英文個案題最容易被誤解成英文寫作測試。事實上它考的是：面對一堆互相衝突的資料，你能否分辨輕重、提出可行方案、並誠實交代取捨。英文只是載體。
The English case question is most often mistaken for a test of English writing. What it actually tests is this: faced with a pile of conflicting material, can you tell what matters from what does not, propose something that could be implemented, and be honest about the trade-offs. English is only the vehicle.
典型的 JRE 英文個案題會給你：一個政策情境、若干數據、三至四方持份者的意見，然後要求你識別議題、評估建議、提出保障措施、說明如何評估成效。
留意題目的角色設定通常是「You are a member of the policy team in the XXX Bureau」。這句不是裝飾——它界定了你的身分。你是政策主任，不是評論員、不是倡議者、不是研究員。
評論員可以說「這個方案荒謬」。政策主任要說「此安排在執行上存在三項困難」，然後逐項處理。前者展示立場，後者展示能力。JRE 考的是後者。
閱卷者當然會看語文，但語文分數的上限很快就到。真正拉開距離的是：你有沒有看見資料裡的矛盾、有沒有處理它、建議是否可以明天就執行。
換句話說，一篇文法完美但論證空泛的答案，分數會低於一篇有小瑕疵但判斷準確的答案。
建議把 100 分鐘分成：讀題 20 分鐘、大綱 15 分鐘、書寫 60 分鐘、檢查 5 分鐘。
二十分鐘讀題聽起來很多，但這是回報最高的投資。分三輪讀：
只看題目要求（REQUIRED 那一段）和小標題。先知道要交什麼，再回頭找材料。很多人反過來，讀完全部資料才看要求，結果重讀一次。
逐段讀，用四種標記：
合上題目，用一句話回答：這個方案最可能失敗在哪裡？
答得出，你的整篇文章就有主軸。答不出，回去重讀標了星號的部分。
這是整篇文章的核心，值得單獨一節說明。
JRE 個案題的背景資料裡，幾乎必定藏著一項足以令方案不可行的限制。它通常寫得很平淡，像是背景說明，但實際上是考點。
| 資料寫法 | 實際意思 |
|---|---|
| 現時處理容量約覆蓋所需的六成 | 全面推行會產生無法處理的量，必須分階段 |
| 四個地盤中有兩個未能如期落成 | 目標與供應時間錯配，要調整推進速度 |
| 相關專科現時輪候時間已經偏長 | 篩查出來的新個案會令輪候更長，識別之後沒有承接 |
| 現行法例未完全涵蓋此類操作 | 法律基礎未備，先行推展會有風險 |
| 試行依賴校外義工提供支援 | 模式不可持續，推廣時會斷裂 |
看見了卻在建議中略過，是最大的失分位。反過來說，正面處理它——說明如何補上、或如何調整推行節奏令兩者同步——是最快拿分的方法。
| 段落 | 任務 | 建議篇幅 |
|---|---|---|
| 1. Issues | 識別三項最重要的議題並排序，說明排序理由 | 約 25% |
| 2. Options | 評估可行方案，交代各自的取捨 | 約 25% |
| 3. Recommendation | 明確建議 + 保障措施 + 過渡安排 | 約 35% |
| 4. Evaluation | 成效指標與檢討時點 | 約 15% |
如果題目明確列出 (a)(b)(c)(d) 四項要求，就照它的次序寫，並用小標題對應。不要自作聰明重新編排。
題目通常要求「identify and rank the three most critical issues」。留意是 rank，不只是 list。
很多考生列出三項就停，沒有說明為何是這個次序。排序理由本身就是分數。
The three issues are ranked by irreversibility. The absence of a legal basis for the proposed operations ranks first: proceeding without it would expose the Government to challenge and any resulting suspension would be costly to reverse. Data retention ranks second, as decisions taken at the design stage are difficult to revisit once systems are built. Cost-effectiveness ranks third — significant, but capable of adjustment during implementation.
三十多個字就交代了三項議題、排序，以及排序的邏輯。
如果題目給了選項（例如 a vehicle cap versus quality-based requirements），逐個評估，並明確交代 trade-offs。
常見錯誤是只講優點不講代價。每個選項都應該有一句「代價是什麼」。
A quota would protect existing operators and limit disruption, but would leave under-served districts unchanged and require a mechanism to adjust the ceiling over time.
Quality-based requirements would allow supply to respond to demand, but would place the adjustment burden on the incumbent trade without a transition path.
