The A in LEAF is where the essay earns its grade, and the part most students underestimate. The moves are concrete. They just take practice.
Analysis is your reasoning about the evidence, the work of turning a fact into a defence of your thesis.
Evidence is what happened. Analysis is what it means, and why the reader should care.
If a sentence could appear in a summary of the source, it isn't analysis. If it needs you to have read both the source and the rest of your essay, it almost certainly is.
You don't need all six in one paragraph. Two or three, executed well, lift a paragraph from competent to strong. When analysis stalls, pick a move by name and write one sentence that does exactly that.
| Move | What it does in a sentence |
|---|---|
| Interpret | Says what the figure means, not just what it is. |
| Causal | Names the most plausible cause behind the number. |
| Comparative | Sets it against a benchmark, an average, a past year, a rival. |
| Limit | Says what the evidence can not show. |
| Stake | Spells out what follows if the trend holds. |
| Reframe | Flips the obvious reading into a sharper one. |
A 2023 KOFICE study found Korean-language course enrolments rose 17% worldwide in the year after Squid Game. The figure is striking less for its size than for its direction: language study here follows the content rather than preceding it, reversing the usual assumption that fluency comes first and interest second [interpret]. Set beside the flat enrolment of the prior five years, a single show moved the curve more than a decade of institutes did [comparative]. If content, not curriculum, is now the driver, the policy lever shifts from classrooms to exports [stake]. The enrolment spike, then, is not a footnote to Hallyu's success but evidence of how cultural reach now manufactures its own demand for fluency.
Notice the ratio: one sentence of evidence, four of analysis. That is the shape of the long A.
Korea ranked 105th on the 2023 gender-gap index.
A 105th ranking, far below Korea's economic peers, signals the gap is institutional, not developmental: wealth alone hasn't moved it, so wealth alone won't fix it.
Seoul's subway ridership fell 7% after the 2022 fare rise.
A 7% drop from a modest fare rise suggests ridership is more price-sensitive than planners assumed, which makes fare policy, not service, the real lever on congestion.
Find a sentence in your draft that only restates a source. Which move would turn it into analysis?
Academic voice is neither bullying nor mealy-mouthed. Two registers tell the reader exactly how confident you are, and the trick is matching them to your evidence.
When the evidence supports a tendency, not a certainty.
“These findings suggest hydrogen adoption is plausibly industry-led, at least in 2018–2021.”
When the evidence is strong enough for certainty.
“The data clearly establishes that logistics density, not the app, is the structural advantage.”
Boosting past the evidence. If a source supports a tendency and you write “clearly demonstrates,” you've overdrawn, and reviewers spot it fast.
Turn three flat summaries into analysis, then calibrate four claims, hedged, calibrated, or over-boosted?
Each line summarises evidence. Add analysis, naming the move you use.
Write the move in brackets after each sentence, [interpret], [stake], [compare]. If you can't name it, it isn't analysis yet.
Tick one. If it's over-boosted, rewrite it to match its evidence.
| Sentence | Hedged | Calibrated | Over-boosted |
|---|---|---|---|
| This study proves once and for all that K-pop is the most powerful diplomacy tool ever invented. | ☐ | ☐ | ☐ |
| The figures suggest the four-day trial improved retention, though samples were small. | ☐ | ☐ | ☐ |
| Coupang's density is, without question, the only reason it stays competitive. | ☐ | ☐ | ☐ |
Swap “proves once and for all” for what the evidence earns, often “suggests” or “indicates.”
Chapter 10, pages 44 to 46, including the “one datum worked all the way up” page and the hedge/boost word banks, plus Exercises 10.1 to 10.2.
Chapter 10 · Next, citing the evidence — APA & IEEE