Record a testable hypothesis, primary metric, allocation, exclusions, and stopping rules before results can influence the experiment plan.
No public result for version 1 yet
Version 1 · 4 variables · 4 required
Fill in the values below — the customized prompt updates as you type. marks a required value.
The proposed intervention and decision the experiment should inform. Default: Will a shorter project-creation form increase the share of new accounts that create their first project?
Population, assignment unit, traffic, and any baseline information. Default: New self-serve accounts; randomize at account level; approximately 400 eligible accounts per week. Baseline completion rate is not yet available.
Outcome definition and practical impact threshold. Default: Primary outcome: account creates at least one project within seven days of signup. A gain of 3 percentage points would be worth considering; this is a business threshold, not a computed sample-size target.
Timing limits, concurrent changes, measurement risks, and dependencies. Default: One treatment and one control. Four-week launch window. Avoid changing onboarding emails during the experiment. Some users may belong to multiple accounts.
Draft a preregistration-style protocol for this proposed experiment. The protocol is a draft for review, not a completed registration.
Decision and proposed change:
Will a shorter project-creation form increase the share of new accounts that create their first project?
Population, assignment unit, and available traffic:
New self-serve accounts; randomize at account level; approximately 400 eligible accounts per week. Baseline completion rate is not yet available.
Primary outcome and smallest worthwhile effect:
Primary outcome: account creates at least one project within seven days of signup. A gain of 3 percentage points would be worth considering; this is a business threshold, not a computed sample-size target.
Operational constraints and known risks:
One treatment and one control. Four-week launch window. Avoid changing onboarding emails during the experiment. Some users may belong to multiple accounts.
Include:
1. The decision the experiment will inform, the falsifiable alternative hypothesis, and its null counterpart in plain language.
2. Eligibility, randomization unit, allocation, exposure definition, and handling of repeat participants or cross-group contamination.
3. One primary outcome with numerator, denominator, observation window, and unit; distinguish secondary metrics and guardrails.
4. A proposed analysis method and the assumptions it requires. Specify needed inputs for a sample-size calculation, including baseline variability or rate, effect size, significance level, and power. Do not invent a sample size or treat a business threshold as a computed detectable effect.
5. Rules to decide before launch: duration or stopping criterion, missing data, exclusions, outliers, multiple comparisons, and protocol deviations. Mark unresolved choices clearly.
6. A decision table for meaningful benefit, meaningful harm, and inconclusive evidence, keeping practical impact separate from statistical significance.
End with a checklist of facts that must be confirmed before enrollment. Do not look at outcome data, backdate a registration, or state that this plan guarantees an unbiased or adequately powered study.No public examples yet for version 1
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
Curated prompt only. No example output or model test is attached; model compatibility has not been verified. Default inputs are illustrative practice material; replace them with your own facts and context.
Reports go to moderators and are never shown publicly.
No public results yet
Turn reviewer comments into specific revision actions and courteous responses without claiming changes or analyses that have not been done.
No public results yet
Prioritize the main finding, section text and figures for a conference poster while preserving evidence limits and venue constraints.
No public results yet
Build a source-grounded glossary that preserves differences in definitions, assumptions and measurement across fields.
No public results yet
Find unmatched citations, unused references, inconsistent author-year labels and missing bibliographic fields without inventing metadata.
No public results yet
Translate a research question into observable events, recording fields and pilot checks that different observers can follow consistently.
No public results yet
Develop provisional codes with inclusion rules, exclusions and evidence anchors while keeping interpretation distinct from the source material.
Oldest first · one level of replies
Sign in to join the discussion. Share what you learned or made with this prompt.
No comments yet
Be the first to share how this prompt worked for you.