Four dimensions of background, and every protected characteristic reported beside them.
An eight minute questionnaire in five sections, with a decline option on every question.
001
The scored dimensions
( Instrument / Standard )
standard measures
These four follow the Social Mobility Commission's recommended employer measurement standard, now carried into the Government Analysis Function harmonised standard. None of them is a protected characteristic under the Equality Act 2010, which is what makes the resulting index lawful to use in workforce policy.
Occupational background
The single most informative question in the instrument, and the primary measure the government standard recommends: what the main household earner did for a living when the respondent was about fourteen.
How it is assessed.Respondents select from the standard occupational categories used across UK social mobility measurement, running from traditional professional and senior managerial roles, through clerical, intermediate and small business ownership, to technical, craft, routine manual and service occupations, with a category for long term unemployment. Because the wording matches the government instrument, results are comparable with any other employer using the same standard.
Age fourteen is used deliberately. It sits before most people's own decisions begin to affect their trajectory, and it is late enough that respondents reliably remember it.
Educational background
The qualifications held by the people who raised the respondent, which shape expectations, networks and confidence long before any application form.
How it is assessed.Two questions capture the highest qualification reached by each parent or guardian, from no formal qualifications through to postgraduate study. We do not ask respondents to identify parents by sex, because the distinction adds nothing to the estimate and imposes an assumption about household structure that many people do not share. First in family to reach higher education is derived rather than asked.
Material circumstances
What the household actually had, as distinct from what it did for work. Two families in the same occupational category can live very differently.
How it is assessed.Three questions: eligibility for free school meals at any point during schooling, which is the standard's recommended secondary measure and a well validated marker of childhood poverty; the housing the family lived in for most of the respondent's childhood, separating owned, private rented and social housing; and a self assessment of how the household managed financially.
Childhood social housing is included because it carries independent signal. In the 1958 birth cohort it significantly predicted later unemployment, holding other factors constant.
Educational context
Not what someone achieved at school, but the conditions they achieved it in.
How it is assessed.The type of school mainly attended between eleven and sixteen, and the typical class size. School type separates independent, selective state and non selective state, with a distinct category for schooling outside the UK.
One detail worth noticing. Independent schooling on a full bursary is its own category, scored differently from independent schooling on full fees. Someone educated at a fee paying school on a scholarship had the institutional advantage without the household resources that normally accompany it. Treating the two identically would misclassify a real and growing population, and most instruments do exactly that.
002
Collected, reported, and never scored
( Demographics / Cross-cuts )
religion, disability
This is the design decision that makes everything else possible.
PRISM collects the standard equality monitoring characteristics as an optional fifth section, using the Office for National Statistics harmonised categories your organisation almost certainly already uses. Sex, ethnic group, disability or long term health condition, age band, region of upbringing, country of birth, caring responsibilities and relationship status.
How they are assessed.Every question is optional at question level rather than merely at section level. They sit behind a clear statement that they play no part in the background score, are held separately from the scored responses, and are linked only by a pseudonymous identifier.
Use one
Reporting segments
The background premium is reported separately for women and men, by ethnic group and by disability status, so you can see whether background effects compound with other characteristics.
Use two
Statistical controls
Where group sizes allow, these variables enter the models as controls so the background estimate is not confounded by something else.
Use three
Non-response checking
Comparing who answered with who did not, so a finding is never quietly built on an unrepresentative half of the workforce.
The research behind PRISM did estimate effects for demographic characteristics, and those effects are real. We report them as aggregate cross-cuts. We do not put them in anyone's score, because an index that weights sex or ethnicity and then informs pay policy would expose the employer using it, whatever the intention behind it.
003
Five sections, eight minutes
( The questionnaire )
scoring rules
Short enough to complete in a tea break, and written so that nothing in it feels like an interrogation. Every question is about parents, childhood and schooling rather than the person's current life, finances or beliefs.
| Section | Covers | Scored |
|---|---|---|
| A | Occupational background of the main household earner | Yes |
| B | Parental qualifications and first in family | Yes |
| C | Free school meals, housing tenure and household finances | Yes |
| D | School type and class size | Yes |
| E | Optional equality monitoring | Never |
What happens when someone declines
A missing answer is treated as missing, and never quietly filled in with a middle value.
Assigning a mid point to someone who declined would silently give them an advantage or a penalty they did not choose, and would break the promise made at the top of the survey. Where enough of the instrument is answered the remaining questions carry the estimate. Where too little is answered no score is produced at all, and the record goes into the non-response report instead.
How results are grouped
Scores resolve into four bands, labelled using the same terminology as the government standard rather than any evaluative language of our own.
These labels get applied to your own colleagues and will be read by managers, which is why they describe origin rather than rank.
004
What we deliberately do not ask
( Omissions )
excluded
Several things the research found predictive are absent from the instrument. Each omission is a choice, and each has a reason.
Birth circumstances
Health adjacent, invasive in an employment setting, and poorly recalled by adults in any case.
Physical fitness and appearance
The 1958 cohort study found physical attributes predicted both income and unemployment. Collecting that from employees engages disability discrimination risk directly, and neither construct can be measured reliably by self report.
Personality
A single agree or disagree statement per trait is not reliable enough to score anybody on, and a properly validated inventory would not fit inside eight minutes.
Number of siblings
Carries almost no independent signal. It adds noise and a question nobody enjoys answering.
In every case the trade is the same: a small gain in predictive accuracy against a real cost in respondent trust, legal exposure or measurement validity. Response rates on a voluntary survey depend on people believing the instrument is proportionate, and these are the questions that cause a survey to be refused.
The full technical specification is available to clients under NDA.