Introduction

Lesson 1 followed the research cycle from a topic to a problem to a question, and it sorted research questions into four kinds: descriptive, explanatory, predictive and exploratory. Knowing the kind of question is a first step. A question also has to be precise enough that a research team can decide whom to recruit, what to measure and what to compare. This lesson gives questions that precision. Section 1 introduces the PECO and PICO formats for questions that compare groups, Section 2 introduces SPIDER and related formats for qualitative questions, Section 3 judges candidate questions against the FINER criteria, and Section 4 turns the chosen question into aims, objectives and hypotheses and shows how to find a handful of key papers.

1.1 Why a Research Question Needs Parts

Consider the question "Does loneliness affect older adults' health?" It names a topic that matters, but it leaves almost every decision open. It does not say which older adults (everyone over 65 in Canada, or residents of one region), how loneliness will be defined, whom the lonely older adults will be compared with, which aspect of health will be measured, or over what period. Each of these gaps is a decision that someone will have to make later. If the team makes those decisions one at a time during data collection, the study can drift, and different team members can end up working toward different questions.

A structured question format is a short list of the parts that a well-built question should contain, usually remembered by an acronym. The best-known format comes from evidence-based medicine. Richardson et al. (1995) described "the well-built clinical question" as having four parts: the patient or problem, the intervention, the comparison and the outcome. These parts became known as PICO. Researchers in environmental and occupational health later adapted the format for questions about exposures that people experience without anyone assigning them, giving PECO (Morgan et al., 2018).

Structured formats help a research team in three practical ways. They make missing parts visible, because an empty box in the format is easy to see. They supply the words for a literature search, which Section 4 uses to find key papers. They also point toward the study's variables: the exposure and outcome named in a PECO question become the first two nodes of the directed acyclic graph that Lesson 3 teaches you to draw, and Lesson 7 turns each part into a population definition and a measure.

1.2 The Four Parts of a PECO Question

PECO stands for population, exposure, comparator and outcome. Each part answers a question that the research team must settle before it can design the study. Select each card to see what the part means, what to specify, and how the Cedar Valley team filled it in.

Put together, the four parts produce one sentence. The Cedar Valley team wrote the following question for the quantitative strand of its study.

The question uses the word "associated" deliberately. Loneliness cannot be assigned to people, so the team will observe it and compare groups that already differ. Older adults who are lonely may also differ from other older adults in age, income, chronic illness and whether they live alone, and some of those differences could explain any link with emergency department use. An association is a statistical link between two variables, and it may or may not reflect cause and effect. Lesson 3 shows how a causal diagram identifies the other variables that the team needs to measure, and HSCI 341 develops the reasoning used to move from association toward a causal interpretation.

1.3 PICO: Questions About Interventions

PICO replaces the exposure with an intervention: a program, treatment, service or policy that a researcher or an organization delivers on purpose, and that could in principle be assigned to some people and withheld from others. The comparison is usually usual care, no intervention, a placebo or a different intervention. PICO questions are the natural form for trials and program evaluations.

A simple test helps decide between the two formats. If a researcher could, in principle and ethically, decide who receives the factor, the question is a PICO question. If people bring the factor with them, or circumstances assign it, the question is a PECO question. Loneliness fails the test, because no researcher can make people lonely, so the Cedar Valley question is a PECO question. Some topics can be studied either way. Physical activity can be an exposure that people report (PECO) or a supervised exercise program that a trial assigns (PICO), and the two questions lead to different studies.

The Cedar Valley Health Authority has asked the team whether a future study could test a community connector service, in which a trained worker links lonely patients with local groups and activities. The team drafted a PICO question for that possible study, while keeping its current study observational.

PICO and PECO are also used to structure the questions that guide systematic reviews, where each letter becomes a block of search terms. HSCI 241 Lesson 2 teaches review questions, and Lessons 3 to 5 of that course teach searching. In this course the formats structure the question for a single primary study.

