1.1 What a Study Design Is

A study design is the overall plan for how a study gathers and compares information to answer its question. It specifies who will be studied, what will be measured, when each measurement will be taken, and whether the researchers will change anything about the participants' circumstances. Designs that share the same basic logic form a design family. Cohort studies, case-control studies and cross-sectional studies, for example, belong to the analytic observational family, because each compares groups of people without the researchers assigning anything to them.

Lessons 1 and 2 gave you the two inputs this section needs. Lesson 1 sorted research questions into four kinds: descriptive (what is happening and how often), explanatory (why something happens or whether one factor affects another), predictive (who is likely to experience something in the future) and exploratory (how people experience something that is not yet well understood). Lesson 2 gave the question a structure. A PECO question (population, exposure, comparison, outcome) or a PICO question (population, intervention, comparison, outcome) already contains much of its design, because the comparison signals that groups will be compared and an intervention signals that someone will deliver something. A SPIDER question (sample, phenomenon of interest, design, evaluation, research type) points toward qualitative methods.

The principle that organizes this section is that the question chooses the design. Beginning researchers often start with a familiar method, such as a survey, and then look for a question it can answer, which tends to produce findings that do not match what the team or its partners needed to know. A second common error is to treat designs as a ranked list headed by the randomized controlled trial. A well-conducted trial provides the strongest evidence about the effect of an intervention that can ethically be assigned, and it cannot tell you how common loneliness is or how older adults experience a move to a new town. The most suitable design fits the question and can be carried out well with the resources available.

1.2 A Decision Map in Five Steps

A decision map is a sequence of questions that leads from a research question to a design family. The version in the figure adapts the classification of study types published by Grimes and Schulz (2002) and adds steps for evidence synthesis, qualitative questions and mixed methods. Work through the steps in order and stop at the first exit that fits your question.

Step 1: Do existing studies already answer the question?

Before planning new data collection, check whether the answer already exists. If many studies have addressed the question, a systematic or scoping review may be the most useful project. The three to five key papers you learned to find in Lesson 2 are usually enough to tell whether a new study is needed.

Step 2: Does the question ask about numbers, about meanings, or about both?

Questions about how many, how much, how often, or how strongly two factors are related call for quantitative data. Questions about how people experience something, what it means to them, or how a process unfolds call for qualitative data. Questions with parts of both kinds call for mixed methods (Section 2). Two Cedar Valley questions show the split. "What proportion of older adults are lonely?" asks for a number, while "How do older adults living alone experience social connection after a move to a smaller town?" asks about meaning and experience.

Step 3: Is the exposure or intervention assigned on purpose, and is the assignment random?

Some exposures and interventions are assigned on purpose, either by the research team or by another agent, such as a government, a health authority or a program. Studies of deliberately assigned interventions form the experimental family. When the assignment uses a random process, such as a computer-generated allocation sequence, the study is a randomized controlled trial. When the assignment is deliberate and not random, the study is quasi-experimental. This includes a team offering a program at clinics it selects and comparing them with other clinics, and it also includes natural experiments and policy evaluations, in which a government or program decides who is exposed and the researchers compare groups or time periods, for example before and after a new policy. If no one assigns the exposure and the team only records exposures that people already have, such as loneliness or income, the study is observational, and you move to Step 4.

Step 4: Does the study compare groups to estimate an association?

An observational study with no comparison group is descriptive. It describes how often a condition occurs and how it is distributed by person, place and time. Prevalence surveys, case reports, case series and surveillance systems belong here. An observational study that compares groups to estimate whether an exposure is associated with an outcome is analytic, and you move to Step 5.

Step 5: What is the unit, and when are exposure and outcome measured?

If the units of analysis are groups, such as health regions or neighbourhoods, the study is ecological. If the units are individuals, the timing of measurement separates the three main designs. A cross-sectional study measures exposure and outcome at the same time. A cohort study starts with people who differ in their exposure and follows them forward to see who develops the outcome. A case-control study starts with people who have the outcome (cases) and people who do not (controls), and looks back to compare their earlier exposures.

Predictive questions use the same map with a different purpose: a prediction study estimates who is at risk of an outcome without asking whether each predictor causes it. Such studies usually draw on cohort data, because the predictors must be measured before the outcome occurs.

1.3 The Design Families at a Glance

The tabs summarize each family, the questions it answers, an example, and the course in which you will study it in depth.

