A TV crew asked 100 people on the street to pick a new Suica mascot. The screen showed A with 44, B with 32, C with 19, and 5 other answers. That raises a better question than "who won this tiny poll?"
If an online vote got 1 million votes and a street survey asked only 50 people, which result would be worth more?
A statistics class would say, "Random sampling matters." Marketing makes it more interesting. Someone who goes looking for a voting page is a statistical headache, sure. But that person also cared enough to react to the product or the character.
The Suica vote has another twist. These designs were not thrown online by random people. JR East's selection committee narrowed the field to three official finalists before public voting began. So candidate B is not some creature the internet smuggled in.
For fun, this article calls the purple rabbit, B, "B-kun (Poison)". That is not an official name. Nobody has tested anything for toxins.
1. The real issue is sample size, not "the denominator"
People often say "the denominator is too small." What actually changes here is the sample size, meaning how many people you really asked.
Assume a simple random sample. Near 50%, where the result is least certain, the rough 95% margin of error is about ±14 percentage points for 50 people, ±10 for 100, ±3 for 1,000, and ±0.1 for 1 million.
So if two surveys use the same method on the same group of people, the bigger one is far more precise.
The catch is that phrase: "the same method."
A street poll of 100 might come from one spot, one time of day, one weekday, and only people willing to stop and talk. An online vote with 1 million entries might include only people who cared enough to find the page and click.
A bigger sample shrinks random noise. It does not wipe out bias in who gets into the sample.
2. 1 million votes can be "very precise, and very wrong"
Say the real market is split 50-50 between fans of A and fans of B.
Now say A fans are fired up and 20% of them vote online, while B fans don't care much and only 1% vote. With 50 million people in each group, that gives 10 million votes for A and 500,000 for B.
The result looks like roughly 95% for A and 5% for B.
That is 10.5 million votes. The sample is huge.
But it does not show what everyone prefers. It shows who is willing to vote.
This is the trap of self-selection. Your preference can change both what you pick and whether you show up at all.
A bigger sample cuts random swings, so the numbers look steady from one run to the next. But if the way people get in is biased, you get numbers packed tightly around the wrong answer.
Pew Research Center has shown this with opt-in online surveys. Going from 2,000 to 8,000 respondents made the estimates much steadier, but it barely reduced the bias compared with outside benchmarks.
More data helps. It is not holy water that cleans up a bad sample.
3. In marketing, the "biased" people may be exactly who you want
Now the argument flips.
If you want to know what everyone would choose if they all saw the options equally, a well-designed random sample is the right tool.
But a company may care more about people who:
- visit a campaign page on their own,
- already use Suica or notice it,
- talk about characters online,
- buy character merchandise,
- join campaigns,
- and might help a new character catch on.
For them, the self-selected crowd is not just a mess in the data. It can be a segment that really matters.
The mistake is not using an opt-in vote. The mistake is calling its result "what everyone thinks."
The opposite mistake exists too. Mix in huge numbers of people who don't care at all, and you can drown out the signal from the people who will actually join in.
Marketing doesn't need "zero bias." It needs the right people for the decision in front of you.
4. The Suica vote is not an anything-goes popularity contest
JR East's official process matters here.
According to JR East, the project first gathered ideas from inside the JR East Group, young creators drew up the first concepts, and a selection committee picked three designs in August 2026. Public voting runs from September 8 to September 23 at 23:59. After invalid votes are checked, the design with the most votes becomes the new mascot.
A is an orange squirrel. B is a purple rabbit meant to suggest leaping into the future. C is an orange creature carrying a mole in a pouch.
So "B-kun (Poison)" is an official finalist that passed the selection committee.
That changes how to read the vote. It works in two steps:
company and creative screening → public vote
People are not choosing from every design in the world. The company decided what was acceptable first, and the vote measures how people react inside that range.
B-kun is not wild poison found behind the station. He is poison that passed inspection. Just a joke, of course.
5. Street polls and big online votes answer different questions
The street result of A 44, B 32, C 19 still has value.
