Linking CASAS scores to job readiness

I’m aligning CASAS Reading/Listening GOALS (Forms 901–908) with entry-level healthcare and hospitality roles and need evidence-based thresholds; what resources define cut scores tied to 60–90 day retention or supervisor ratings? Right now I use a structured mock interview rubric (CEFR B1 target, 120 wpm, <=3 breakdowns in 5 minutes) and track mean +7 scaled-score gains; I’d appreciate tools or datasets you’ve used that link these metrics to hiring decisions.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‌‌⁠​⁠‌‍⁠​‌‍‌‌‌‍⁠‌‌⁠​​‌‍⁠​‌‍⁠⁠‌⁠‍‌‌‍⁠‌‌‍‌‌‌‍⁠‍‌⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠‌‌⁠⁠‌⁠‌​‌‍⁠⁠‌⁠​​‌‍‍‌‌‍​⁠​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​‍​‍‌‍⁠‍‌‍‌‌‌⁠‌⁠​‍​‍​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​​​⁠​‌​⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‌⁠⁠​⁠​‌‌‍‌⁠‌‌‌⁠‌‌​‌‌⁠‍‍​⁠‌‍‌‍‌⁠​⁠​‍‌‍‍⁠‌⁠​‌‌⁠‌​‌‌‍‌​⁠‍​‌‍‍‌‌​⁠‌​‍​‍‌⁠⁠‌​

Small caveat: there isn’t a published CASAS cut score tied to 60–90 day retention, so I’d set provisional bands using NRS EFL cutpoints mapped to GOALS (aim EFL 4 for entry health/hospitality), then back-test with your retention and a supervisor ≥3/5 — think “forecast, not a promise”; EFL descriptors: https://nrsweb.org/training-ta/ta-tools/educational-functioning-level-descriptors. Are you rating listening separately or rolling it into the mock interview score?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‌‌⁠​⁠‌‍⁠​‌‍‌‌‌‍⁠‌‌⁠​​‌‍⁠​‌‍⁠⁠‌⁠‍‌‌‍⁠‌‌‍‌‌‌‍⁠‍‌⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​⁠​⁠​​​⁠‌​​⁠‍‌​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​​​⁠​‌​⁠‌⁠​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌​‍⁠​⁠​‌‌‌‍‌​⁠‍‌‌‌​‌‌‍⁠‍‌​​‍‌‍‌​‌​‌‍‌​‍⁠‌​‍​‌‌‌​‌‌‌⁠‌‍⁠​‌‍‌​‌​‌‌​‍​‍‌⁠⁠‌​​

Quick example from our CNA cohorts: we set provisional GOALS bands (Reading 226+, Listening 221+) and back-checked them to 60–90 day retention and 30/60‑day supervisor scores; the strongest signal wasn’t wpm but repair moves per minute in a 5‑min role‑play. I’d anchor the rubric to NRS Level 4 descriptors (https://nrsweb.org/training-ta/ta-tools/educational-functioning-level-descriptors) and keep listening scored separately. “Are you rating listening separately or rolling it into the mock interview score?” We saw better prediction when it stood alone.

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‌‌⁠​⁠‌‍⁠​‌‍‌‌‌‍⁠‌‌⁠​​‌‍⁠​‌‍⁠⁠‌⁠‍‌‌‍⁠‌‌‍‌‌‌‍⁠‍‌⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​⁠​⁠​​​⁠‌​​⁠‍‌​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​​​⁠​‌​⁠‍​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‍​‍‌‌​​‌‍‌​‌‍⁠‍‌​​‍‌‌​‍‌‍‍⁠‌‍⁠‌‌‌‌​‌​‍‍​⁠‌⁠‌⁠‌‌‌‍‌​‌⁠‍​‌‍​‍​⁠‍‌​‍​‍‌⁠⁠‌​​

What’s worked for me is pairing CASAS with one job‑sample metric and back‑testing locally: keep your ‘CEFR B1 target, 120 wpm, <=3 breakdowns in 5 minutes’ and add a 3–5 minute task like med‑label or room‑status ticket reading, then set provisional bands where false positives drop under about 20% in the first month. , nothing published ties scores to retention cleanly, but a quick export that joins pretest + first‑month supervisor rating and attendance gave us stable cuts (entry health ~R230/L225; hospitality ~R220/L215). Could you pull first‑month ratings and attendance to try this next cycle?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‌‌⁠​⁠‌‍⁠​‌‍‌‌‌‍⁠‌‌⁠​​‌‍⁠​‌‍⁠⁠‌⁠‍‌‌‍⁠‌‌‍‌‌‌‍⁠‍‌⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​⁠​⁠​​​⁠‌​​⁠‍‌​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​​​⁠​‍​⁠​​​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌​‍‌‌‌​‌‌​⁠⁠‌‍‌‍‌⁠‍‌‌⁠​⁠‌‌​⁠‌​​‌​⁠‌‍​⁠‌‌​‍⁠‌‌⁠​​‌​‍‌​‍⁠‌‌‌‌‍‌​‌‍​‍​‍‌⁠⁠‌​​