1.4 Constraints That Narrow the Choice

The decision map tells you which designs could answer a question. Ethical and practical constraints then tell you which of those designs you can carry out. Select each card to read about one of six constraints that researchers weigh at this stage.

Earlier lessons add two further considerations. Interest holders may value some questions more than others (Lesson 4), and where a study involves First Nations, Inuit or Métis communities, the design must respect the community's authority over its data and the agreements reached with it.

1.5 Applying the Map to the Cedar Valley Study

Two features of the table deserve attention. First, the main questions fall in different families, which is why the Cedar Valley study uses mixed methods (Section 2). Second, the PECO question becomes a cohort design because the team links each survey response to emergency department visits that occur after the survey date. If the team compared loneliness with visits in the year before the survey, the same data would support only a cross-sectional comparison, and the team could not tell whether loneliness came first. Decisions about timing of this kind are written into the protocol (Section 3).

2.1 What Makes a Study Mixed Methods

Mixed-methods research combines quantitative and qualitative data within one study and integrates them to answer the research questions. Creswell and Plano Clark (2018) describe its core characteristics: the researcher collects and analyzes both kinds of data rigorously, integrates the two forms of data and their results, organizes these procedures into a recognizable design, and frames the work within theory and a stated philosophical position. Each part of the study that uses one kind of data is called a strand. The Cedar Valley study has a quantitative strand (the survey, the linked administrative data and the chart review) and a qualitative strand (the interviews and focus groups).

Integration is the feature that separates a mixed-methods study from a study that happens to contain two kinds of data. A study that runs a survey and a set of interviews and reports them in separate chapters, without relating one to the other, has collected two kinds of data and has not yet produced a mixed-methods analysis. Integration means that the results of one strand shape the other strand, or that the two sets of results are brought together and compared, so that the study as a whole says more than either strand could say alone.

Why combine the two kinds of data?

Greene, Caracelli and Graham (1989) reviewed published mixed-methods evaluations and identified five purposes for combining methods. A clear statement of purpose helps a team choose its design, because each purpose implies a different sequence. Select each card for a definition and a Cedar Valley example.

2.2 Timing, Priority and Notation

Two decisions describe the structure of any mixed-methods study. The first is timing: the strands may run at the same time (concurrently) or one after the other (sequentially). The second is priority, sometimes called weighting: the strands may carry equal weight in answering the study's questions, or one strand may be the main source of evidence while the other plays a supporting role. Morse (1991) introduced a short notation for these decisions that is still widely used.

2.3 Three Core Designs

Creswell and Plano Clark (2018) describe three core designs that combine timing, priority and integration in recognizable ways: the convergent design, the explanatory sequential design and the exploratory sequential design. Most mixed-methods studies in the health sciences use one of them, either alone or as a building block within a larger study. The figure shows the order of steps in each, and the tabs describe when to choose each design and what it demands of the team.

Many health studies embed these core designs in larger projects. A randomized trial may include a qualitative process evaluation, a program evaluation may move through several sequential phases, and a community-based participatory project may combine strands at each stage of a partnership. HSCI 826 Lesson 5 applies the three core designs to program evaluation.

2.4 Integration and Joint Displays

Fetters, Curry and Creswell (2013) describe integration at three levels. At the design level, the choice among the three core designs sets when and how the strands meet. At the methods level, the strands are linked through four procedures, described in the items below. At the interpretation and reporting level, the team presents the integrated results in a narrative that weaves the two kinds of findings together, through data transformation (for example, counting how many participants raised a theme), or in a joint display.

A joint display is a table or figure that places quantitative and qualitative results side by side, organized by topic, together with the conclusion drawn from reading them together. That conclusion is called a meta-inference, an overall inference that integrates the inferences from both strands (Teddlie & Tashakkori, 2009). Fetters and colleagues (2013) describe three ways the strands can fit: confirmation, when both lead to the same conclusion; expansion, when they agree in part and one adds something the other lacks; and discordance, when they disagree. Discordance is a result to be reported and examined, because it often points to a measurement problem, a subgroup difference or a new question. The table below is an illustrative joint display for the Cedar Valley study; Lesson 12 returns to the design of joint displays for a report.

2.5 Choosing a Design for the Cedar Valley Study

The choice has practical consequences that later sections take up. The ethics application must describe both phases, include a draft interview guide, and state that the final guide will be submitted as an amendment (Section 3). The survey consent form must include optional consent to be contacted again and to have survey answers linked to health records (Lesson 5). The timeline must place the interviews after the survey closes, which makes the qualitative strand part of the project's longest chain of tasks (Section 4).

