HCI 201 Β· Data Analysis Project

Under the Sea &
Under the Microscope

What 2,000+ Letterboxd reviews of Disney's The Little Mermaid (2023) taught us about race, representation, and the limits of star ratings.

πŸŽ“ Team Roles Β· 5 researchers🐚 HCI 201, Intro to HCI Methods
~ ~ ~
The Setup

We went fishing for racism in movie reviews

When Disney cast Halle Bailey as Ariel in their 2023 live-action remake, the internet had opinions. Loud ones. Our team dove into Letterboxd, a beloved movie journaling app , to study those opinions up close.

Our research question: In what ways are people expressing explicit or implicit racism in reaction to watching The Little Mermaid (2023)? We collected reviews across the full star rating spectrum, each team member covering a different range, from 0.5 stars all the way to 5.

The Method

Qualitative analysis, collaboratively

Letterboxd reviews are a rich mix of short journal entries and numerical star ratings, which made them perfect for our study. We analyzed the text qualitatively, grouping reviews into thematic "buckets" using FigJam as our collaborative workspace.

We weren't just counting racial slurs (though yes, some appeared). We were looking for subtler patterns, the microaggressions, the coded language, the "I'm not racist but..." framing.

πŸŽ‰

Celebration of representation

Reviews that celebrated the emotional impact on Black and Brown girls seeing themselves in Ariel.

🀷

Race acknowledged, dismissed

Reviews that noted the casting but claimed race played no role in their (often negative) opinion.

😀

"Woke agenda" framing

Explicit backlash using coded language around "political correctness" or "forced diversity."

The Unexpected Finding

Stars don't tell the whole story

⭐ , β˜…β˜…β˜…β˜…β˜…

Low ratings β‰  racist reviews

We assumed the lowest-rated reviews would be the most racially charged. We were wrong. Many 1-star reviews praised the casting enthusiastically, and criticized the film for terrible CGI or weak writing instead.

This was our biggest takeaway: quantitative data (star ratings) doesn't capture the texture of qualitative sentiment. A 1-star review and a 5-star review could share identical feelings about representation, while diverging completely on the film's production quality.

It's a classic HCI lesson about the gap between the metric we can measure and the thing we actually care about.

From the Data, Real Reviews

Voices from Letterboxd

Here's a sample of what we actually read, reviews that capture each of our three buckets. Some made us cringe, some made us tear up a little.

Explicit negative / racially coded
β˜…Β½Nobita NobiExplicit

"If mermaids can be of any race, humans can also be any race… Let's cast Ryan Gosling to play Black Panther. Please don't tell me they can't find a red headed white actress who can sing as good as Halle. So it's not about talent/merit."

β˜…Β½SolidSnake31Explicit

"I don't like this film. I don't like Ariel. I don't like the plot changes. I don't like modern Disney. And I don't like 'the message.'"

β˜…Β½Hatrick87Explicit

"They better make Tiana white."

β˜…β˜…anonymousExplicit

"Can we stop being 'woke' and just stay w the classic story? 'Ur just racist' yeah sure ok sorry I like the classic story and feel like it didn't need to be messed with."

β˜…β˜…anonymousImplicit

"They were halfway there with the casting she does look like a fish irl she just ain the right skintone."

Race acknowledged, dismissed
β˜…Belle ForgerExplicit

"This is a disclaimer… I'm not a racist and I didn't hate this movie because the lead was of color… I do actually watch a lot of films with a whole lot of diversity."

β˜…β˜…Β½anonymousExplicit

"Yes, Ariel is black. What a concept. How brave. But Prince Eric is super white even though he has a black mother because he was adopted. Cause you can't have THAT love story."

β˜…Β½anonymousImplicit

"I'm not racist but ts wasn't good, watched with my sister."

β˜…β˜…anonymousExplicit/Implicit

"it's not about the skin tone (actually, that's the least important thing, but if you put it together with everything else thats wrong about the movie, it's a little annoying)... why the f* is Eric's mother black while he is 100% white?"

Celebration of representation
β˜…β˜…β˜…β˜…β˜…jodieExplicit

"as a little girl i was obsessed with mermaids, so seeing someone who looks like me play one delighted my inner child!! and this film is going to mean so much to so many little black girls out there."

β˜…β˜…β˜…β˜…β˜…anonymousExplicit

"I'm a black woman whose favorite princess has always been Ariel, since I was itty bitty. So imagine my absolute, sheer delight when Disney announced their next Ariel would be… A BLACK WOMAN!!!!!!!!"

β˜…β˜…β˜…β˜…Β½anonymousExplicit

"I love watching a black mermaid and I am so thankful that it opening up the dialogue of black people and black kids/girls in aquatic spaces. This really healed my inner child."

β˜…β˜…β˜…β˜…emanuelbuExplicit

"to those hate-reviewing y'all seem to forget how important it is for black kids to finally have REPRESENTATION."

β˜…β˜…β˜…Β½MomoExplicit

"Fuck the racist! She was made for this role!"

⚑ The surprising part

Many of the 1-star reviews above actually praised Halle Bailey's performance and called race a positive, while tanking the score over CGI or writing. Star rating β‰  racial sentiment. That's the whole finding.

"The star ratings were not always directly aligned with users' opinions on race."

, Our key finding, HCI 201 Data Analysis

What I Learned

HCI is about listening to data

Doing this project reinforced something I want to carry into every UX project: don't let your hypothesis blind you to what the data is actually saying. We went in expecting a clean correlation between low ratings and racist sentiment. The data pushed back.

It also made me think deeply about the limitations of any review platform's rating system. A single number flattens nuance. When your "users" are expressing something as complex as racial bias, you need mixed-methods, qualitative and quantitative, to even begin to understand what's going on.

And honestly? Seeing reviews from Black women and girls describing crying tears of joy watching Halle Bailey sing "Part of Your World"? That was the most powerful data point of all.

The Team

Group 1

Five researchers, each reviewing a different band of the rating spectrum, then synthesizing together.

Eesha Gupta β€” 1.5–2β˜… reviews + presentation
Eric Khuu β€” 2.5–3β˜… reviews + Figjam
Tereese Bangayan β€” 4.5–5β˜… reviews + presenter
Lance Hansen β€” 0.5–1β˜… reviews + dataset description
Tanvi Reddy β€” 3.5–4β˜… reviews + research questions
🐚

HCI 201 Β· Introduction to HCI Methods Β· Group 1

← Back to Playground