Psychometric Analytics & Test Mapping
Licensing | Certification
Evidence that helps you understand how assessments perform — from individual questions to learning objectives, complete examinations, and the credentialing programs they support.
LearningSuite™ connects the evidence created during assessment delivery with the objectives, examinations, and credentialing programs those assessments support. Core psychometric measures are calculated from examination response data inside LearningSuite, while objective-level and program-level analytics allow reviewers to see how that evidence changes as the scope widens. The result is a connected view of assessment quality rather than a collection of disconnected reports.
Four connected levels
Evidence that scales from every question to the credentialing program.
LearningSuite organizes assessment intelligence at four interconnected levels. A reviewer can start with an individual item, move into the objective it measures, evaluate the examination as a whole, and then place those results in the context of the credentialing program.
Item level
Difficulty, discrimination, distractor behavior, response distributions, and item-level evidence.
Learning objective level
Objective mastery, assessment coverage, mapped items, and drill-down to the evidence behind each objective.
Exam level
Reliability, score distribution, pass rate, measurement uncertainty, and governed standard setting.
Program level
Credential status, renewal risk, employer and license-type performance, administrative activity, and reporting.
Test mapping & objective performance
Connect the question to the objective it is supposed to measure.
Traditional item analysis answers whether a question is performing well. Learning Objective Performance adds the next layer: how well candidates are performing against the objectives the assessment was built to measure, and whether those objectives are adequately represented on the examination.
Objective-Level Analytics
For each mapped learning objective, LearningSuite can report the number of associated items, response volume, average percent correct, pass rate where applicable, average difficulty, discrimination, point-biserial performance, and an objective status such as Strong, Monitor, or Needs Review.
Assessment Coverage
Identify objectives with no mapped items, objectives represented by only one item, objectives with unusually high item counts, and the share of examination coverage assigned to each objective. Reviewers can drill from an objective back to its linked items and distractor evidence.
Exam-level measurement
Reliability and measurement uncertainty, shown in context.
No single coefficient tells the whole story. LearningSuite reports reliability alongside the score distribution and related measures so reviewers can evaluate the examination using the population that actually produced the results.
Item performance
Find the questions that deserve attention before they disappear into an average.
LearningSuite calculates item difficulty and point-biserial discrimination for scored questions, then groups results into review bands so subject-matter experts can see the shape of an examination and drill into individual items that may warrant investigation.
Difficulty Index
The proportion of candidates who answered the item correctly.
Difficulty bands help reviewers prioritize items for examination. The appropriate level of difficulty depends on the purpose of the item, the credential, and the assessment design.
Discrimination / Point-Biserial
The relationship between performance on the item and performance on the examination overall.
Negative discrimination is surfaced separately because it can indicate an item that deserves closer review. A negative value is evidence to investigate, not an automatic determination that an item is defective.
Distractor analysis
Not just which option was chosen — how the candidates choosing it performed.
A useful distractor should behave differently from the keyed response. LearningSuite goes beyond simple option counts by examining the performance profile associated with each answer option, including distractor-level point-biserial correlation.
| Captured or calculated per option | What it helps a reviewer understand |
|---|---|
| Selection count | Shows which distractors are attracting responses and which are rarely or never selected. |
| Mean total score of selecting candidates | Shows whether an option tends to attract stronger or weaker performers. |
| Score standard deviation of selecting candidates | Shows the spread of overall performance among candidates selecting that option. |
| Distractor Point-Biserial | Helps distinguish a functioning distractor from one that may attract stronger candidates or provide little diagnostic value. |
| Correct option and response population | Provides the answer key and sample context needed to interpret the option-level statistics. |
For organizations subscribed to SolyraAI, these deterministic item and option statistics can also be interpreted in a written review that summarizes strengths, concerns, problematic distractors, and items that warrant human review.
Governed standard setting
Use psychometric evidence to support a documented standard-setting decision.
