Graduate Employability Skills Assessment with a Portable RecordA score you own, with the error bar printed on it.
Four sections, forty exercises, every figure with its band — and one section that gets no figure at all, because its reliability cannot carry one.
The campus market sells a number the vendor disclaims. This one publishes what it can and refuses what it cannot.
A graduate in India can pay well over a thousand rupees for an employability test whose own provider disclaims the number it gives back. One major test shows employers the most recent score rather than the best one and enforces a gap of weeks before a retake, so a single bad afternoon costs a month and a half. What the student gets for the money is a figure with no error bar, no reliability and no key logic — and then a year of job recommendations attached to it.
This is the honest version and it costs a fraction of that. Forty exercises measure four things every employer in the country tests for and almost nobody reports properly: reading a workplace document for what it actually says rather than what it seems to say, working with the numbers that turn up in an ordinary job, following a written instruction exactly, and writing a short message a reader can act on.
What comes back is a record rather than a verdict. Every section carries its own band, drawn on the page rather than mentioned in a technical manual, with both spans written out in plain words. Every section carries its reliability. And one of the four is deliberately shorter than the others, lands below the threshold, and gets a placement instead of a number — with the arithmetic printed beside it, to three decimals, and a note saying what rounding it for display would have done. That refusal is the point of the product. A report that will not give you a figure it cannot support is the only kind whose other figures mean anything.
The scale is drawn as a modelled distribution built from our own declared answer priors, labelled as modelled with the number of simulated respondents on the same panel, and you are marked on it. No percentile appears anywhere, because a percentile needs a real norm group with a stated size and composition and this instrument does not have one yet. When it does, the curve will be replaced and the first thing to change will be its shape.
There is no pass mark, no placement promise of any kind, and nothing here should be the whole of anybody. It is a record you own and can send to somebody, carrying the key logic, the reliability and the error bands, so that whoever reads it can judge it rather than trust it.
What you walk away with
Each section drawn as a band rather than a point, with the 68 and 95 per cent spans written out where the reliability supports a number at all.
Omega per section to three decimals, and a section that carries a placement instead of a number with the reason beside it.
Built from our own answer priors, labelled as modelled with its simulated sample size, and marked with your score. No percentile is claimed anywhere.
Each with what it gets you and what it costs when it runs unchecked, written so you can hold it without a coach.
Left out, and added. An instruction is failed as completely by extra as by a gap, and a single accuracy figure hides which one you do.
Chosen from your own weakest section or your own lean, naming a specific situation and a specific behaviour.
Inside your report
Illustrative sample — your report is generated from your own responses.
One section is short on purpose. Its reliability cannot carry a number, so it gets a placement and the arithmetic is printed.
| Section | Items | Omega | Permits |
|---|---|---|---|
| Reading a document | 12 | 0.772 | A number with its band |
| Working with numbers | 10 | 0.738 | A number with its band |
| Following instructions | 10 | 0.738 | A number with its band |
| Writing that lands | 8 | 0.693 | A placement, no figure |
Rounded for display, .693 prints as .69 and stays refused. Taking the decision on the rounded value would have promoted it.
The curve is for seeing how crowded the middle is. A percentile needs a real norm group and this instrument does not have one yet.
Built for
- Final-year students and recent graduates preparing for placement season
- Early-career job seekers who want something specific to work on
- Anybody who has paid for an employability score and learned nothing from it
- Career services teams who want a record students can act on rather than a rank
Four sections, forty exercises, every figure with its error bar
40 exercises across five formats · about 40 minutes · a record you own, with published reliability, per-section bands, a modelled curve and no placement promise.
Take free · full report ₹249 (incl. GST)
Frequently asked questions
Because eight exercises cannot support one. That section's reliability estimate comes out below the threshold at which a number is defensible, so it carries a three-way placement and the arithmetic is printed beside it. Printing a precise-looking figure a short section cannot support is the commonest dishonesty in this category.