3.1 What a Protocol Is and Who Reads It

A research protocol is the written plan for a study. It records the decisions the team has made about the question, the design, the participants, the data, the analysis and the safeguards, in enough detail that another qualified researcher could carry out the study as intended. Sections 1 and 2 produced two of those decisions for the Cedar Valley study: the design family for each question and the mixed-methods design that joins them. The protocol is where such decisions are written down, dated and kept together.

A grant proposal and a protocol share much of their content, and teams often adapt one from the other. They serve different purposes. A proposal argues that a study deserves funding, so it emphasizes the importance of the problem, the strength of the team and the value of the expected results. A protocol tells the people who will carry out, approve and oversee the study exactly what will happen, so it emphasizes procedures: who is eligible, how they are recruited, what is measured and when, how data are stored, and how they will be analyzed. A funded proposal usually needs to be expanded into a full protocol before data collection begins.

Several groups read a protocol, and each reads it for a different reason. Select each card to see what one group looks for.

The protocol also serves as the source document for the other papers a project produces. Each of the documents in the figure draws on the same decisions about design, participants, data and analysis. When those decisions are written once in the protocol and copied from there, the documents stay consistent with one another. When each document is written from memory, small differences appear, and an REB or a data steward may ask the team to explain them.

3.2 The Sections of a Protocol

No single format is required for every protocol. Funders, REBs and journals each publish templates, and the template you are given should take precedence. For randomized trials, the SPIRIT 2025 statement (Chan et al., 2025), which updated the 2013 statement, lists the items a trial protocol should contain. For systematic reviews, PRISMA-P (Moher et al., 2015) serves the same function, and HSCI 241 Lesson 2 teaches it. Protocols for observational, qualitative and mixed-methods studies usually contain the twelve sections described below. Open each item to see what the section holds, an example from the Cedar Valley protocol, and the lesson of this course that prepares you to write it.

What the design and conceptual framework section contains

Card 4 deserves a closer look, because it is where a protocol explains how theory shaped the study. The section usually has three linked parts. The first is the guiding theory or perspective, a general account of why the outcome occurs that the team draws on to choose its questions and variables. The second is a conceptual framework, which names the constructs the study will measure or explore and states how the team expects them to relate to one another. The third is the causal diagram derived from the framework, which turns the expected relations into explicit assumptions about which variable affects which, and therefore which variables must be measured.

In the Cedar Valley protocol, the guiding perspective is the web of causation together with Krieger’s (1994) critique of it, which led the team to treat income, transport and rural residence as causes in their own right (Lesson 3). The conceptual framework is the causal web, which names loneliness, its individual, relationship and structural causes, and the pathways through depression and physical inactivity to emergency department visits. The causal diagram is the DAG from Lesson 3, with its justification table. For further reading, HSCI 230 Lesson 7 Section 1 (Theory Before Instruments) shows how frameworks shape what a study measures, and HSCI 841 Lesson 2 Section 2.6 describes how theory enters a qualitative study through sensitizing concepts and conceptual frameworks.

3.3 Worked Example: A One-Page Protocol Summary

A full protocol for a study like Cedar Valley may run to twenty or thirty pages with appendices. Most teams also keep a one-page summary for partners, new team members and committee meetings. The summary below follows the twelve sections in compressed form. Writing a summary of this kind early is a useful test: any row that you cannot fill in identifies a decision that has not yet been made.

3.4 Versions, Amendments, Registration and Publication

Version control for protocols

A protocol changes as a study develops, and every reader needs to know which version is current. A simple convention works for most projects. Working drafts carry decimal numbers (0.1, 0.2 and so on). The first version submitted to the REB becomes version 1.0. Drafts prepared after that carry numbers such as 1.1 and 1.2, and the next version submitted for approval becomes 2.0. Each version is saved as a separate file, and the protocol keeps an amendment log that records every approved change.

A protocol amendment is a change to research that the REB has already approved. Under TCPS 2, researchers submit proposed changes to the REB and wait for approval before making them, with an exception for changes needed to remove an immediate risk to participants. Adding a survey question, changing recruitment materials, adding a site and changing how data are stored all require an amendment. The amendment log for the Cedar Valley protocol shows how changes are recorded.