1.4 Writing and Checking a Structured Question

Most students find it easier to build a PECO or PICO question in a fixed order than to write the sentence all at once. The following steps work for most topics.

1. Start from the problem statement (Lesson 1), and state in one sentence why the problem matters to someone.
2. Name the outcome first, because it is what the study is trying to explain or change, and decide how and when it will be measured.
3. Name the exposure or intervention, and say how a person will be classified as exposed or as receiving the intervention.
4. Name the comparator explicitly, so that a reader knows who the exposed group will be compared with.
5. Define the population by person, place and time.
6. Write the question as a single sentence that ends with a question mark, using "associated with" for an observed exposure and words such as "reduce" or "increase" only for an assigned intervention.
7. Check that each part is defined well enough that two team members would classify the same participant in the same way.

Drafts often contain one of a small number of problems. Open each item to see a weak version, the problem it contains and a stronger version.

The draft names no population or place, gives no definition of screen time, has no comparator, and uses "hurt" for an outcome that could mean sleep, mood, eyesight or school grades. One stronger version reads: "Among students in grades 9 to 12 in a British Columbia school district, is recreational screen time of more than four hours on school days, compared with two hours or less, associated with sleeping fewer than eight hours on school nights during the spring term?" Another student might choose a different outcome, such as symptoms of anxiety measured with a validated scale, and that choice would be equally acceptable if it were defined and measurable.

A structured question is a first draft. Section 2 turns to questions about experiences and meanings, which these formats fit poorly, and Section 3 tests whether a well-structured question is also a good one to study.

2.1 What Qualitative Questions Ask

Lesson 1 described exploratory questions as questions about how people experience, understand or respond to something, often in areas where little is known. Qualitative research answers these questions with words, observations and other non-numerical data. A qualitative research question asks how or what: how a group of people experience a situation, what a condition or service means to them, or how a process unfolds in a particular setting.

The Cedar Valley team noticed something in early conversations with its advisory group of six older adults. Several advisors had moved from Cedar City to smaller towns in the region after retiring or after a spouse died, and they described the move as the point at which their social lives changed. The team decided that its qualitative strand would ask about this experience. A first attempt to write the question in PICO form produced the following.

The problem lies in the format. PICO and PECO were built for questions in which groups are compared on an outcome that can be counted or measured. A question about how people experience social connection after a move has no intervention, no comparison group and no single outcome. It needs a format whose parts match what qualitative studies actually specify: the people who will be interviewed or observed, the experience or process of interest, and the way data will be collected.

Features of a good qualitative question

Writers on qualitative design offer consistent advice. Creswell and Poth (2018) recommend a single broad central question supported by a small number of sub-questions that divide it into parts. A good qualitative question usually begins with "how" or "what", uses open verbs such as explore, describe or understand, names the people and the setting, and focuses on one experience or process. It avoids words that imply measurement or cause, such as "effect", "impact", "increase" or "to what extent", and it avoids wording that presumes the answer. Agee (2009) describes qualitative questions as the product of a reflective process in which the question is refined as the researcher learns more about the setting and the participants, so the question in the protocol is a starting point that may be sharpened during data collection.

2.2 SPIDER: Sample, Phenomenon of Interest, Design, Evaluation, Research Type

Cooke, Smith and Booth (2012) developed the SPIDER tool to help researchers search for qualitative and mixed-methods studies when preparing evidence syntheses. They kept the idea behind PICO, a short list of parts, and changed the parts to match qualitative work. "Population" became sample, because qualitative studies usually involve small groups chosen for the depth of their experience and do not aim to represent a population statistically. "Intervention" became phenomenon of interest. The remaining letters describe the design (how data are collected), the evaluation (what kind of findings the study seeks) and the research type.

SPIDER was designed as a search tool, and a later comparison found that SPIDER searches returned fewer irrelevant records than PICO-based searches but also missed some relevant studies (Methley et al., 2014). For systematic reviews, the choice of search framework is a question for HSCI 241. For a primary study, SPIDER works well as a checklist that makes sure a qualitative question and its methods are both stated. Select each card to see the element and the Cedar Valley decision.