LearningSuite extends psychometric analysis into a Modified Angoff standard-setting review within the existing Rescore workflow. Aggregate examination evidence can be evaluated with optional SolyraAI support, but the recommendation remains advisory and the final decision remains with the qualified reviewer.
Psychometric Context
The review can consider the current passing score, score distribution, high and low scores, mean, median, standard deviation, reliability, Standard Error of Measurement, item difficulty, point-biserial discrimination, answer-option patterns, and the reviewer-approved definition of a borderline candidate.
Accept, Modify or Reject
SolyraAI can recommend an exam-level adjustment and show projected effects on the mean, pass rate, and candidates moved to passing. A reviewer must accept, modify, or reject the recommendation, provide a rationale, and approve the result before it can enter the existing Rescore process.
Advisory by design. SolyraAI does not directly change candidate scores. Approved standard-setting adjustments apply at the examination level and remain subject to required human approval and audit documentation.
Credential program intelligence
Assessment quality is one part of program health.
Exam Analysis answers whether an assessment is performing well. Program Analytics answers a different question: how the credentialing program itself is performing. LearningSuite connects both views within the same credential-lifecycle System of Record.
Status & distribution
Compliance rate, credential status, license distribution by type, and expirations over time.
Renewal risk
Expiring credentials, late renewals, aging buckets, and records requiring administrative attention.
Compliance rankings
Employer and license-type views that help administrators identify where compliance risk is concentrated.
Administrative activity
Program actions and recent activity that support operational review and audit preparation.
Reporting & documentation
Turn governed data into reports built for review.
LearningSuite supports interactive reports and export workflows across its analytics capabilities. Learning Objective Performance supports web review with PDF and Excel export, while Program Analytics supports configurable report profiles, on-screen preview, and PDF output for agency-facing and leadership use.
One governed source, multiple audiences. Psychometricians can work at the item level, credentialing leaders can evaluate program health, and regulators or boards can receive a documented report without rebuilding the underlying evidence in a separate system.
Deterministic evidence, optional AI
The measurements are calculated. The interpretation can be assisted.
LearningSuite separates empirical psychometric measurement from optional AI interpretation. That distinction preserves a reproducible evidence layer while still giving reviewers a faster way to understand what the numbers may mean.
Deterministic Measurement
Difficulty, item point-biserial, option-selection statistics, distractor point-biserial, reliability measures, descriptive statistics, and SEM are calculated by the platform from the supplied examination data or derived deterministically from those values.
Explanatory & Advisory Analysis
SolyraAI can explain psychometric evidence, identify items that warrant review, summarize strengths and concerns, and support governed workflows such as standard setting. It does not replace the underlying calculations or the qualified human decision-maker.
Frequently Asked Questions
Difficulty measures the proportion of candidates who answered an item correctly. Discrimination measures the relationship between performance on that item and performance on the examination overall. LearningSuite™ reports both so reviewers can distinguish a difficult item from an item that may not be functioning as intended.
LearningSuite calculates reliability statistics from examination response data and reports them with the score distribution and related descriptive measures. Depending on the workflow, this includes Cronbach’s Alpha, KR-21, standard deviation, and Standard Error of Measurement.
Assessment items can be associated with learning objectives. Learning Objective Performance aggregates the item evidence by objective so reviewers can evaluate objective mastery, assessment coverage, objectives with little or no coverage, and the individual items supporting each result.
Yes. LearningSuite includes a governed Modified Angoff standard-setting review within the Rescore workflow. SolyraAI can provide an advisory recommendation using aggregate psychometric evidence, but a qualified human reviewer must accept, modify, or reject the recommendation and provide a rationale before an approved adjustment can proceed.
No. Core psychometric values are calculated by LearningSuite from examination response data. SolyraAI can optionally interpret those values, summarize findings, and prioritize items for review, but the empirical measurements do not depend on AI.
Distractor point-biserial helps show whether an incorrect option is attracting lower-performing candidates as expected, attracting stronger candidates in a potentially problematic way, or functioning weakly. It complements option-selection counts and the overall-score profile of the candidates selecting that option.