What a Skill Vector Can See
This chapter examines what a skill vector means when it is used to describe an occupation. The Bureau of Labor Statistics’ Public Skills Data workbook provides the concrete case: it records broad skill requirements for U.S. occupations alongside projected employment, while the OECD’s adult-skills evidence helps distinguish occupation requirements from individual capabilities. The practical goal is to read a profile as a pattern of requirements, understand what its numbers can illuminate, and recognise what they cannot establish about a worker.
An occupation is a pattern across dimensions
A title compresses many kinds of work into a short label. A vector unpacks some of that compression by representing an occupation as an ordered set of coordinates. If the coordinates are kept in the same order for every occupation, each role can be compared with another role dimension by dimension. The result is not a complete description of work. It is a consistent surface on which similarities and gaps become visible.
The BLS workbook contains 832 six-digit occupation codes and 17 broad EP skill dimensions. They include capabilities such as adaptability, computers and information technology, creativity and innovation, critical and analytical thinking, customer service, detail orientation, interpersonal work, leadership, mathematics, problem solving and decision making, project management, science, speaking and listening, and writing and reading. It also includes dimensions that are easier to overlook when jobs are described in prestige-oriented language: fine motor work, mechanical capability, and physical strength and stamina.
That breadth changes the meaning of “transferable skill.” Transfer is not limited to a technical tool. A person may carry forward careful attention to detail, structured problem solving, customer interaction, physical coordination or the ability to organise work. A target role may preserve some of these coordinates while demanding more of others. A title-to-title comparison cannot show that shape; a profile can.
The dimensions are not all the same kind of thing, and the workbook does not claim that they are. “Mathematics” and “critical and analytical thinking” point toward different aspects of cognitive work. “Customer service” and “speaking and listening” overlap but are not identical: one describes the service relationship, while the other describes a mode of communication used across many settings. “Fine motor” and “physical strength and stamina” capture different physical demands. Keeping them as separate coordinates prevents a broad word such as “hands-on” or “technical” from hiding an important difference.
For a reader, the useful mental model is a row of dials rather than a single ranking. The dials do not say that one occupation is better than another. They show where the work places more or less importance. An occupation with a high detail-oriented score and a moderate mathematics score is asking a different question of a worker from one with high mathematics and analytical scores, even if both require computers and careful decisions. A person's career history may contain evidence for several of those dials, but the profile alone cannot tell us how much of that evidence they possess.
The workbook's structure makes this distinction precise. It records an occupation title and code, employment in 2024 and projected employment in 2034, numerical and percentage employment change, an EP skills identifier and title, and an EP skills score. It also records the ONET-SOC code, ONET element identifier and name, and O*NET data value that sit beneath the broader EP skill. The employment fields and the skill fields therefore answer different questions. One concerns the occupation's expected size and direction; the other concerns which capabilities matter in the occupation.
The EP skill score is a requirement signal. The workbook's field description places it on a scale from 1.0, “Not important,” to 5.0, “Extremely important,” and defines it as calculated from related O*NET elements. A score of 4.0 for writing and reading means that this dimension is important to the occupation. It does not mean that a worker in the occupation has achieved 4.0 out of 5, that every task requires the same level, or that a candidate with a lower self-rating cannot learn the work.
The O*NET elements also explain why the workbook contains many more rows than there are visible skill dimensions. One occupation can have several underlying elements associated with a single EP skill. In the retained data, the occupation-level EP score remains constant across those related elements. Counting every row as a new coordinate would inflate the apparent size of the vector and give extra weight to dimensions simply because they have more underlying elements. The correct reading is one occupation-level coordinate per EP skill title, with the elements providing the basis for that aggregate.
A profile can therefore be written conceptually as: occupation identity, seventeen requirement coordinates, and a separate employment outlook. The profile is more informative than a title because it exposes a pattern. It is still narrower than the job itself because it does not tell us the exact tasks, tools, work setting, supervision, schedule, licence, local hiring conditions or quality of evidence an employer will accept.
The distinction between requirement and capability is not a technical footnote. It is the point at which many career comparisons go wrong. Suppose a target occupation places strong emphasis on problem solving and decision making. A person who has handled unusual customer cases, diagnosed process failures or coordinated a response to an unexpected event may have relevant evidence even if their previous title did not contain the word “analyst.” But the data do not prove that the person has mastered the target domain. They tell us what to investigate: which past actions demonstrate the dimension, which target tasks demand it, and what additional practice or knowledge is missing.
