How scoring works

Every element in Theta — audiometry, otoscopy, tympanometry, acoustic reflexes, the speech tests — is scored the same way underneath. Once you understand the model on this page, each element's own Scoring & Report tab is just a short list of which fields that element exposes. This page is the hub those pages point back to.

The universal model

Scoring always compares two things:

• the designed value — the true answer you set when you built the case (the "answer key"), and
• the submitted value — what the learner recorded while working the case.

For each field that's turned on, Theta checks whether the submitted value matches the designed value. A match counts as correct; a mismatch counts as incorrect. Nothing else feeds the score — not how long the learner took, not how many times they re-tested. (Those details are captured separately for the report's activity trail, but they don't change the number.)

What counts as a "match": exact vs. template

Most elements support two ways of being correct, and many cases use both at once depending on which field is being checked.

Exact-value scoring

The submitted value must equal the designed value (within any tolerance — see below). This is used for concrete measurements: a threshold of 45 dB, a tympanogram peak, a reflex threshold at a given frequency, an SRT in dB.

Template-category scoring

Instead of a single literal number, the field is judged against a category or classification the learner is expected to identify. Examples:

Because elements can be templates (the underlying values regenerate per submission from a chosen category) or fixed (the same values every time), template-category scoring lets you grade the clinical conclusion even when the exact numbers differ from one learner's case to the next. A single element can mix both: e.g. score the exact thresholds on the audiogram and the degree/type/configuration profile category.

The per-field tri-state: Off / Scored / Report-only

You don't score a whole element all-or-nothing. On each element's Scoring & Report tab, every individual field has a small control you click to cycle through three states:

Clicking the control cycles Off → Scored → Report-only → Off. A scored field is always shown on the report — there's no separate "include on report" toggle, which keeps the dense scoring grids uncluttered.

Use Report-only for things you want the instructor and learner to see on the PDF for reference, but that shouldn't move the grade — for example, showing the expected masking levels next to the learner's, without penalising the masking decision itself.

Lenient ("optional") scoring

Some clinically defensible choices can't be judged in foresight — for instance, whether to mask the better cochlea. Theta marks those rows as accepted either way: the learner gets the point regardless, and the report annotates the row with an asterisk and a short reason. You don't configure this; the scorer applies it automatically where it applies.

The clearest case is bone conduction where one cochlea is better than the other by even a single 5 dB step and the poorer ear has been masked: its masked bone threshold proves it worse, so the unmasked line must belong to the better ear. Theta records that better ear as a plain threshold (no masking dilemma), but a learner who tries to mask it, finds no plateau, and writes MD instead is not marked down — the label is optional there, and doesn't count as an invented situation. A true dilemma — where the masker crosses back into the test cochlea before the required plateau can form, as with a symmetric bilateral conductive loss — still requires the MD flag. The plateau is counted from where the learner reaches it: the initial masking level, or the point where the threshold stops climbing if that is higher. So when both cochleae are about equal and the initial masking level (with its safety pad) already sits at the top of the plateau, the point is a dilemma even though a plateau exists below where masking starts.

Masking air where the other ear's bone isn't required

Whether an air-conduction point needs masking depends on the non-test ear's bone threshold at that frequency. When the lesson doesn't require that bone threshold (250 Hz by default, the other ear under "bone in one ear only", or a custom frequency list), Theta judges the masking decision against what the learner actually knew:

• They tested the other ear's bone there (a real threshold — not a masking dilemma, no response, or vibrotactile): masking is judged against bone, the standard clinical rule, and graded strictly. The row carries a † and a note saying why, and the expected audiogram on the results screen and in the PDF draws that point with the symbol for the answer the row expects (masked, where the lesson's own answer was not to mask), so the plot and the row agree.
• They didn't: there's nothing to compare against but air, so where comparing air to air and comparing against bone give different answers, the decision is accepted either way (*). Masking there is the conservative call; skipping it follows the air comparison.

The comparison always uses the patient's true bone threshold, so a mistyped bone value can't change the answer. Above 4 kHz the comparison is always air to air — bone conduction isn't reliable up there. Only the masking decision adapts: the final threshold is still graded against the truth, and a lesson's points available are the same for every learner.

Where the decision is accepted either way, the Effective masking row — and the masking-fault counts in Advanced Analytics that read it (Masking omitted, Masking that was not needed, Masking faults in total, Worst masking fault) — judge the learner's own choice: not masking reads as correct, not as masking omitted, and masking anyway isn't counted as unnecessary. A learner who does mask there is still judged on the masker level, so over- and under-masking still show.

This applies to attempts started since this rule reached Theta. An attempt started before it keeps the lesson's fixed answer at these points — in its grade and in the instructor's threshold-tracing panel — so its score doesn't change when it is reopened or re-scored.

Thresholds entered without testing

On an assessment that asks for it — set on the course item, not the lesson, and so far only on Skill Assessments: Theta's (its exams and practice cases), and an organization's own, set up by Theta (its exam and practice copies, while the organization holds the Skill Assessment) — a final threshold with no testing behind it is marked wrong, whatever its value. The row says why ("no tone was presented at this point before it was recorded").

Its masking is marked wrong too. Nothing earns credit without testing, so the Effective masking row of such a point is wrong whatever the masker did — and so is the masking row of a required point where no threshold was recorded at all. (Without this, "masking not needed, masker left off" would earn a point at every such frequency, even on an attempt where nothing was tested.)