From these decisions, the team wrote a central question and three sub-questions.

The full SPIDER sentence is long, because it records the design and research type inside the question. In a protocol, many researchers write a shorter central question, "How do adults aged 65 and older who live alone experience social connection after moving to a smaller Cedar Valley community?", and state the design and research type in the methods. Either form is acceptable if all five elements are stated somewhere.

2.3 Other Formats: SPICE, PEO and PCC

SPIDER is one of several formats for questions that PICO fits poorly. Three others appear often in health sciences writing, and the tabs below compare them with the same running case.

2.4 Choosing a Format and Writing the Question

The choice of format follows from what the answer will look like. If the answer will be numbers that compare groups, the question is a PECO or PICO question. If the answer will be numbers that describe a single group, a population and an outcome are usually enough. If the answer will be an account of experiences, views or meanings in words, SPIDER is the usual starting point, and SPICE is a good alternative when the setting and the viewpoint of a particular group are central. Figure 2.1 sets out these choices.

Questions for mixed-methods studies

The Cedar Valley study is a mixed-methods study, so it needs both kinds of question. Its quantitative strand asks the PECO question from Section 1, and its qualitative strand asks the SPIDER question above. Many mixed-methods researchers also write a third, integrating question that states how the two strands will inform each other. The Cedar Valley team wrote: "In what ways do the experiences of older adults who live alone help explain the association between loneliness and emergency department use found in the survey?" Lesson 6 describes the mixed-methods designs that determine when and how the two strands are combined.

Repairing weak qualitative questions

Draft qualitative questions often slip back into the language of measurement or presume their own answers. Open each item to see a weak draft, the problem and a revision.

One strong answer reads: "How do international undergraduate students at a Canadian university (S) describe their decisions about seeking help for stress or low mood (PI), as explored in semi-structured interviews (D) about their views, concerns and experiences with campus services (E), in a qualitative study (R)?" Two sub-questions could ask "What sources of support do students turn to first, and why?" and "How do language, cost, cultural expectations and knowledge of services shape decisions about counselling?" The question avoids assuming that students use services less, which is an empirical claim that the team would need data to support, and it keeps the focus on decisions and experiences.

By the end of this section, the Cedar Valley team has two well-structured questions and an integrating question. Structure, however, says nothing about whether a question is worth the time and resources a study requires. Section 3 introduces the FINER criteria, which the team uses to judge its candidate questions.

3.1 Why Judge a Question Before Studying It

Sections 1 and 2 produced questions that are well structured. A well-structured question can still be a poor choice. It may need more participants than anyone can recruit, it may have been answered convincingly already, or it may expose participants to risks that no ethics board would accept. Judging candidate questions before committing to one finds these problems at the planning stage, when a question is cheap to change.

Hulley and colleagues proposed a widely used set of criteria for this judgment in their textbook Designing Clinical Research (Hulley et al., 2013). A good research question, they argued, is feasible, interesting, novel, ethical and relevant, which gives the acronym FINER. The criteria were written for clinical research and apply equally well to public health, health services and qualitative studies. A team drafts several candidate questions, judges each one against the five criteria, revises the promising ones, and carries the strongest forward, as Figure 3.1 shows.

3.2 The Five Criteria

Feasible

A feasible question is one that the team can actually answer with the people, skills, time, money and data available to it. Hulley et al. (2013) list several parts of feasibility: an adequate number of participants, adequate technical expertise, affordability in time and money, and a manageable scope. Public health studies add a further part, which is access to data. The Cedar Valley PECO question depends on linking survey responses to administrative records, which requires participants' consent and an approved data access request through Population Data BC (Lesson 9). The team's 1,600 completed surveys are enough to compare 392 lonely respondents with 1,208 others, but the time needed for data access approval had to be built into the plan.

