. Within these variables were two categories: behaviors and conversations. In the former, attention was paid toward whether Asian/American characters engaged in dating/courtship behavior (a direct manifestation of changing the relationship in question [such as building or ending a relationship]), light kissing/touching (touching another’s body in a way that was meant to be loving [e.g. light or closed-mouth kisses between partners]), physical flirting (behavior intended to promote sexual interest), passionate kissing, intimate touching (touching of another’s body in a way meant to be sexually arousing), masturbation, implied intercourse (strong inferences that physical acts involving sex had occurred, were occurring, or would immediately occur), and depicted intercourse (a direct view was shown of any person engaged in sexual intercourse regardless of the degree of nudity shown), all with yes or no options. We also asked whether contraception was used, if the sexual acts were mundane or kinky (followed traditional norms of sexuality), what attitude the characters had towards sex (callous, casual, neutral, committed), and how explicitly naked they were. In the conversations category, we focused on whether Asian/American characters ever discuss these relationships, their histories, or even just said words that insinuated sex. In addition, a variable on who initiated the behaviors or conversations was included to see how much agency Asian/American characters have in these interactions.
To ensure that the coding scheme was applied reliably, I recruited a second coder, a Chinese woman in her late 50s unfamiliar with content analysis or the full scope of the study. After taking a half hour to explain the different variables, my second coder and I double-coded roughly 10% of the overall sample. Out of 55 variables, the second coder and I only disagreed on 12: Beauty (89% agreement), Business Prowess (67%), STEM Prowess (89%), Humanities Prowess (67%), Extroverted (89%), Anger (67%), Fear (78%), Physical Flirting (89%), Implied Intercourse (89%), Attitude Towards Sex (72%), and Explicit Nudity (93%). With our lowest average being only 67%, the coding scheme proved to be reliable. Consequently, if a variable is discarded, it is due to its (lack of) relevancy to the study rather than a statistical matter.
Autoethnography
In addition to the content analysis, autoethnography was incorporated as another, qualitative methodology for this study. For myself, I wrote down my personal gut reactions to what was on the screen in a separate document. My second coder is also Asian/American; consequently, I include her remarks in our discussion following the overlap sampling. Through these qualitative examples, I add a dimension to the quantitative data accrued through coding the texts. I made the decision to consider my personal subjectivity as someone who is of Asian/American descent, watching these Asian/American characters being depicted on screen. While content analysis is generally supposed to be objective, it is conducted by people with their own subjectivities, and I wanted to honor that as well as the numbers put forward by this study.
[2] The list was curated on February 2020 through Wikipedia, thus excluding series like The Babysitter’s Club (2020) and Never Have I Ever (2020), which came out later. As this study focused on Asian/Americans, we excluded K-Dramas and other foreign-produced Netflix Originals. Considering the limited scope of this study, and how reality television and nonfiction shows use less of a traditional “narrative” than that of fictional live-action comedies and dramas, they were excluded. As children’s programs are created with different audiences in mind, they were also excluded.
[3] For the sake of this study, we tried to be mindful of specific ethnicities within the labels of East Asian and South Asian. However, as American hegemony tends to lump all Asians together, this was not always possible.
[4] The concept of narrative role was drawn from Igartua et al. (2014). A narrative role is the role the character plays in the story—e.g. whether they were a protagonist (the hero of the story, who the audience is meant to sympathize with), the antagonist (the villain, who works against the hero’s aims), a secondary or tertiary protagonist/antagonist, or a combination thereof.