Pre-registration template

When this is used

Before any experiment starts in the product (the software refuses to start one without a pre-registration link), and before any study on TaleTykes with an outside measure. The filled-in form is posted on a public registry and linked from the Evidence page.

1. Study information

• Title and short key (the key is what the product stores).
• Authors and affiliations, with conflicts of interest. TaleTykes staff always have one and say so.
• Research question in one sentence.
• Hypotheses, numbered, each with a direction. "No difference" is a legitimate expected result.

2. Design

• Design type: randomized by child, randomized by class, waitlist, or a quasi-experiment with a matched comparison. Say why.
• Unit of assignment and unit of analysis. If classes are assigned and children analyzed, say how clustering is handled.
• Blinding: who knows the assignment and who does not. Outside scorers never know.
• Duration, with the date after which no decision will be made before. In-product experiments run at least two weeks.

3. Sample

• Population and setting (grade, band, home or school).
• Inclusion and exclusion rules, written before the data are seen.
• Consent basis: research consent from a parent, or a school agreement that covers research use, and the opt-out path.
• Planned sample size with the calculation behind it (the smallest effect worth detecting, power, alpha, expected clustering). Stopping rules, if any.

4. Variables

• Manipulated variables: the exact settings and their values. In the product these come from an allow-list of defensible knobs.
• Primary outcome: one, named, with how and when it is measured and by whom.
• Secondary outcomes, listed in advance.
• Covariates and how they were chosen.

5. Analysis plan

• The model for the primary outcome, written out.
• How missing data are handled.
• Any sequential monitoring and the boundary used. The product uses a one-sided mixture sequential probability ratio test at alpha 0.05 as a monitor, not as the confirmatory analysis.
• Exploratory analyses, labeled as such.

6. Reporting

• Where results will be posted, and the commitment to post them whatever they show.
• What a null result changes in the product.
• Data availability: the de-identified transaction export format and who may request it.

7. Ethics

• Review board or the reason none is needed.
• Risks to children (none beyond ordinary use; nothing withholds teaching) and how they are watched.
• How a parent withdraws a child mid-study.

Sign-off

Name, role and date of the person who checked the form against the claims policy before the study started.