An annotated example brief
Project 1 · a competent draft, marked up with what would raise it
How to read this page
Below is a short example brief on invented findings, written the way a competent team often writes a first draft. It is not a model answer — it is a competent answer, which is exactly the level most first drafts reach. It is also compressed: your brief will run 750–1,500 words, and this one keeps only enough of each section to annotate.
After each section you will find a boxed note — What would raise it — naming the specific move that separates competent from strong. The notes are the point of this page. Read each one against the section above it. The rubric’s own line is that strong submissions show the team’s thinking; competent submissions show the team’s output — the notes are what that difference looks like in prose.
The numbers, the wave, and the findings here are fabricated for illustration. Do not cite them, and do not reuse the question.
The example brief
The question
Among U.S. adults who say they get news on social media, how does confidence in that news differ by age?
Social media is now a routine news source, and confidence in what people encounter there shapes whether they act on it. Our intended reader is a campus communications office deciding which channels to use for public-health messaging.
What would raise it
The question is genuinely good — narrowed to a population it is actually about, and the answer is not obvious in advance. The reader sentence is doing less work than it looks: “a campus communications office” is named but never used again. A strong brief lets the reader shape a real choice later — for instance, closing with what this result would mean for which platform that office picks. The brief requires an intended reader, so naming one is not optional — the move is to serve the reader you named, not just to name one.
Measurement
Confidence was measured with a single item asking how much respondents trust the news they see on social media, on a four-point scale from “a lot” to “not at all.” We collapsed this to a binary indicator of high confidence (“a lot” or “some”) versus low.
Age came from the fielded age-category variable, in four bands.
What would raise it
The construct-to-variable chain is stated, which is the requirement. What is missing is the cost of the collapse: folding four categories into two discards the distinction between “a lot” and “some,” and those are not the same claim. One sentence — “we chose the binary because the cell sizes in the oldest band were too small to describe reliably, at the cost of hiding whether confidence is enthusiastic or merely tolerant” — would even do double duty as the measurement limitation this section is asked to name, which this draft defers to its closing instead. Naming the cost turns a reported decision into a defended one.
Data and analytic sample
We used a single ATP wave. Of the 5,097 respondents, 3,842 said they get news on social media and answered the confidence item; those respondents are the analytic sample. Respondents who refused either item were dropped.
What would raise it
The denominator is stated, which many drafts forget. But those 1,255 exclusions mix two very different groups, and the prose folds them together: people who do not get news on social media are out of scope — excluding them is the question working as intended — while people who refused the confidence item are missing data, the distinction M05 drilled. “Were dropped” also passes over the question a careful reader asks next: were the people who refused different from the people who answered? Even a single line — how many refused, and whether they skewed toward one age band — is the difference between reporting a sample and understanding one.
Table 1 and the figures
Table 1 describes the 3,842 respondents in the analytic sample by age, education, and gender. Figure 1 shows the share with high confidence in each age band. Figure 2 shows the same breakdown separately for respondents who use social media daily versus less often.
What would raise it
Figure 2 is well chosen — it qualifies Figure 1 rather than restating it, which is the standard the rubric sets. The prose does not yet say what either figure shows. “Figure 1 shows the share…” describes the axes; a strong brief describes the finding: which band is highest, how large the gap is, and whether confidence falls at every step of age or dips and recovers along the way. If a reader could get the same information from your axis labels, the sentence is not earning its place.
What the findings mean — and do not mean
Confidence in social-media news declines with age in this sample. The gap between the youngest and oldest bands is about 20 percentage points. This is a descriptive pattern in one wave; it does not tell us that ageing causes distrust, and we cannot say whether it holds in other years.
One measurement limitation is that a single item is a thin measure of confidence. One thing this analysis raises but cannot answer is whether the age gap reflects different platforms rather than different ages.
What would raise it
All five required elements are present, and the causal disclaimer is correctly stated. The closing is where competent drafts most often stop early. The final limitation — that the age gap might be a platform gap — is the most interesting sentence in the brief, and it arrives as an afterthought in the last line. A strong brief would have noticed that earlier and either checked it (a third figure splitting by platform) or said plainly why the wave cannot support the check. The best limitation is one you went looking for, not one you thought of while writing the conclusion.
What this example is not
It does not show you a strong brief, on purpose. If it did, teams would pattern-match to it and every submission would look the same. The point of reading a competent one is to see how much of the distance to strong is covered by five specific moves — one per note above:
- serving the reader you named
- naming the cost of a measurement decision, not just the decision
- separating who was out of scope from who was missing — and asking who the missing were
- describing the finding, not the axes
- following up the interesting limitation rather than listing it
Each of those is a sentence or two, or at most one more small computation. None requires a new method — they require having asked one more question of work you had already done.