NR-702B Week 3 The Evidence Matrix: How to Write It

What NR-702B Week 3 asks for

What makes a matrix worth building rather than a chore to complete? The fact that it forces the same questions on every source. A rural critical access clinic evaluating structured medication reconciliation will find its evidence spread across designs that look nothing alike: a multi-site trial, two before-and-after improvement reports, a qualitative study of pharmacist workflow, and a professional society statement. Reading them one after another produces four unrelated impressions. Extracting them into identical fields makes the differences visible, and the differences are where the argument lives, because the moment you can see that the strongest effects came from settings with an on-site pharmacist, your translation problem has come into focus.

The second demand is that extraction be disciplined. Each row records design, sample and setting, what was actually delivered, the outcome and how it was measured, the result in clinical units, the limitations that matter, and a verdict about weight. Students commonly fill the first fields carefully and let the last two decay into a sentence copied from the abstract's conclusion. That inversion is exactly backwards: the descriptive fields are transcription, and the verdict is the doctoral work.

Third, the extra hours in this variant make a properly built matrix realistic rather than aspirational. Reading a study well enough to complete a row honestly takes a substantial part of an hour, and doing that across a set of a dozen or more is a real block of time. Spending it now is efficient across the whole sequence, because a completed matrix carries the synthesis, the intervention decision, much of the proposal and most of the answers to committee questions without being rebuilt.

The boundary that governs every page in this manual. Clinical hours, hour logs, encounter counts, preceptor evaluations, site paperwork and signatures are the student's own record. They are never drafted, reconstructed or estimated with help, and no tutor performs, observes or documents clinical activity for a student. Writing support covers the scholarly layer alone: making an appraisal reach a verdict, keeping fields consistent, and tightening prose. Where real encounters inform your writing, de-identify every detail and describe the setting by type and size rather than by name.

The NR-702B Week 3 method, step by step

Seven moves for building a matrix that does analytic work rather than clerical work.

1. Choose your columns from the decision you have to make

   If the open question is whether the intervention transfers to a small clinic without specialist staffing, then who delivered it is a column. Generic templates produce generic matrices; columns chosen from your own uncertainty produce useful ones.

2. Fill design and sample from the methods section

   Abstracts describe studies generously and titles overstate them. Read the methods for whether there was a comparison group, how allocation happened, how many were enrolled and how many were analyzed after attrition.

3. Record what was actually delivered, in operational detail

   Who did it, how often, for how long, with what support. This column is where translation projects find their answers, and it is the one most often left as the name of the intervention rather than its content.

4. Extract the effect in clinical units with its base

   Percentage point change with numerators and denominators, events avoided per hundred, minutes saved. Where the source reports an interval, carry it, because precision travels with the estimate and disappears the moment you round to a headline.

5. Apply one named appraisal tool across every row

   Different designs need different instruments, but the family should be consistent and named. Using a structured tool means your quality judgments can be checked against a standard rather than read as preference.

6. Write the verdict column as weight for your project

   Three tiers work well: carries the argument, supports with qualifications, informs feasibility only. Add a clause saying why, because the reason is what a reader will use when they disagree with your tier.

7. Read down each column and note the patterns

   Where do the strong designs cluster, which settings are missing, which outcome measures repeat, where does effect size covary with intensity. Those observations are the raw material for next stage's synthesis and are worth writing down as you go.

A layout and word budget for a matrix deliverable

Our frame for the written material that surrounds a matrix, sized for roughly 1,200 to 1,500 words alongside the table itself. It is our own outline rather than anything the university issues, and your week's rubric outranks it wherever they disagree. If your program supplies a matrix template, use its fields and keep this narrative structure around it.

Evidence craft for matrix and appraisal writing

Name the design from the methods, then use verbs that match it. Observational work supports was associated with. Reduced, improved and prevented belong to designs where an intervention was assigned and compared. Mismatched verbs in a matrix are especially visible because they sit beside the design column that contradicts them.

Keep fields consistent even when a study is silent. Where a source does not report attrition, the cell says not reported rather than being left blank. Blank cells read as incomplete extraction; not reported is a finding about the study and often belongs in your quality judgment.

Do not let the abstract write your outcome column. Reported conclusions are frequently stronger than the results table supports, particularly in small improvement reports. Extract from the results, and where the authors' conclusion overstates their data, note it in the appraisal column rather than repeating it.

Attribute the appraisal tool and any grading system. Levels of evidence hierarchies and quality instruments are published by named bodies in named versions. Cite them, and apply the categories as they are defined rather than as you remember them, since a misapplied level is easy for a reader to catch.

Treat quality improvement reports as their own category. They rarely establish effect and often carry the richest implementation detail. Say explicitly that you are weighting them for feasibility rather than for effect size, which keeps your hierarchy honest while still using what they offer.

Five mistakes that cost points in this week's territory

• A matrix with no verdict column. Without a judgment field the table is a set of summaries, and the appraisal the stage exists to teach has not happened.
• Conclusions transcribed from abstracts. The outcome column filled from the authors' framing rather than their results table transmits their overreach into your project.
• Inconsistent fields. Rows extracted at different depths make column reading impossible, which removes the only analytic advantage a matrix has.
• Effect sizes reported as significance. A p value in the effect column cannot tell a reader whether the change is worth the disruption it will cause in a clinic.
• Delivery detail omitted. A matrix that records what an intervention was called but not what was actually done cannot support a translation decision at all.

Before you submit

• Every column earns its place against a decision you have to make
• Design and sample were taken from the methods section, not the abstract
• The delivery column records who did what, how often and with what support
• Effects appear in clinical units with bases and, where available, precision
• A named appraisal tool was applied consistently and is cited
• Every row ends in a weight verdict with a reason clause
• Fields say not reported rather than being left empty
• A written paragraph reports what reading down the columns revealed

Questions students ask about this stage

Keep going