Permanent product recording

The product remains available for review

Direct observation records the response while it occurs. Product measurement examines a result afterward. A reviewer can often rescore the same product, which supports calibration and audit.

The word “permanent” is relative. A saved digital entry may persist, while a cleaned table or assembled package can change quickly. Define the review window and preserve evidence when later changes could erase the product.

A product must map to the response

A valid product should have a dependable relationship to the behavior of interest. Ask:

• Could someone else have created it?
• Could the response occur without leaving this product?
• Could the product appear without the response?
• Does it show quality, completion, count, or only presence?
• Was there a genuine opportunity to create it?
• Did ordinary assistance, prompts, or automation contribute?

A submitted form can show that a submission occurred. It may not show who entered the information, how much help was used, or whether the person understood it.

Write a scoring rubric

Define what qualifies before reviewing products. For a packed kit, list required components, acceptable substitutions, quality criteria, and treatment of partially completed kits. Train reviewers with examples and nonexamples.

If the score is 17 correct products out of 20 eligible opportunities, report 17 of 20, or 85%. Preserve which three failed and why. A percentage without the denominator or error pattern loses much of its usefulness.

Set the review deadline too. A same-day check may support quick correction, while a weekly audit serves a different decision. Store the product date, review date, reviewer, rubric version, and final disposition.

A fictional vocational example

Keira is a fictional 19-year-old who wants to prepare seed packets at a community greenhouse. With Keira, the team defines a complete packet as one labeled envelope containing the correct seed type, specified quantity range, and sealed edge. Her chosen visual guide and counting tray remain available.

At the end of one shift, 20 packets are linked to Keira’s workstation and job log. Seventeen meet all three criteria, so product accuracy is 17 of 20, or 85%. Two contain the wrong seed type and one is unsealed.

The supervisor independently rescored six packets and agreed on all six. That small agreement sample supports scoring consistency for those products. It does not prove authorship, independence, or quality across every packet. The team also asks Keira which supports felt useful before changing the routine.

Track opportunity and authorship

Products can be misattributed in shared spaces. Use a reasonable provenance method such as a workstation log, individual file history, assigned batch, or direct sample. Avoid invasive monitoring when a less intrusive control works.

The denominator should include eligible opportunities due for review. If a printer failure prevents three labels from being produced, record three system-blocked opportunities separately. They should not become client errors or vanish from operations reporting.

Digital products need audit context

Timestamps, form submissions, messages, and workflow records can serve as products. Verify account access, automation, edits, imports, shared credentials, time zone, and audit history. A final database state may hide who changed a field and when.

Preserve original evidence and correction history. If software scores the product, validate the scoring logic against human review and track version changes, failures, and overrides.

Permanent products miss response process

The final result may omit latency, effort, prompts, discomfort, communication, problem solving, partner behavior, and safety. A perfect worksheet could follow extensive prompting. An imperfect product could reflect inaccessible materials or a device failure.

Add direct observation or another measure when process matters. Pair the score with the person’s report and relevant environmental measures.

Protect privacy and access

Products may contain protected health information, school records, images, messages, or another person’s data. Collect only what the measurement purpose requires. Follow applicable consent, assent, privacy, security, access, retention, disclosure, and deletion requirements.

Keep AAC, interpreters, mobility and sensory supports, breaks, food, water, bathroom access, prescribed care, and emergency help available. Record support use rather than removing it to create an artificial independence score.

Review measurement quality

Useful controls include products with verified opportunity and provenance divided by products due for review, products scored under the current rubric, sampled reviewer agreement, system-blocked opportunities, missing products, scoring corrections, and automated-score validation.

The BACB BCBA Test Content Outline, Sixth Edition distinguishes direct, indirect, and product measures and covers reliability, validity, and representative-data selection. It defines examination content and does not validate a particular product.