這種「would... but would...」句式在 policy paper 很常用，因為它同時交代收益與代價，語氣中性。
這一段篇幅最大，因為它包含三件事。
不要含糊。「It is recommended that the proposal proceed, subject to the safeguards set out below」比「The proposal has both merits and drawbacks」好得多。
建議可以有條件——有條件不等於騎牆，而是說明在什麼情況下做什麼。
這是題目明碼標價的分數（safeguards / mitigation measures）。要具體到可以執行：
如果現況已有既得利益者（例如現行牌照持有人、現有服務使用者），必須交代過渡安排。略過這一塊，建議就不完整。
同時，這裡是你處理那項關鍵限制的地方。例如容量只有六成，就在此說明分階段納入的次序，以及以什麼條件進入下一階段。
題目通常會指定時限：「within 24 months」「over the first three years」。答案要對應那個時限。
成效指標與代價指標並列，是成熟答案的標誌。例如攔截可疑轉賬：除了「攔截金額」，也要看「誤判率」與「平均延誤時間」。只講成效，閱卷者會覺得你沒想清楚。
另外交代檢討時點與觸發條件：什麼時候回頭看、看到什麼數字就要調整、什麼情況下應該叫停。
JRE 要求的是 policy paper 的語域：清楚、中性、可執行。不需要華麗詞藻，也不要過度口語。
| 避免 | 改用 |
|---|---|
| This proposal is a disaster waiting to happen | This proposal carries significant implementation risk |
| We must act now! | Early action is recommended, for the following reasons |
| It's clear that everyone agrees | Submissions from three of the four groups supported the direction |
| a lot of / a big problem | a substantial proportion / a material concern |
最後一種——主動承認建議的邊界——最少人用，但最能顯示成熟度。
官方明言不獲分數。資料只在支撐判斷時引用，而且要立刻接上分析。
上面已詳述，這是最大的一項。
題目要 rank，你只 list。排序理由本身就是分數。
「Enhance inter-departmental collaboration」不是建議。要說明由誰做、做什麼、多久一次。
區議會屬諮詢組織，其意見供參考，不構成批准或否決。同類錯誤包括把公眾諮詢的多數意見當成必須執行的結論。
這是題目明確要求的部分，卻常被時間不足的考生放棄。寧可寫得簡短，也不要不寫。
英文題練的是格式熟練度——四段框架要練到不用想。建議分兩階段：
階段一 · 只練讀題與大綱。限時 35 分鐘，讀題 20 分鐘、寫大綱 15 分鐘，不寫全文。這樣一小時可以練兩題，而且集中訓練最關鍵的判斷環節。
階段二 · 每週一篇完整計時。100 分鐘，不查資料。寫完隔一天再讀。
本站線上練習系統提供 100 分鐘計時模式，附字數統計與草稿自動儲存；模擬試題每題附答題提示，會直接指出該題的關鍵限制在哪裡，可以用來核對自己有沒有看漏。
中文題的框架不同，可參考中文論述題四段結構。兩題分數比重相同，不要只練一邊。
A typical JRE English case question gives you a policy situation, some figures, and the views of three or four groups of stakeholders, then asks you to identify the issues, assess a proposal, propose safeguards, and say how you would measure whether it worked.
Note that the question normally casts you as "a member of the policy team in the XXX Bureau". That line is not decoration — it defines who you are. You are a policy officer: not a commentator, not an advocate, not a researcher.
A commentator can write "this proposal is absurd". A policy officer writes "this arrangement presents three implementation difficulties", and then deals with each. The first displays a position; the second displays capability. The JRE is marking the second.
Markers do of course look at language, but marks for language reach their ceiling quickly. What actually separates scripts is whether you saw the contradiction in the material, whether you dealt with it, and whether your recommendation could be acted on tomorrow.
Put plainly: a script with flawless grammar and hollow reasoning scores below a script with small blemishes and sound judgement.
A workable split of 100 minutes: 20 reading, 15 outlining, 60 writing, 5 checking.
Twenty minutes of reading sounds like a lot, but it is the highest-return investment on the paper. Read in three passes.
Read only the task itself (the REQUIRED paragraph) and the subheadings. Know what you have to hand in before you go looking for material. Most people do it the other way round, read everything first and then the task, and end up reading it all again.
Work through paragraph by paragraph with four marks:
Close the paper and answer one question in a sentence: where is this proposal most likely to fail?
If you can answer it, your whole script has a spine. If you cannot, go back to the starred paragraphs.
This is the heart of the script and deserves its own section.
The background material in a JRE case question almost always hides one constraint severe enough to make the proposal unworkable. It is usually written very flatly, as though it were background colour, when in fact it is the thing being tested.
| How the material puts it | What it means |
|---|---|
| Current processing capacity covers about 60 per cent of what is required | Full roll-out would generate volume that cannot be handled; it has to be phased |
| Two of the four sites will not be completed on schedule | Targets and supply are out of step; the pace has to be adjusted |
| Waiting times in the relevant specialty are already long | Newly identified cases will lengthen the queue; identification with nothing behind it |
| Existing legislation does not fully cover operations of this kind | There is no legal basis; proceeding carries risk |
| The pilot relied on volunteers from outside the schools | The model is not sustainable and will break on scaling |
Seeing it and then skipping it in your recommendation is the single biggest way to lose marks. Conversely, dealing with it head on — saying how the gap would be filled, or how the pace of implementation would be adjusted so the two move together — is the fastest way to gain them.
| Section | Job | Suggested share |
|---|---|---|
| 1. Issues | Identify the three most important issues, rank them, and justify the ranking | ~25% |
| 2. Options | Assess the workable options and state the trade-offs | ~25% |
| 3. Recommendation | A clear recommendation, safeguards, and transitional arrangements | ~35% |
| 4. Evaluation | Performance measures and review points | ~15% |
If the question sets out requirements (a), (b), (c) and (d), follow its order and use subheadings that match. Do not rearrange them to be clever.
