Generating Evidence Through Research

Unit 9 named four domains of evidence use, and this is the first of them — research — approached from the side that generates it rather than the side that reads it. The relationship to Part 1 is the thing to hold. Appraising a study and designing one are the same knowledge viewed from opposite ends. Every question unit 6 taught you to ask of someone else's trial is a question you now have to answer in advance about your own. That is why this unit references units 4, 6 and 7 rather than repeating them — without that restraint it would have duplicated about a quarter of Part 1. By the end you can say how appraising and designing relate; why inclusion criteria are set before recruitment; why ethics review is a substantive judgement rather than paperwork; why uneven dropout changes the answer rather than shrinking the sample; and what randomisation does that careful matching cannot.

Five videos. This is unit 9's first domain — research — approached from the generating side, and the relationship to Part 1 is the thing to hold: appraising a study and designing one are the same knowledge viewed from opposite ends. Unit 6 was the preparation for this unit, and this unit does not repeat it.

Where this sits in the four domains.

EBP 3151 The Fourth Domain

Unit 9 named four domains of evidence use. This is the first one — research — approached from the generating side.

The relationship to Part 1.

EBP 3152 Appraising and Designing Are Different Views

Appraising a study and designing one are the same knowledge from opposite ends, which is why unit 6 was the preparation for this.

Where a study actually starts.

EBP 3153 The Design Meeting

A study begins in a conversation about what is being asked, before any method is chosen.

Choosing the outcome.

EBP 3154 Choose Outcomes That Matter to Someone

Outcomes are chosen because they matter to patients or clinicians, not because they are convenient to collect.

The rule that governs everything.

EBP 3155 The Design Follows the Question

The question decides the design, which is unit 2's question types arriving where they have consequences.

Three videos on precision. Inclusion criteria are set before recruitment so they cannot be shaped by who turns up, and the instrument has to capture the thing you mean — which is unit 6's validity question, asked before the study rather than after it.

Who is in the study.

EBP 3156 Inclusion and Exclusion Criteria

Who is eligible, set before recruitment so the criteria cannot be shaped by who turns up.

What gets measured.

EBP 3157 Variables

What is being measured and what is being changed, named precisely enough to be repeated.

Validity from the designer's side.

EBP 3158 Measures Must Measure What Is Intended

The instrument has to capture the thing you mean, which is unit 6's validity question asked before the study rather than after.

Four videos. Some questions cannot be asked whatever their scientific interest, the risk to participants cannot exceed the expected benefit, and review happens before the study runs rather than as a submission afterwards.

The first constraint.

EBP 3159 A Question Has to Be Ethical

Some questions cannot be asked, whatever their scientific interest, and that is settled first.

The balance required.

EBP 3160 Risk Cannot Exceed Benefit

The risk to participants cannot exceed the benefit expected, and the judgement is made before recruitment.

When review happens.

EBP 3161 Review Before the Study Runs

Human subjects review happens before the study runs rather than as a submission afterwards.

The other legal requirement.

EBP 3162 Privacy Regulation Applies

Health privacy regulation governs what can be collected and how it is held.

Five videos. The one to carry is attrition — if the people who leave differ from those who stay, dropout changes the answer rather than shrinking the sample, and the remaining group is answering a different question from the one you asked.

What can undermine the answer.

EBP 3163 Threats to Internal Validity

The named ways a study's internal validity can fail, each addressed by design rather than in the discussion section.

The first threat.

EBP 3164 Attrition

People leaving the study, which is planned for rather than discovered.

Why attrition matters more than it looks.

EBP 3165 Uneven Dropout Changes the Answer

If those who leave differ from those who stay, attrition changes the answer rather than shrinking the sample.

Two further threats.

EBP 3166 Maturation and Observer Drift

People change over time regardless of the intervention, and observers change how they measure.

The mechanism, restated from unit 6.

EBP 3167 Randomization Controls What You Did Not Think Of

Randomisation distributes the factors nobody identified, which is what no amount of careful matching achieves.

Four videos referencing rather than re-teaching unit 6's design hierarchy, and closing on where the research process ends — in dissemination, which is unit 21's step six arriving from the other direction.