Registration and publication

Registering a protocol creates a public, dated record of what a study planned to do. Clinical trials are expected to be registered in a public registry, such as ClinicalTrials.gov or a registry in the World Health Organization's network, and journals that follow the recommendations of the International Committee of Medical Journal Editors consider a trial for publication only if it was registered at or before the enrolment of its first participant. Registration is optional for most observational, qualitative and mixed-methods studies, but many teams preregister them, for example on the Open Science Framework, so that readers can compare the final report with the plan. Some journals, including BMJ Open and JMIR Research Protocols, publish full protocols.

Both practices respond to a documented problem. Chan and colleagues (2004) compared published trial reports with their protocols and found that outcomes were often reported selectively, with outcomes that showed clear differences more likely to appear in the published report. A public protocol lets readers see whether all planned outcomes were reported and whether any change to the plan was explained.

Students who have taken HSCI 241 met registration for reviews, in PROSPERO and on the Open Science Framework, in its Lesson 1 Section 4.3, and HSCI 230 Lesson 1 (Research Integrity and Reform) explains the problems that registration addresses. Neither course is required before this one, so this section stands on its own.

4.1 Timelines and Gantt Charts

Project management is the set of practices that keeps a study on schedule, within budget and in control of its files and data. This section applies five of them to the Cedar Valley study: timelines, roles, budgets, file organization and data management.

A timeline begins with a work breakdown structure, a list of everything the project must produce, broken into tasks small enough that someone can estimate how long each will take. The team then works through five steps.

1. List the deliverables, such as the approved ethics application, the cleaned survey file and the final report, and break each into tasks.
2. Estimate how long each task will take, and add time to the estimate. People routinely underestimate how long their own tasks will take, even when similar tasks have run late before, a pattern known as the planning fallacy (Buehler et al., 1994).
3. Identify the dependencies between tasks. The most common is a finish-to-start dependency, in which one task cannot begin until another ends: the survey pilot cannot start until the REB has approved the study.
4. Place the tasks on a calendar and mark the milestones, the dated points at which a key result is reached, such as REB approval or the close of the survey.
5. Find the critical path and protect it with buffer time.

A Gantt chart displays the result as horizontal bars on a calendar, one bar per task. The format is named after Henry Gantt, who developed bar charts for scheduling work in the 1910s. The critical path is the longest chain of dependent tasks from the start of the project to its end. Its length sets the shortest time in which the project can be finished, so any delay to a task on the critical path delays the whole project. Tasks off the critical path have slack, an amount of time by which they can slip without delaying the finish.

The chart shows the consequences of the design chosen in Section 2. Because interview participants are selected from survey respondents, the qualitative strand cannot begin until the survey closes in month 10, so both strands sit on the critical path. The data access request and the linkage run alongside them with some slack: a delay of two months could be absorbed, and a longer one would push back integration and the final report.

Three tasks depend on decisions made by others: the REB review, the data access request and the agreements with the six partner clinics. The team starts them early, builds buffer time after them, and plans work it can do while it waits, such as building the survey during the ethics review. It reviews the chart at each monthly meeting.

4.2 Roles and Responsibilities

Many problems in research teams come from unclear responsibility. A RACI matrix assigns one of four roles to each person for each task. The person responsible (R) does the work. The person accountable (A) answers for the task being completed and makes the final decision, and each task has exactly one. People who are consulted (C) give input before decisions are made, and people who are informed (I) are kept up to date.

The matrix sits inside a short team charter that records how often the team meets, how decisions are made, how members communicate, who may access which data, and how authorship will be decided. Under the criteria of the International Committee of Medical Journal Editors, an author has made a substantial contribution to the design of the work or to the collection, analysis or interpretation of its data; has drafted the work or revised it critically; has approved the final version; and has agreed to be accountable for the work. Advisory group members and community partners who meet these criteria are authors, and those who do not are named in the acknowledgements. A research agreement with a First Nations partner may also set out how the partner reviews publications before submission (Lesson 4). Agreeing on these matters at the start is usually easier than settling them when a manuscript is ready.

4.3 Budgets

A research budget lists the expected costs of the project, line by line, and a budget justification explains how each amount was calculated and why the expense is needed. Reviewers use it to check that the budget matches the protocol: 24 interviews in the protocol should mean 24 honoraria in the budget and enough staff time to transcribe 24 recordings. Each line shows its calculation as a quantity multiplied by a unit cost. The budget below is illustrative; real rates come from salary scales, vendor quotes and the funder's rules.