The question normally asks you to "identify and rank the three most critical issues". Note that it says rank, not merely list.
Many candidates give three issues and stop, without saying why that order. The justification for the ranking is itself worth marks.
The three issues are ranked by irreversibility. The absence of a legal basis for the proposed operations ranks first: proceeding without it would expose the Government to challenge and any resulting suspension would be costly to reverse. Data retention ranks second, as decisions taken at the design stage are difficult to revisit once systems are built. Cost-effectiveness ranks third — significant, but capable of adjustment during implementation.
Three sentences covering three issues, the order, and the logic behind the order.
Where the question supplies options — a vehicle cap versus quality-based requirements, say — assess each and state the trade-offs explicitly.
The common error is giving the advantages without the cost. Every option should carry a sentence saying what it costs.
A quota would protect existing operators and limit disruption, but would leave under-served districts unchanged and require a mechanism to adjust the ceiling over time.
Quality-based requirements would allow supply to respond to demand, but would place the adjustment burden on the incumbent trade without a transition path.
This "would… but would…" construction is a staple of policy papers precisely because it states the gain and the cost together in a neutral voice.
This is the longest section, because it carries three things.
Do not hedge. "It is recommended that the proposal proceed, subject to the safeguards set out below" is far better than "the proposal has both merits and drawbacks".
A recommendation may be conditional — conditions are not fence-sitting; they state what would be done under what circumstances.
These are marks the question advertises openly (safeguards, mitigation measures). Make them specific enough to act on:
Where there are existing interests — current licence holders, current service users — you must deal with the transition. Leave that out and the recommendation is incomplete.
This is also where you address the binding constraint. If capacity covers only 60 per cent, this is where you set out the order in which groups are phased in, and the condition for moving to the next phase.
The question usually specifies a horizon: "within 24 months", "over the first three years". Your answer should match that horizon.
Setting a cost measure alongside a performance measure is the mark of a mature answer. For intercepting suspicious transfers, for instance: as well as "value intercepted", look at the false-positive rate and the average delay imposed. Report only the upside and the marker will conclude you have not thought it through.
Also state when it would be reviewed and what would trigger a change: when you would look again, what figures would prompt an adjustment, and in what circumstances the scheme should be halted.
What the JRE wants is the register of a policy paper: clear, neutral, actionable. No ornament, and nothing too colloquial.
| Avoid | Use instead |
|---|---|
| This proposal is a disaster waiting to happen | This proposal carries significant implementation risk |
| We must act now! | Early action is recommended, for the following reasons |
| It's clear that everyone agrees | Submissions from three of the four groups supported the direction |
| a lot of / a big problem | a substantial proportion / a material concern |
The last of these — volunteering the boundary of your own recommendation — is the least used and the most telling sign of maturity.
The official guidance says plainly that this earns no marks. Cite the material only where it supports a judgement, and follow the citation immediately with analysis.
Covered at length above. This is the largest of the six.
The question asks you to rank; you list. The justification for the order is itself worth marks.
"Enhance inter-departmental collaboration" is not a recommendation. Say who would do what, and how often.
District Councils are consultative bodies: their views inform, but do not approve or veto. The same error includes treating the majority view in a public consultation as a conclusion that must be implemented.
This is a part the question explicitly asks for, and it is routinely abandoned by candidates who run out of time. Better brief than absent.
What the English question rewards is fluency with the format — the four-section framework should need no thought. Practise in two stages.
Stage one · reading and outlining only. Set 35 minutes: 20 to read, 15 to outline, and do not write the script. That gets you two questions in an hour, with the training concentrated on the part that matters most.
Stage two · one full timed script a week. 100 minutes, no looking things up. Leave it a day before you read it back.
The online practice system on this site offers a 100-minute timed mode with a word count and automatic draft saving. Each question in the mock papers carries notes that point out where that question's binding constraint lies, so you can check whether you missed it.
The Chinese question uses a different framework — see the four-section structure for the Chinese essay. The two questions carry equal weight, so do not practise only one side.
自己看自己的答案，很難察覺論證的漏洞。計劃 B 提供 3 份模擬試卷及 3 次導師逐份批改，附詳細評語，5 個工作天內回覆。也可以先用免費的線上練習系統操一題。
查看收費計劃 →Reading your own script, the holes in your reasoning are very hard to see. Plan B provides three mock papers and three rounds of paper-by-paper marking by a tutor, with detailed comments returned within five working days. You can also work through one question first on the free online practice system.
See plans and pricing →