A percentile needs a real norm group with a stated size and composition, and this instrument does not have one yet. The curve on your report is modelled from our own declared answer priors, is labelled as modelled with its simulated sample size, and is there so you can see how crowded the middle is rather than to rank you.
No, and nothing here implies it will. There is no placement promise, no employer network attached and no pass mark. It is a record of four measurable things with the error around each one printed, and it should sit alongside what you have actually done rather than in place of it.
Yes, with no waiting period. Holding somebody to a single afternoon for weeks is a commercial decision rather than a measurement one, and the band printed on every figure already tells you how much a second sitting would be expected to move.
About forty minutes for forty exercises. ₹249 in India, inclusive of GST, or US$2.99 elsewhere, one-time, for the sitting and the full report. Below ₹300 the sitting is take-free rather than pay-first.
Each one takes a single capability, puts you inside the situations where it is actually tested, and scores your choices against published evidence — with a report designed for that capability alone, not a template. They span hiring, compliance, education, operations and personal skill.
Browse the catalogue →Methodology: Forty original exercises across five formats: fourteen keyed single-choice items, eight select-every-that-applies items, eight true or false claims, five match-the-following exercises and five ordering exercises. Construct statement: it measures four workplace-readiness skills - reading a document for what it states rather than what it implies, working with everyday workplace quantities, following a written instruction exactly, and writing a short message a reader can act on. It does not measure general intelligence, personality, motivation, degree subject knowledge, interview performance, or the likelihood of getting any particular job, and it carries no placement promise of any kind. Declared response instruction: knowledge throughout. Every exercise has an answer that follows from the material given in it. Scoring: a RELIABILITY-WEIGHTED COMPOSITE. Each section is scored as chance-corrected graded credit against an authored prior on every option, so zero on a section means no better than a respondent answering the way people typically answer rather than no better than a coin. McDonald's omega is estimated for each section from its item count at a stated assumed average inter-item correlation, and the weighting decision follows from the spread: with the spread below .15 the composite uses unit weights, and that is stated on the report rather than left implicit. The threshold decision is taken on the unrounded omega, so a section at .69 is not promoted to a number by a rounding function. Any section whose omega falls below .70 carries a three-way placement and no figure, with the arithmetic printed beside it. Every reported figure carries its standard error of measurement, drawn as a band and written out at both the 68 and the 95 per cent span. An unanswered exercise leaves both sides of the ratio rather than scoring zero. The reference distribution on the report is MODELLED from the declared priors, labelled as modelled with its simulated sample size, and no percentile is claimed anywhere. Construct grounding, drawn across sources rather than from one framework: the standard error of measurement and the requirement that a score be reported with its band (AERA, APA and NCME Standards for Educational and Psychological Testing, 2014); McDonald's omega as the reliability estimate to report in place of alpha, because alpha assumes equal factor loadings that real item sets do not satisfy (McDonald, 1999; Dunn, Baguley and Brunsden, 2014); the unreliability of short subscores and the case against reporting them as numbers (Sinharay, Haberman and Puhan, 2007; Haberman, 2008); classical test theory on difficulty and discrimination targets (Crocker and Algina, 1986; Ebel and Frisbie, 1991); item-writing guidelines on plausible distractors and cue control (Haladyna, Downing and Rodriguez, 2002); the base-rate and percentage-change errors that dominate everyday workplace numeracy (Gigerenzer, 2002 on natural frequencies); the reading research on distinguishing what a text states from what a reader infers (Kintsch, 1988); plain-language and front-loading findings in workplace writing (Plain Language Association guidance; Redish, Letting Go of the Words, 2012); and score-report design research on what makes a report interpretable to a non-specialist reader (Zenisky and Hambleton, 2012; Hambleton and Zenisky, 2013). All items are original works. No item, scale name or report section is taken from any commercial instrument, and no affiliation with or endorsement by any vendor, employer, university or examination body is claimed or implied.