Claim, Evidence, Limitation
The Core Pattern
Every scientific communication product in this course should keep three pieces together:
- Claim: what you want the audience to understand.
- Evidence: the plot, table, dashboard output, code result, or citation that supports it.
- Limitation: what the evidence cannot prove.
If one piece is missing, the message is either hard to trust or too easy to overstate.
Build the Table First
Before opening revealjs, make a small table.
| Piece | Draft |
|---|---|
| Audience | Public-health analyst preparing a briefing |
| Question | How does mean BMI vary by income group in the classroom NHANES dataset? |
| Claim | Mean BMI differs across income groups in the prepared dataset. |
| Evidence | A5 corrected plot and A6 Table 1 |
| Limitation | Descriptive, unweighted, classroom snapshot; not causal or population-weighted |
| Source note | CDC/NCHS NHANES public-use data, prepared for classroom use |
This table is not an extra deliverable. It is a drafting tool for A8.
Reuse Earlier Work
From A5
Use the corrected plot and caption. The caption already names the statistic, source, exclusions, and limitation. Shorten it for a slide, but do not remove the limitation.
From A6
Use the Table 1, methods note, and provenance/stewardship note. They help the audience understand who is included, which values are missing, and why the result is descriptive.
From A7
Use the audience note and limitation/privacy check. They remind you that a KT product is designed for someone, not for everyone.
Plain-Language Revision
Technical draft:
The grouped BMI means show heterogeneity across IncomeGroup strata.
Plain-language revision:
Average BMI is not the same across income groups in this classroom dataset.
Limitation sentence:
This is a descriptive, unweighted comparison and does not show that income causes BMI differences.
Here is a second example, this time about missing data.
Technical draft:
Rows with missing BMI values were excluded via
filter(!is.na(BMI))prior to aggregation.
Plain-language revision:
Before calculating the group averages, we left out people whose BMI was not recorded.
Limitation sentence:
The averages describe only people with a recorded BMI, so groups with more missing values are less fully represented.
What Not to Do
Avoid these common shortcuts:
- turning a descriptive pattern into a causal statement;
- hiding the limitation in tiny text;
- showing a figure without naming the data source;
- copying a dashboard screenshot that exposes row-level data;
- asking AI to “make it more persuasive” without checking accuracy.
Slide-Level Check
For each slide, ask:
- What is the one claim?
- Where is the evidence?
- Where is the limitation?
- Is the source visible or easy to find?
If a peer cannot answer those questions, revise before submission.