Interesting

An interesting question matters to the researcher, to other researchers and to the people who would use the answer. Interest to the researcher sustains the work when it takes longer than planned. Interest to interest holders matters as much: the Cedar Valley question about emergency department visits arose from clinic staff, and the health authority wanted the answer to plan services. Lesson 4 shows how to identify the interest holders whose views should inform this judgment.

Novel

A novel question adds something to what is already known. Hulley et al. (2013) describe a novel study as one that confirms, refutes or extends earlier findings, or that provides new findings. A study of a known association in a new population or setting can be novel, as can a study with better measures or a stronger design than earlier work. Judging novelty requires reading what has been published, and Section 4 gives a short method for finding key papers.

Ethical

An ethical question can be studied in a way that a research ethics board (REB) would approve. In Canada, REBs apply the Tri-Council Policy Statement, known as TCPS 2, which Lesson 5 introduces. At the question stage, the team considers whether the risks to participants can be justified, whether consent and privacy can be secured, and whether community agreements are needed, such as the agreement the Cedar Valley team is developing with the Cedar Valley First Nations Health Centre (Lesson 4).

Relevant

A relevant question has an answer that someone could use. Hulley et al. (2013) describe relevance to scientific knowledge, to clinical practice and health policy, and to future research. In community-engaged work, it also means relevance to the priorities of the communities affected. The Cedar Valley advisory group of six older adults asked that the study include their experience of moving, which raised the relevance of the qualitative question.

3.3 A Scoring Rubric for FINER

Hulley et al. (2013) present FINER as a checklist. This lesson turns it into a scoring rubric so that judgments are recorded and comparable. Each criterion is scored 1 (serious concern), 2 (some concern) or 3 (strong), and each score is justified in one sentence that cites evidence, such as a number, a source or a planned safeguard. The tabs below give the guiding questions and descriptors.

3.4 Interactive: Score a Candidate Question

Paste a candidate question into the scorer below, choose a score for each criterion and write a one-sentence reason for each. The scorer applies the decision rules and assembles a summary of the scores and reasons. Your entries are saved in this browser and included when you download your lesson responses.

3.5 Scoring the Cedar Valley Candidate Questions

The Cedar Valley team drafted four candidate questions. Dr. Hart, the graduate research assistant and the community research associate each scored them independently and then met to agree on the scores below.

Question A scored 2 for feasibility because the linked records depend on an approved data access request, and 2 for novelty because studies in other countries have examined loneliness and health care use among older adults, so the Cedar Valley study would extend that work to a rural Canadian region. Ethics scored 3 because participants consented to linkage and the records would be analyzed in a secure environment.

Question B shows why a single score of 1 outweighs a respectable total. Its total of 11 reflects high interest and relevance, but dementia develops over many years and the study has 12 months of linked follow-up. Early memory changes can also lead people to withdraw socially, so a short study could not separate the two directions of influence. No rewording could fix these problems, so the team set the question aside for a future long-term cohort study.

Question C failed on two criteria. About 70 of the 1,600 survey responses came from Kestrel Lake, which is too few for a meaningful comparison, and the regional survey had no items on social media use. The relationship between internet use and loneliness in older adults has also been studied extensively, so the team replaced the question.

Question D scored 2 for ethics because interviews about loss and isolation may cause distress, so the team planned a distress protocol and a list of local supports (Lesson 5). It scored 3 for novelty because the team found few qualitative studies of older adults who had moved to smaller communities in rural British Columbia.

3.6 Using FINER Scores Well

FINER scores are structured judgments. Two careful people can score the same question differently, and the value of the rubric lies in the reasons it makes them write down. Scoring independently and then discussing differences, as the Cedar Valley team did, exposes hidden assumptions. Each revision should be scored again and the versions kept. Open each item below to see a common scoring error.

