Almost every IA that goes wrong went wrong here, weeks before anyone wrote a sentence.
A topic that fails one of the five questions below cannot be rescued later by good writing, however well you write. The problem shows up at the end, in the discussion, when there is nothing believable to recommend. By then you have an introduction and a methodology built on it.
If you already have a topic, treat this as a health check. It takes five minutes and it is worth doing.
The idea underneath all five questions
This IA is not about studying something interesting.
The guide frames the task around the power of psychology to lead to real improvement in people's quality of life. If nothing in your idea could change for anybody, it is the wrong idea, however good the psychology is.
Hold that while you read the rest.
Question 1: Is there a problem?
Not a topic. Not a curiosity. Something going wrong.
"Social media and teenagers" is a topic. "Students who join a school mid-year report feeling they do not belong" is a problem. You can only build an IA on the second kind.
Write it as one sentence before you go any further.
Question 2: Who does it hurt?
Your population of interest has to be specific enough that a stranger would know whether it means them.
It can be broad, like teenagers in Indonesia. It can be narrow, like students aged 14 to 18 who joined an international school in Stockholm within the past twelve months. What it cannot be is "students" or "young people", because those describe two hundred different studies.
Narrowing the population is the single cheapest improvement available to most IAs.
Question 3: What could change?
This is the question people skip, and skipping it is what produces the panicked, invented policy recommendation in the discussion three weeks later.
If your findings came out as you expect, what could a school, a clinic, a workplace or a family actually do differently?
It has to be something specific and plausible. Not "raise awareness". Something a real person could implement on a Monday.
Ask it now, because it returns as a whole paragraph in your discussion, worth part of six marks. Answers invented at that stage read exactly as invented as they are.
Question 4: Is there research?
You need exactly two studies, and they have to be findable.
What the two need to do. Study one usually shows the problem is real: evidence that this group experiences this thing, or experiences it more than others. Study two usually shows why it matters or what affects it: evidence linking the problem to an outcome, or identifying something that makes it better or worse.
The rules to check. Peer-reviewed psychology journals. Education and business journals are risky, so the safe advice is to stay in psychological literature. They do not have to be on your exact population: if you are studying Indian adolescents, a study on adolescents is fine, but if you are studying kindergartens, it should be about kindergartens. You must be able to state the findings and conclusions of each in about forty words, and if you cannot, you have not understood it well enough to use it.
Two only. A third full study earns nothing and eats words you will need in your discussion. Short citations to evidence the scale of the problem are normal and expected.
One rule people miss: your own proposal must not repeat either study. Same variables and same population as your background research is not allowed, even if you design your own tool.
Where to look. Google Scholar. Your school library databases, usually PsycINFO, PubMed or EBSCO. ResearchGate has many free abstracts and full articles. Search the problem plus the population plus the outcome, rather than the topic on its own.
How to read a paper without drowning. You do not need the whole article. Read the abstract, then the discussion, then the first paragraph of the introduction if you still need context. You can skip the method and results sections almost entirely, because you are not reporting them.
The forty-minute test. Do this before you fall in love with the idea. If forty minutes of searching turns up nothing usable, the topic is too niche and you will be writing an introduction with nothing underneath it.
Question 5: Is it allowed?
Quick, but non-negotiable.
Banned outright. Abuse of any kind, whether sexual, physical or emotional. Self-harm, including suicide. Pornography, rape or other sexual acts. Serial killers or acts of torture. Anything involving animals. Ingestion, although investigating self-reported dietary choices is permitted. The guide notes this list is not exhaustive, and your teacher can advise on what is sensitive in your local context.
Allowed, but handle with care. Mental health, body image, neurodiversity and cultural difference are all permitted here. Several of them are not allowed in class practicals, so this is your opportunity. Write the proposal as though an ethics committee has to approve it, because that is the standard being applied. Studies that are unjustifiably unethical, such as withholding food, forcing participation or public humiliation, are out.
What the hypothetical nature does allow. Because you never run it, you may propose things that would be unfeasible for a school student. You could propose interviewing people with depression, or a longitudinal study running over many years. That is explicitly permitted.
What you cannot do is propose something unethical and call it hypothetical.
Also check. Your proposal is individually developed. Same topic as a classmate is fine, same population is fine, but avoid coordinated research questions, tools and wording that could make the work look jointly produced. You are not reusing an aim from one of your class practicals. And nothing in your design would humiliate, deceive without justification, or pressure anyone into taking part.
Where to look for ideas
Start from the context, then look for the kind of problem that tends to work inside it.
In education: exam stress, procrastination, disengagement after a school move, homework and sleep, feedback students never act on, subject choice anxiety.
The same approach works across health and well-being, human development, human relationships, and learning and cognition. Find something going wrong for a group you can name.
Two examples of the shape a workable proposal takes: an observational study of how space is used on a school sports pitch during lunchtimes; a questionnaire investigating which factors a specific group finds most damaging to their mental health. Both name a population, both point at something that could change.
The step that saves the most time
Sketch five items for your data collection tool before your topic is approved.
Some perfectly good ideas cannot produce a tool. You find that out in twenty minutes now, or in week three when you have already written two sections.
If your five draft items feel forced, vague or impossible to write, that is information about your topic, not about your writing.
The order of week one
No prose at all. Five days of decisions.
Take three ideas through the five questions until one answers all of them. Nail down the population and the problem. Find your two studies and save them with full references. Sketch five items for your tool, or decide to change the topic. Then lock your aim or research question and plan the discussion in note form.
If you can do that, the writing takes a week and mostly writes itself.
If you want the full topic planner, the sentence banks and a complete model proposal, that is what the PsychExcel Complete IA Handbook is built for.