If many of those people had not followed the campaign, it may show the first impression of a wide, casual audience.
The official online vote shows something else. Among people who heard about the campaign, reached the page, and actually voted, which finalist did they pick?
These two don't have to compete.
A street poll leans toward first impressions from a wide crowd. An online vote leans toward choices made by people who took action.
If the real goal is "get more people to use Suica," you need other numbers too: awareness, liking, intent to use, campaign participation, merchandise sales, social sharing, and real usage.
A vote shows what people like. It does not directly show who will use Suica more.
Votes are one metric. They are not the final boss.
6. Why B-kun is interesting for marketing
When the three finalists were revealed, many people online loudly asked for the Suica penguin back. But fan art showed up quickly, and news reports quoted comments like "they're starting to look cute" and "just adopt all three."
That pattern is a big deal for character marketing.
A character people love at first sight is strong. But another kind is strong too. People first say "what is that thing?", then give it a nickname, draw it, turn it into a meme, and slowly get used to it.
Seeing something again and again can make you like it more, up to a point. This is called the mere-exposure effect. A meta-analysis of 81 articles found a mostly positive effect, but also an inverted-U shape. Repeating something does not make people love it forever.
The lesson is not "design something people hate first."
The better lesson is this: a character people keep talking about creates its own repeat exposure.
If nobody reacts, nobody complains. But nobody draws fan art either.
7. Did Myaku-Myaku save the Expo because people hated it at first?
The obvious comparison is Myaku-Myaku, the official character of the Osaka-Kansai Expo 2025.
When it first appeared, many people found it strange and said so. Then events, merchandise, social media, and playful remixes helped more people grow fond of it. News coverage of the Suica finalists has directly compared today's shift from rejection to affection with Myaku-Myaku.
The Expo itself ended with big numbers. A 2026 review by Japan's Ministry of Economy, Trade and Industry reported about 29.02 million visitors over 184 days, an operating surplus of up to ¥37 billion, and an estimated economic impact of about ¥3.6 trillion.
But don't stretch the cause and effect.
Nobody has proved that early dislike of Myaku-Myaku made the Expo a success.
Exhibits, events, transport, tickets, operations, the season, media coverage, and word of mouth all played a part.
Still, for the character alone, this path is plausible:
strange look → people talk → seen again and again → memes and play → people want one → affection.
Backlash is not guaranteed fuel for success. But sometimes it works as friction that builds awareness.
In game terms, Myaku-Myaku may have had a passive skill that turns early hate into experience points. Just a metaphor.
8. As of the night of September 23, 2026, there is no official "current leader"
JR East's official page says voting runs until September 23 at 23:59, each person gets one vote, invalid votes will be checked, and the design with the most valid votes will be chosen. The new mascot is due to be announced on November 18.
As of when this article was researched, JR East has not published live vote counts or an official interim ranking.
So "1 million online votes vs. 50 street interviews" is a thought experiment about what different kinds of data are worth. It does not mean the Suica vote has reached 1 million votes.
What you can see online is a mood shift from early pushback to fan art and support for more than one design. You can't turn that into "B is in the lead."
The election results for B-kun are still waiting on the oracle.
9. Bottom line: a vote's value depends on who you want to understand
1 million responses are not automatically better than 50. A well-designed random sample of 50 can represent the whole public better than 1 million self-selected clicks.
But self-selected votes are not automatically low in value either. If the decision is about your most engaged users, a huge opt-in vote can be very direct evidence.
A simple way to think about it:
data value ≈ sample size × fit with the target group × quality of selection × closeness to real behavior
Do you want to know about the whole public? Current users? Possible new users? Fans willing to vote? Or which design actually changes how people use and buy?
Until you decide, "1 million vs. 50" is just two numbers fighting like monsters.
For Suica, the company screened three official finalists first and then asked the public to choose. So the official vote is better read not as a national opinion poll but as a way to find out which approved finalist gets the strongest response from the people who take part.
Will A win, will C surge, or will purple B-kun take it all on suspiciously strong rabbit legs? We don't know yet.
That's as far as statistics can honestly take us.