"Testing" is read from the attempt's own record of the tones presented, with the same evidence and the same allowances as the testing confidence instructors see beside an attempt: only tones count (not masker auditions), only threshold saves (not MCL / UCL), an unmasked bone threshold is supported by a search on either mastoid, and a value typed into the saved-values panel to correct a point that was searched counts as tested. A point with a little testing behind it — a single tone, say — is not marked wrong; a point with none, or a value the record never saw, is.

The record is read in the order things happened: work the connection held back and sent late — or sent twice — is read where it belongs, so a late Clear never wipes testing done after it. Clearing one frequency removes only the points the audiometer cleared (the ears and transducers its two channels were set to), and an entry the record can't read is skipped rather than failing the attempt's grade.

The rule is fixed onto each attempt as it starts, so switching it on or off later never regrades an attempt already made. On the same assessments an attempt whose testing confidence is Low doesn't pass, whatever its score (see the exam result in The submission flow).

Technique flags

A lesson in a course can flag how pure-tone thresholds were found — the Hughson-Westlake starting level and steps, the two ascending responses behind a threshold, tone length, coverage, masking — and, if you choose, take points off for them. It is set on the course item (the lesson's settings in the course, beside Practice / Exam): Off, Report only (listed on the results and the report, no points off) or Scored (listed, and each flag's points come off the points the learner earned — 18 of 20 with 2 points of flags reads 16 of 20, never below zero). What each flag checks is described under Audiometry → Technique flags.

Hide flags from learners keeps the list for instructors only; when flags are scored, a learner still sees how many points came off. The flags' settings are fixed onto each attempt as it starts, so changing them later never changes an attempt already made by itself — Re-score applies the current settings to an attempt, so switching flags on for a lesson and re-scoring shows them on earlier attempts too.

Tolerances

Exact-value fields can allow a margin so a near-miss still counts. Tolerances live on the element's Scoring & Report tab where they apply:

Category/template fields don't use a numeric tolerance — the submitted category either matches the designed category or it doesn't. Elements that don't expose a tolerance field score their values exactly.

How an element's score is computed

For one element, Theta:

1. Looks at every field set to Scored (Off and Report-only fields are skipped for counting).
2. Compares each scored field's submitted value to the designed value, applying any tolerance.
3. Counts correct out of total scored fields — that's the element's points. The percentage is correct / total.

Fields set to Report-only are gathered separately so they can be shown on the report without affecting the count. An element with only report-only fields shows as Reference only — it appears on the report but contributes no points.

How the lesson (submission) total is computed

The submission's overall score is simply the sum across all elements: total correct divided by total scored points, expressed as a percentage — rounded to the nearest whole percent, a half rounding up. Elements with nothing scored don't dilute the total — they're not counted. In the instructor's Scoring Summary, you'll see the overall percentage at the top and each element's own correct/total and percentage below, with expandable breakdowns field-by-field. On a lesson whose technique flags are Scored, the overall correct is after the flags' points come off; the Technique card under it says how many.

On an item with a pass mark (a Skill Assessment exam — Theta's, or an organization's own), an attempt passes when that rounded score is at or above the mark and its testing confidence isn't Low. For example, under a 90% mark, 43 of 48 points is 89.6%, which rounds to 90% — a pass. An attempt's verdict is recorded when it is submitted and keeps it; a re-score judges it again by this rule. Advanced Analytics judges a percentage pass rule the same way: on the whole percent its cell shows.

Re-scoring after you change settings

Scores are computed against the scoring settings as they were when the submission was graded (a snapshot). If you later edit what's scored on the lesson, already-graded submissions don't silently change. The Scoring Summary flags this with a "settings changed" notice and a Re-score button, so you choose when to apply the new rules. Re-scoring recomputes against the current settings.

A change to how Theta itself grades — rather than to a setting you chose — applies to attempts started after it. An attempt started before keeps the rule it was graded under, and Re-score doesn't change that: old attempts keep their original grades. The adaptive air-masking decision above, the untested-threshold rule, and the retired per-frequency OAE and acoustic-reflex exclusions, work this way. A lesson's technique flags are a setting you choose, so they follow the scoring-settings rule instead: an attempt keeps the flags it started with until staff re-score it, and a re-score applies the lesson's current flag settings. So does an assessment's pass mark: each attempt keeps the mark it was started under, so changing it later never changes an earlier attempt's pass or fail. (The one exception is how the score is rounded against the mark: since 2026-10-06 a re-score judges an earlier attempt by today's rule — nearest whole percent — though its recorded verdict stands until staff re-score it.) The Skill Assessment's frozen 2026 exam key and the grades of two reference attempts are also pinned by a test, so a future change to the engine can't quietly regrade it.

Where each element's scoring is described

Each element's page lists the exact fields it exposes and any element-specific behaviour. Start there once you've read this model:

• Audiometry — thresholds, PTA, masking work, profile (degree/type/configuration), tolerances, auto-fill.
• Otoscopy — category and referral-needed.
• Tympanometry — tympanogram type.
• Acoustic Reflexes — ipsi/contra thresholds and classifications by frequency.
• Speech in Quiet — SRT, WRS, and sound-sensitivity fields.
• Information — reference text; collected responses are shown on the report, not auto-scored.

When you're ready to see how scored and report-only results turn into a PDF, read Reports & PDFs.