A reasonable scoring is feasible 2 (40 nurses are reachable if unit managers agree, but approval is pending), interesting 2 (the topic matters to nurses, but no hospital decision depends on the answer), novel 1 (the association has been studied extensively), ethical 3 (an anonymous survey of adults carries minimal risk) and relevant 2, for a total of 10 with one score of 1. The decision is to revise. One revision ties the question to an unstudied local change: if the hospital has just introduced self-scheduling, the question could compare nurses' sleep quality in the three months before and after the change. Managers would use that answer to decide whether to keep the system, which raises novelty and relevance.

The Cedar Valley team has now chosen its quantitative question (A) and its qualitative question (D). Section 4 turns those questions into a study aim, specific objectives and a hypothesis, and shows how the team found the key papers that supported its novelty scores.

4.1 From Question to Aim, Objectives and Hypotheses

A research question states what the study wants to know. A protocol, a grant application or an ethics application also needs to state what the study will do. Researchers do this with three further statements that follow from the question. The research aim is a single sentence stating the overall purpose of the study. The research objectives are the specific steps that, taken together, achieve the aim, and each one usually corresponds to a set of data collection or analysis activities. A hypothesis is a testable prediction about the relationship between variables, written for a quantitative explanatory question before the data are analyzed. Descriptive and qualitative objectives usually have no hypothesis, because they set out to describe or understand rather than to test a prediction.

4.2 Writing the Aim and Objectives

An aim is usually written as an infinitive phrase that begins with a broad verb, such as "to understand", "to examine" or "to describe". It covers the whole study, so a mixed-methods study has one aim that both strands serve. Objectives are narrower. Each objective names an action, the variables or experiences involved, the people concerned and, where possible, the method and time frame. A useful test is that a reader should be able to tell from each objective what data the team will collect and what the team will do with them.

Many researchers check their objectives against the SMART criteria. Doran (1981) proposed that management objectives be specific, measurable, assignable, realistic and time-related, and later adaptations for research usually read the letters as specific, measurable, achievable, relevant and time-bound. For a research objective, "measurable" means that the objective names something the study will measure or document, and "time-bound" means that it states when or over what period.

The verb in an objective signals the kind of question it serves. Select each card to see the verbs that fit each kind of question from Lesson 1.

The Cedar Valley aim and objectives

The team wrote one aim and four objectives. Objective 1 is descriptive, objective 2 serves the PECO question, and objectives 3 and 4 serve the qualitative strand.

The objectives were written before the survey was analyzed. Objective 1 was later answered directly: 392 of the 1,600 respondents, or 24.5 percent, scored 6 or higher. Objective 2 was checked against the SMART criteria: it is specific (two defined groups and one outcome), measurable (visit counts from administrative records), achievable (the team has consent for linkage and an application under way), relevant (it answers the PECO question that the health authority asked for) and time-bound (the 12 months after the survey).

4.3 Writing Hypotheses

A hypothesis turns an explanatory question into a prediction that the data could contradict. Researchers usually write it as a pair of statements. The null hypothesis states that there is no association or no difference between the groups being compared. The alternative hypothesis states that there is one. An alternative hypothesis is directional when it predicts which group will have the higher value, and non-directional when it predicts only that the groups will differ. A directional hypothesis should rest on earlier evidence or on a stated reason.

Several conventions apply to hypotheses. A hypothesis names the same population, exposure, comparator, outcome and time frame as the question, so that it can be tested with the data the study will collect. It uses the language of association for an observational study. It avoids the word "prove", because data can be consistent or inconsistent with a hypothesis and cannot establish it with certainty. The statistical tests used to weigh the evidence against a null hypothesis belong to HSCI 341 and HSCI 410, and this lesson writes hypotheses without testing them.

Hypotheses are written before the data are analyzed. Kerr (1998) coined the term HARKing, for hypothesizing after the results are known, to describe the practice of presenting a hypothesis developed after seeing the data as if it had been predicted in advance. HARKing makes chance findings look like confirmed predictions. Writing the hypothesis into the protocol, which Lesson 6 describes, creates a dated record of what was predicted. Analyses that the team decides on later are still allowed, but they should be reported as exploratory.