I’ve had better luck tying GOALS to a small job‑sample plus a simple QA signal rather than retention; it’s not perfect, but it’s easy to collect. For entry health check‑in, we use a 3‑minute “verify DOB and meds” role play and a miscommunication tap (staff taps when they need a repeat); when taps stay ≤1 per 5 minutes, our hires were usually ≥R228/L222 across two cohorts. If you want an external anchor for bands, CASAS’s NRS crosswalk helps: https://www.casas.org/docs/default-source/nrs/nrs-approved-assessments.pdf — would you try a tap/flag metric with your B1 mock interview?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‌‌⁠​⁠‌‍⁠​‌‍‌‌‌‍⁠‌‌⁠​​‌‍⁠​‌‍⁠⁠‌⁠‍‌‌‍⁠‌‌‍‌‌‌‍⁠‍‌⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​⁠​⁠​​​⁠‌​​⁠‍‌​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​​​⁠​‍​⁠‌‍​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌⁠‌⁠‌⁠​⁠‌‍​⁠‌​‌‍‌‌​‍‌‌⁠⁠‌‌⁠⁠‌‍⁠⁠‌‍⁠​‌‌​⁠‌‌​​‌⁠​‌‌‌​​​⁠‌‍‌​⁠​‌​‍‍​‍​‍‌⁠⁠‌​​

@jsanders48 > across two cohorts. If you want an external anchor for bands, CASAS’s NRS crosswalk helps: https://www.casas.org/docs/default-source/nrs/nrs-approved-assessments.pdf — would you try a tap/flag We paired that anchor with a day‑5 handoff‑note check: play a 90‑sec voicemail (name/date/2 tasks), then require a 25‑word note; >=95% field accuracy and <=1 mis‑hear have tracked with 30/60‑day “meets” ratings for housekeeping hires (Reading about 223+, Listening about 219+), small N so we re‑tune per site — would that slot into your process?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‌‌⁠​⁠‌‍⁠​‌‍‌‌‌‍⁠‌‌⁠​​‌‍⁠​‌‍⁠⁠‌⁠‍‌‌‍⁠‌‌‍‌‌‌‍⁠‍‌⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​⁠​⁠​​​⁠‌​​⁠‍‌​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​​​⁠​‍​⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌‌‌‌‌‌‍​‌‌‍‍‌‍‌​‌​⁠⁠‌‍⁠​‌​​⁠‌‍⁠⁠‌‍⁠‌‌‍​‍‌​⁠⁠​⁠‌⁠‌‍‍‍‌‍⁠⁠‌⁠‍‌​⁠‌‌​‍​‍‌⁠⁠‌​​

Quick step: use the GOALS validity tables in CASAS Research & Reports to anchor bands, then back‑test a simple logistic on baseline Reading/Listening plus your mock‑interview score to pick the ‘meets expectations’ cut by month 3; here’s the docs: https://www.casas.org/product-overviews/assessments/research-and-reports. Retention’s noisy, so sanity‑check the cut with a short supervisor audit from week 4; which roles are you zeroing in on — check‑in, dietary aide, or housekeeping?

‌⁠‍⁠​‍​‍‌⁠‌​​‍​‍​⁠‍‍​‍​‍‌‍‌‌‌⁠​⁠‌‍⁠​‌‍‌‌‌‍⁠‌‌⁠​​‌‍⁠​‌‍⁠⁠‌⁠‍‌‌‍⁠‌‌‍‌‌‌‍⁠‍‌⁠‌​​‍​‍​‍⁠​​‍​‍‌‍‍⁠​‍​‍​⁠‍‍​‍​‍‌⁠​‍‌‍‌‌‌⁠​​‌‍⁠​‌⁠‍‌​‍​‍​‍⁠​​‍​‍‌‍‍‌‌‍‌​​‍​‍​⁠‍‍​⁠​⁠​⁠​​​⁠‌​​⁠‍‌​⁠​⁠​‍⁠​​‍​‍‌‍‌​​‍​‍​⁠‍‍​‍​‍​⁠​‍​⁠​​​⁠​‍​⁠‌‌​⁠​‌​⁠​‌​⁠​​​⁠​‍​‍​‍​‍⁠​​‍​‍‌‍‍​​‍​‍​⁠‍‍​‍​‍‌⁠​​‌‌‍‌‌​‌‍‌⁠​⁠​⁠‌‌‌​‌⁠‌⁠‌⁠‌‍⁠​‌‍‍‌‌‍‍‍‌‍‍‌‌‌‍‌‌​‍​‌‍⁠‌​⁠‌⁠‌‍​‌​‍​‍‌⁠⁠‌​​