The same caution applies to a low score. If an occupation places less emphasis on mathematics, that does not mean mathematics is irrelevant to every task or that the role offers no route into a more quantitative destination. It means only that mathematics is less important on this broad occupation-level measure. A transition decision needs the actual target tasks and a reason to weight one coordinate more heavily than another.
The relationship between a worker and this profile can be represented as a small data model:
The diagram's asymmetry matters. The occupation owns the requirement coordinates and the outlook. The person supplies a different kind of evidence. The two can be compared, but they should not be merged into one score before the comparison has even begun.
Requirement profiles are not personal proficiency tests
The OECD report supplies a second measurement layer that is useful precisely because it is different. It treats literacy, numeracy and adaptive problem solving as core information-processing skills. Literacy concerns accessing, understanding, evaluating and reflecting on written texts. Numeracy concerns using and reasoning critically with mathematical content. Adaptive problem solving concerns defining a problem, searching for information, applying a solution in changing environments and monitoring progress. The report also considers willingness to delay gratification and social and emotional skills such as conscientiousness, emotional stability and open-mindedness.
These constructs help explain the foundations on which people acquire specialised skills and navigate changing work. They are not alternate labels for the BLS dimensions. The BLS “Computers and information technology” score describes the importance of that capability to an occupation. The OECD adaptive-problem-solving measure describes how an adult performed on a standardised assessment under low-stakes conditions. One is a demand-side profile; the other is a person-level observation. A number from one cannot be substituted for a number from the other.
The OECD is unusually clear about what its adult assessment captures. The assessment is untimed and low stakes, and it is administered by trained interviewers. Scores represent typical performance under routine conditions rather than the maximum performance a person might display in a high-pressure job search, a selective educational setting or a consequential work task. That makes the assessment valuable for understanding broad distributions and learning needs, but it also prevents a simplistic interpretation such as “this score is the person's fixed capacity.” Motivation, context and familiarity with the assessment affect observed performance.
The distinction is even sharper for social and emotional skills. In the OECD report, these measures are based on self-report and are context-dependent. A higher level of one behavioural tendency is not always better: agreeableness can support collaboration but make it harder to assert a decision in a conflict, while persistence can help with long projects but hinder flexibility when conditions change. The report does not establish an absolute cross-country scale for these measures. They can describe relative tendencies within a context; they should not be used as universal personality requirements for an occupation.
This gives the reader a reliable rule for interpreting a vector. First ask what kind of observation produced the number. Then ask whose perspective it represents. Finally ask what decision the number can legitimately inform. A BLS requirement score can help compare the broad demands of two occupations. An OECD proficiency measure can help explain a population's foundation skills and the importance of learning opportunity. Neither one alone can certify an individual for a role.
There is a common illusion of competence here: recognising a label can feel like possessing the capability behind it. Someone may recognise “adaptive problem solving” and “cybersecurity” as related ideas without being able to investigate an incident, explain a threat model or document a decision. Likewise, someone may have performed the underlying work without knowing the occupational vocabulary used to describe it. The vector is useful because it prompts a search for observable evidence, not because it replaces that search.
A practical comparison therefore uses the occupation profile to generate questions about the person. If the target profile emphasises writing and reading, what documents, explanations or records has the person produced, and for whom? If it emphasises mathematics, what quantities, uncertainty or trade-offs have they reasoned through? If it emphasises interpersonal work, how have they handled a confused customer, a disagreement or a handoff? If it emphasises adaptability, what happened when the original procedure stopped working? The answers should be concrete enough for another person to inspect.
The profile also generates questions about the target occupation. Which of the broad dimensions are critical to actual tasks rather than merely present in the aggregate? Which tools or bodies of knowledge sit underneath the broad label? Does the occupation require a licence, a degree, a portfolio, supervised practice, physical capacity or experience in a particular setting? Is demand growing, and is the skill requirement changing quickly? The BLS workbook can supply part of the outlook question, but not all of the entry conditions.
This is why the skill vector should be read as a coordinate system, not as a verdict. It can make an invisible continuity visible: a person who has resolved exceptions, interpreted records and communicated decisions may have a stronger bridge into another role than their title suggests. It can also expose a hidden gap: two roles may both involve computers, but one may place substantially more weight on analytical reasoning, mathematics or written documentation. The vector says where to look. It does not decide what the person can do.
The next step is to compare two profiles explicitly. That comparison should preserve the distinction established here: the source role and target role each have a requirement pattern, the worker has separate evidence, and employment outlook is a separate variable. Once those layers are kept apart, a difference between roles can become a specific transition question—what must be learned, practised or demonstrated—rather than a vague judgement about whether the titles “match.”