The weaker designs, referenced.

EBP 3168 Pre-Experimental and Nonexperimental Designs

The designs that cannot support a causal claim, referenced here rather than re-taught from unit 6.

One of them named.

EBP 3169 Case-Control, Looking Back

Looking backward from outcome to exposure, which suits etiology questions and cannot establish cause.

The process as a whole.

EBP 3170 The Research Process Has Steps

Research follows an ordered process, and the chapter's enumeration of it is partly illegible — flag A-57.

Where it ends.

EBP 3171 It Ends in Dissemination

The research process ends in dissemination, which is unit 21's step six arriving from the other direction.

Three videos. The second is worth stating plainly: ethics review decides whether the question should be asked at all, which makes it a substantive judgement rather than a form to submit.

The first error.

EBP 3172 Misconception: Research and EBP Are the Same Work

Research generates generalisable knowledge; evidence-based practice translates existing evidence into a decision. Different aims, methods and obligations.

The second.

EBP 3173 Misconception: Ethics Review Is Paperwork

Ethics review decides whether the question should be asked at all, which is a substantive judgement rather than an administrative step.

The third, and the technical one.

EBP 3174 Misconception: Dropouts Only Shrink the Sample

If those who left differ systematically from those who stayed, the remaining sample answers a different question from the one asked.

Ten videos closing the unit, added after a re-shot title page revealed an objective this course was not serving anywhere. The distinction that carries the strand is the one most often collapsed: random sampling supports generalising a finding outward, random assignment supports claiming the intervention caused it, and a study can have either without the other.

Sampling, as a design decision.

EBP 3181 Sampling Is a Design Decision

How participants are selected is decided in advance, and it determines who the answer can be said to apply to.

The design that supports generalising.

EBP 3182 Random Sampling

Drawing so that every member of a population has a known chance of selection, which lets a result generalise back to it.

The distinction that catches people.

EBP 3183 Sampling Is Not Assignment

Sampling supports generalising outward; assignment supports the causal claim. A study may have either without the other.

Sampling within subgroups.

EBP 3184 Stratification

Dividing a population into subgroups and sampling within each, so a small subgroup is still represented.

Grouping before assignment.

EBP 3185 Blocking

Grouping participants alike on some variable before assigning, so that variable cannot explain the difference between arms.

The two combined.

EBP 3186 Randomized Block Design

Forming blocks of similar participants and randomising within each, combining blocking's control with randomisation's reach.

How many participants.

EBP 3187 Sample Size Follows the Question

How many follows from the size of effect worth detecting, not from what is convenient to recruit.

The limit every sample sets.

EBP 3188 Who the Sample Excludes

Every sampling decision excludes someone, and the excluded group defines the limit of what the study can claim.

The error worth naming.

EBP 3189 Misconception: A Bigger Sample Fixes a Biased One

Size addresses precision. Only selection addresses bias, so more of a skewed sample is a more precise wrong number.

The strand assembled.

EBP 3190 Choosing a Sampling Design

One sampling design chosen to support the claim intended, stratified where it matters, blocked where it helps, with exclusions written down.

One procedure designing a study that would survive the appraisal you were taught to perform.

The unit assembled.

EBP 3175 Designing a Study That Could Be Appraised

One study designed so that a reader applying unit 6's three questions would accept it.

That is generating evidence. Against the outcomes. Appraising and designing are the same knowledge from opposite ends. The design follows the question — unit 2's question types arriving where they finally have consequences — and every appraisal question you learned is now something you answer in advance. Inclusion criteria are set before recruitment so they cannot be shaped by who happens to turn up. The same logic runs through the whole unit: decisions made before the data arrive cannot be adjusted by the data. Ethics review decides whether the question should be asked at all. That is a substantive judgement about risk against benefit, made before recruitment — not a form submitted once the design is settled. And attrition changes the answer rather than shrinking the sample. If the people who left differ systematically from those who stayed, what remains is answering a different question from the one you set out to ask — which no increase in sample size repairs. Randomisation distributes the factors nobody identified. Matching and adjustment handle the confounders you thought of; randomisation handles the ones you did not, and that is the whole of why it licenses a causal claim. Unit 23 funds it.