The qualitative strand states no hypothesis. Objective 3 asks how older adults experience social connection, and a prediction about what they will say would work against the open listening that interviews require. Descriptive objectives such as objective 1 also have no hypothesis, because they estimate a quantity without comparing groups.

4.4 Finding Three to Five Key Papers

A research question is written in conversation with earlier research. Before committing to a question, a researcher needs to know enough of that research to judge novelty, to borrow good measures and to explain why the question matters. At the question stage, this means finding three to five key papers: published studies or reports that are most directly useful for the question. This is a short, focused search. Systematic searching, with documented search strategies across several databases and formal screening, is taught in HSCI 241.

What to look for

A good set of key papers usually covers several roles. Open each item to see the role and what the Cedar Valley team found.

How to search

The parts of a structured question supply the search words. Take two or three key terms from your PECO or SPIDER elements, list a few synonyms for each, and combine them in simple searches.

Then work through the following steps.

1. Search Google Scholar, PubMed or the SFU Library catalogue with two or three terms at a time, for example loneliness "emergency department" "older adults".
2. Read titles first and then abstracts, and keep only the papers that address your population, exposure or outcome directly.
3. Prefer peer-reviewed articles and reports from recognized organizations, and note the year and the country of each study.
4. Look for a recent review first, because its reference list leads to the main studies.
5. Use backward citation chasing, which means reading the reference list of a useful paper to find the earlier work it relies on.
6. Use forward citation chasing, which means following the "Cited by" link in Google Scholar to find later papers that cite a useful paper.
7. Stop when you have three to five papers that cover the roles above, and record each one in a short table.

Keep the full reference for each paper as you find it. Lesson 12 shows how to manage references with Zotero and format them in APA 7 style. Lesson 3 uses these key papers to build a causal web, so record any other variables that the papers link to loneliness or to emergency department use, such as age, chronic illness, income or living alone.

4.5 Putting It Together

The four sections of this lesson produce a set of linked statements that a team keeps together as it moves into planning: the structured question, its FINER scores with their justifications, the aim and objectives, any hypotheses, and the key papers. For its quantitative strand, the Cedar Valley team’s set reads as follows: the PECO question from Section 1; its FINER scores (feasible 2, interesting 3, novel 2, ethical 3, relevant 3, total 13, ready) with one sentence of justification for each; the aim and objective 2; the null and alternative hypotheses; and a table of four key papers. For a qualitative question, the set would contain a SPIDER question, its FINER scores, an aim and objectives, and key papers that include at least one qualitative study, with no hypothesis. Lesson 3 uses the key papers to build a causal web and a first directed acyclic graph, and Lesson 6 shows where the question, aims and hypotheses appear in a study protocol.

Bringing It All Together

This lesson followed the Cedar Valley team as it turned a broad worry, that older adults in the region seem lonely, into questions a study can answer. Section 1 structured a quantitative question with PECO, naming a population, a defined exposure, a comparator and an outcome with a time frame, and showed when an assignable intervention calls for PICO instead. Section 2 introduced SPIDER for questions about experiences and meanings, compared it with SPICE, PEO and PCC, and showed how to repair qualitative questions that slip into the language of measurement or presume their answers.

Section 3 moved from structure to judgment. The FINER criteria ask whether a question is feasible, interesting, novel, ethical and relevant, and the rubric in this lesson records each judgment as a score with a reason. A single score of 1 outweighs a high total, which is why the team set aside a question about loneliness and dementia despite its interest and relevance. Section 4 turned the chosen questions into an aim, specific objectives and a pair of hypotheses, and showed how to find three to five key papers that support the novelty and relevance judgments.

Together these steps produce a linked set of statements that a team carries into planning. Lesson 3 turns the key papers into a causal web and a first directed acyclic graph.

The final reflection asks you to apply every part of the lesson to a new community scenario.

Final Knowledge Assessment

Definitions of the terms, formats and people introduced in this lesson, in the order of the lesson's sections.