Lying awake at night, feeling you are not trying hard enough, wondering whether something is wrong with you.
That feeling is real. But it is only half. The other half sits outside your character: the year you were born, when you entered your field, and how that field distributes its risks.
Being able to see both halves at once is what C. Wright Mills, in 1959, called the sociological imagination. It is more or less the entire reason this discipline exists.
Mills's example is plain enough to be arithmetic.
In a city of 100,000 people, one man is unemployed. That is his trouble — perhaps he is lazy, perhaps unlucky; the thing to study is this man. But when a nation with 50 million employed has 15 million out of work, the route through personal character is closed: no opportunity structure can be wrecked by 15 million characters at once. Now the thing to study is the economy itself.
Marriage is the same. One couple who cannot make it work is about those two people; but if a substantial fraction of marriages break up in the first few years, that is a structural problem in marriage, employment and housing — even though every couple involved experiences only "we were not compatible".
Mills's sharpest point is not the grand claim that structure determines everything. It is this: misreading an institutional problem as a personal one is a systematic and predictable illusion — because you can feel a trouble, and you cannot feel an institution.
Why is the illusion so stable? Because it is a self-reinforcing loop. Institutions devolve risk onto individuals → each person sees only their own share and explains it by character → people who explain it by character do not go looking for others in the same position, so no collective expression forms → the institution feels no pressure and carries on devolving. Privatising the attribution is the institution's own pressure valve. There is no conspiracy here; it is assembled out of steps that are each locally reasonable.
The clearest case today is burnout. It is usually handled as a problem of personal resilience, with meditation apps and time-management courses attached; but if an organisation's burnout rate stays high for years, that is a problem of rostering, headcount and review cycles — the difference being that the latter costs budget and the former costs one link.
Be clear about evidence: this is a conceptual framework, not an empirical finding, and it cannot be falsified in itself. Its value is a decision rule — how common is this situation, and who does it concentrate on — and that rule can be checked against data. The criticism lands here too: Mills never gave an operational threshold for when a trouble becomes an issue. The reverse abuse is just as bad — if every pain is translated into a structural problem, agency disappears. The usable reading keeps both scales open at once.
The same family upheaval, landing on a seven-year-old and on a seventeen-year-old, grows into two different lives.
The person who turned that sentiment into something measurable was Glen Elder. He followed a group of children born in Oakland, California in 1920 and 1921 — adolescents when the Depression arrived.
Elder found the consequences of family economic loss were not evenly spread. The older children were pushed into odd jobs and into supporting the household, entering adult roles early, and many of them did not do badly at all in later career terms; those who were young when the loss struck, during full dependence, carried the damage longer.
The finding matters less than the methodological turn: Elder made "a historical event" stop being a backdrop and become a variable that allocates consequences by age.
The mechanism has a name: the cohort effect — a cohort being the people born around the same time who therefore meet the same events at the same point in their lives. A historical shock does not land evenly; it is distributed according to the life stage you happened to be in. Two reasons: the same shock triggers different roles at different stages (going out to work, or dropping out of school); and early states lock in later paths — a starting salary sets the baseline for negotiating the next one, and your first industry decides who you will later meet. Small initial differences are amplified by path dependence into large lifelong ones.
The most direct contemporary echo is the luck of your graduation year: several economic studies find that graduates entering the labour market in a recession start on lower salaries and the gap takes many years to close — the same ability, differing only in the moment of entry, and yet almost always assessed as personal ability.
On evidence: Elder's was a longitudinal cohort study, long-running and rich in material, but with only a hundred-odd subjects, in one place, and not experimental, so causal inference is limited; its core idea, however, has been confirmed repeatedly since. There is also a harder problem: age, period and cohort cannot be separately identified mathematically (fix any two and the third follows), so "is this a generational difference or just people getting older" often cannot be cleanly settled. Whenever you meet a claim about what "a generation" is like, that is the first question to ask.
Mills said we are prone to misreading issues as troubles. Half a century later Ulrich Beck said something stronger: modern institutions actively require you to handle issues as troubles.
The risks of education, employment, pensions and care are devolved layer by layer onto individuals, and you are issued a full vocabulary of self-management to go with them. Beck's formulation is precise: people are required to find biographical solutions to systemic contradictions.
One quantifiable facet. Chetty and colleagues used United States tax records to compute absolute mobility — the share of people earning more at thirty than their parents did at the same age. For the 1940 birth cohort it was around 90%; for the 1980 cohort it had fallen to around 50%. "Doing better than your parents" went from something that happened by default to a coin flip.
On evidence: this is descriptive statistics from large-scale administrative data, far more reliable than survey work; but the definitions are genuinely contested (how to adjust for household size, whether to count post-tax transfers and non-cash benefits), and the size of the fall moves under different settings. The direction is robust; do not treat the numbers as a precise scale.
Two gears drive this. The first is risk devolution: institutions move uncertainty off the collective ledger onto individual ones — lifetime employment becomes project work, defined-benefit pensions become individual accounts. The second is a self-optimisation industry: once risk is devolved, a large supply side appears selling courses, methods and mindsets, and it objectively recodes structural problems as skill problems. Once the two mesh, "this is your problem" is no longer merely an illusion; it is an explanation somebody makes a living from.
Parenting is the clearest example: childcare provision, working hours and housing costs are structural variables, yet nearly all the information parents receive daily is about themselves — whether the method is right, whether there is enough time together, whether they are calm enough. Structure is not negotiable and method can be purchased, so all the energy flows to method.
The criticism is due as well: as a grand theory, individualization has uneven empirical support — welfare regimes differ enormously between countries, and treating it as one global trend hides those differences. Conversely, attributing everything to structure is just as easy: a structural explanation lets you see your situation clearly, and seeing clearly is not the same as the situation changing.