Locale: en
1. Bottom line: a bigger salary does not automatically install population-level thinking
Career forums often produce surprisingly high answers to a simple question: “At what income would you call someone a high earner?”
People may answer ¥15 million, ¥20 million, or even more.
That creates an obvious puzzle: if someone can earn that much, how can they fail to notice that the people around them are not a representative sample of society?
Because those are different skills.
A high salary shows that someone is valuable in a particular labor market, company, profession, or role. It does not certify that they always identify the correct statistical population, detect selection bias, or separate a local norm from a national one.
A promotion does not come with the “statistics literacy” expansion pack.
2. National data make ¥10 million clearly unusual
Japan’s National Tax Agency reported an average annual salary of ¥4.87 million among people who worked throughout 2025.[1]
Adding the published salary brackets shows that roughly 6.6% earned more than ¥10 million, about 1.9% earned more than ¥15 million, and about 0.7% earned more than ¥20 million.[2]
So if the population is all year-round private-sector salary earners in Japan, those incomes are plainly uncommon.
Calling ¥15 million “not high” as a statement about the whole country is hard to defend statistically.
But saying “¥15 million does not feel high in my industry” or “it would not make me feel rich” is a different claim.
Much of the confusion comes from treating those claims as interchangeable.
3. Career platforms are selected before the poll even begins
People discussing salaries on a career-oriented platform are not randomly drawn from the population.
There is already a chain of selection:
care about careers
→ care about compensation
→ use a career platform
→ open an income poll
→ choose to comment.
If highly paid comments are then more visible or receive more reactions, the sample a reader sees becomes even less representative.
It is like surveying people at a food festival and concluding that the entire country loves eating out.
Inside the festival, the observation may be perfectly true.
The error begins when the festival becomes “everyone.”
4. Humans compare themselves with nearby people, not a national spreadsheet
Research on income comparison emphasizes the importance of the reference group: colleagues, people in the same occupation, people with similar education, friends, or family can all become the benchmark.[3]
Most people do not begin the morning by checking their percentile among tens of millions of workers.
They look sideways.
A study linking Danish survey responses with administrative income records found systematic misperceptions of people’s positions within different reference groups. Those errors were related to social proximity, how transparent income was, and visible signals of economic status.[4]
So if someone works in a circle where ¥10–20 million incomes are common, “¥10 million is normal” can be an accurate description of their local environment.
Local truth and national truth can coexist.
5. As income rises, the definition of “high” can move away
At ¥5 million, ¥10 million may look extraordinary.
At ¥10 million, the person starts noticing colleagues on ¥12–15 million.
At ¥15 million, ¥20–30 million becomes the next visible tier.
The finish line grows wheels.
Research on income comparisons suggests that people often treat economically relevant peers—and frequently people above them—as especially informative reference points.[3]
This makes it easy for a person to remain psychologically “middle of the pack” even while being statistically near the top of the national distribution.
Top 2% nationally, middle of the team chat.
No contradiction required.
6. “How can someone earn so much and miss this?” is not actually a paradox
The abilities rewarded by the labor market are task-specific.
High pay can reflect technical expertise, sales ability, negotiation, project leadership, scarce experience, organizational responsibility, access to a profitable industry, or simply working for a firm with more money to distribute.
None of those requires a person to say:
“This online sample is self-selected, so we should not extrapolate it to the entire population.”
A brilliant engineer need not be a brilliant survey methodologist. A great salesperson need not think naturally in base rates. A senior executive is not automatically immune to cognitive bias.
Humans do not have one universal “intelligence stat.”
You can be level 80 in your profession and still be carrying starter gear in sampling theory.
7. But answering “¥20 million” is not automatically wrong
There is an important counterpoint.
If the question is “What income feels high to you?”, it is subjective by design.
A respondent might be answering:
- the level at which life finally feels comfortable
- the level that is rare inside their own company
- the income they personally aspire to
- the level that still feels high after taxes, housing, or family costs
In that case, ¥20 million can be a perfectly valid personal threshold.
The reasoning problem appears only when the claim silently changes from:
“high for me”
to
“high in my circle”
to
“high for society as a whole.”
Subjective affluence and statistical rarity are different variables.
8. Four questions that fix most salary arguments
When an income discussion starts drifting, ask:
What is the population?
All workers, the same age group, large-company employees, or the same profession?How did these people enter the sample?
Randomly, or because career- and salary-focused people chose to be there?What does “high” mean?
Rare, comfortable, prestigious, or personally satisfying?Are we exporting a local norm to society?
“Normal at my company” is not the same sentence as “normal in Japan.”
These questions turn a vague status fight into a measurement problem.
9. Conclusion: high income does not include a free “zoom out” function
High earners are often genuinely excellent at something the market values.
That does not mean their mental sample of society is nationally representative.
In fact, success can place them inside increasingly affluent networks, making high incomes more visible and therefore more normal-looking.
Career platforms add another layer of selection through who joins, who answers, who comments, and whose comments get attention.
The world they see is not necessarily false.
It is simply narrow.
The useful question is not “How can this person be so dumb?”
It is:
“Normal relative to which denominator?”
Even at ¥30 million, the denominator does not update itself.