A justification turns each line into a short argument. For transcription, the Cedar Valley justification reads: "Four focus groups of about 90 minutes will produce about 360 minutes of audio. Overlapping voices defeat most automated tools, so professional transcription at $2.00 per audio minute, costing $720, is needed for the groups. The interviewers will transcribe the 24 interviews from drafts produced by speech recognition software on an encrypted university laptop, so that work is costed under personnel. Clean verbatim transcripts are needed for the first-cycle coding described in the analysis plan." The justification also notes in-kind contributions, resources that partners provide without charge: the university hosts the REDCap survey platform, the partner clinics provide meeting rooms, and a health authority analyst advises on the administrative data.

Three checks prevent common problems. Read the funder's rules on eligible expenses (for the federal granting agencies, the Tri-Agency Guide on Financial Administration). Ask your research office whether salaries must include a percentage for employee benefits. Confirm that honoraria match the amounts approved by the REB, which checks that payment does not pressure people to take part (Lesson 5).

4.4 File Organization and Naming

A two-year project with several staff produces thousands of files. A folder structure agreed at the start lets anyone find a file without asking, and it keeps identifiable information away from everyday working files. The Cedar Valley structure numbers its folders in the order the work happens, keeps raw data read-only so the original export can always be recovered, and holds the linking key, consent forms and audio in a separate restricted location. Lesson 5 explained why identifiers are stored apart from responses.

A file naming convention is a fixed pattern for file names that every team member follows. The Cedar Valley convention is date, project, content and version, separated by the underline character (_), as in 2026-05-04_CVSCS_interview-guide_v03.docx. A name such as Interview guide FINAL revised (2).docx gives no date, sorts unpredictably and leaves readers unsure which file is current. HSCI 410 Lesson 1 numbers analysis datasets with a two-digit suffix, such as bp01.csv, and records each version in a file log, which is compatible with this convention. Select each card for one of the rules behind the convention.

4.5 Version Control

Version control is any system that records the changes made to a file over time, so that the team can see what changed, who changed it and why, and can return to an earlier version. Most projects use three systems, one for each kind of file.

4.6 The Data Management Plan

A data management plan (DMP) is a living document that describes how a project's data will be collected, documented, stored, protected, shared and preserved, during the project and after it ends. Lesson 5 introduced the DMP with a focus on privacy; here it becomes an operating document. The Tri-Agency Research Data Management Policy (2021) asks Canadian institutions to develop data management strategies and requires DMPs for some funding opportunities. The DMP Assistant, a free bilingual online tool supported by the Digital Research Alliance of Canada, provides templates.

Two sets of principles guide the plan. The FAIR principles ask that data be findable, accessible, interoperable and reusable, so that they can be found and used again with appropriate permissions (Wilkinson et al., 2016). The CARE principles for Indigenous data governance (collective benefit, authority to control, responsibility and ethics) add attention to the people and purposes behind the data (Carroll et al., 2020). In Canada, the CARE principles sit alongside the First Nations principles of OCAP®, which Lesson 4 turned into agreements. Open each heading below to see what the section of a DMP covers and how the Cedar Valley one-page plan completes it.

The team reviews the plan at each milestone and whenever the protocol changes. When telephone interviews were added in protocol version 3.0, it added a rule that they are recorded only on university-approved devices.

Bringing It All Together

This lesson has moved from a research question to a study that a team can carry out. The decision map in Section 1 asks five questions in order: whether existing studies already answer the question, whether it asks about numbers or meanings, whether the exposure is assigned on purpose and at random, whether groups are compared, and what the unit and timing of measurement are. Ethical and practical constraints then narrow the choice. For the fictional Cedar Valley Social Connection Study, the map placed the prevalence question in the descriptive family, the PECO question in a cohort design built on linked records, and the SPIDER question in the qualitative family.

Because the Cedar Valley questions fall in different families, the study needs a mixed-methods design. Section 2 compared the convergent, explanatory sequential and exploratory sequential designs and showed that integration, through connecting, building, merging and embedding, is what makes a study mixed methods. The team chose an explanatory sequential design (QUAN → QUAL), and that choice shaped everything that followed: the consent to be contacted again, the amendment for the final interview guide, and the critical path of the timeline.

Sections 3 and 4 turned the design into a plan. The protocol records the decisions in one version-controlled document that supplies the ethics application, the data access request, the data management plan and the methods section. The Gantt chart, the RACI matrix, the justified budget, the folder structure and naming convention, version control and the data management plan keep the project on schedule, within budget and in control of its data.

The final reflection asks you to apply the whole lesson to a new study, from the choice of design to the data management plan.

Final Knowledge Assessment

These terms, frameworks and people appear in this lesson on choosing a design and managing a research project.