know.

CookbookResearch and civic lifeNo. 46

A literature reviewpapers read, compared and synthesised into a chapter

Recipe No. 46 · Research and civic life

For: A PhD student with a literature review chapter due

You get: A section per paper, evidence tables, and your own synthesis

Time: An evening, then twenty minutes a paper

See it in know.sh

An evidence table of seven studies, drawn by Claude from the corrected sections: a row per study naming its section, the key paper marked. The papers are invented.

Green space and adolescent mental healthMeasuring green space exposure

15of 16

Evidence tableNDVI in a buffer, 7 studies

Insight, 1 note, 190 words

The seven studies in this theme that measure exposure as mean NDVI in a buffer round the home, in section order. Rows name the section they came from; check a cell there before quoting it.

Five of seven report an association in the same direction; Haugland and Kowalczyk find none. Worth a Question: what do those two share that the others do not?

One paper as a section after the student’s pass: the DOI as its source, the four parts checked, and My reading in the student’s words. The paper is invented.

know., Measuring green space exposure, 6 · Look up, New, ⋯

Green space and adolescent mental health, Measuring green space exposure

6 of 14

Okafor 2023: cohort, NDVI within 500 m and SDQ scores

Evidence, key section, 2 highlights, 1 note, 1 source, 356 words

Highlight, Note, Edit

Okafor, Brandt and Lindqvist (2023). Journal of Environmental Epidemiology Studies. doi.org/10.5555/jees.2023.0412

Methods. Prospective cohort. Exposure is mean NDVI in a 500 m buffer round the home address at ages 11 and 14; outcome is the parent-reported Strengths and Difficulties Questionnaire (SDQ) emotional subscale at 14 and 17. Adjusted for household income, parental education and urbanicity.

Sample. 4,212 adolescents in four Scandinavian cities; 71 per cent retained at 17.

Findings. An interquartile rise in NDVI was associated with a 0.3-point lower emotional score at 17, stronger in the most deprived quintile. No association for conduct problems.

Limitations. Residential NDVI says nothing about whether anyone went outside. Parent report may miss internalising symptoms.

My reading. The deprived-quintile result rests on 380 young people and the confidence interval nearly touches zero. Useful for the equity theme, but not as strong as the abstract makes it sound.

You, 19 September

Compare with Reyes 2021 (section 9), which used distance to the nearest park and found nothing. Is the difference the measure or the setting?

Claude reads the theme through the know.sh connector, adds the tables as new sections and says which cells it could not fill.

Claude, connected to know.sh

You

Using know.sh, read every Evidence section in the document “Measuring green space exposure” on my shelf “Green space and adolescent mental health” and build an evidence table element, one row per study: author, year and section number, design, sample size and finding. Add it as a new Insight section at the end. Do not change the sections.

document_list_sections Measuring green space exposure: 14 sections, 12 of them Evidence

section_get Read sections 1–12

section_add Added “Evidence table: NDVI in a buffer, 7 studies” to Measuring green space exposure, 15

section_add Added “Evidence table: distance to a park and time outdoors, 5 studies” to Measuring green space exposure, 16

Claude

Twelve rows is more than one table reads well, so I split them by exposure measure: section 15 has the seven studies using NDVI in a buffer (250 m to 1 km), section 16 the three using distance to the nearest park and the two using self-reported time outdoors. Two rows need your eye: section 4 gives no sample size, so I wrote “not reported”, and section 11 calls its design “longitudinal” without saying how many waves. I have not looked anything up outside your library.

How it works

Your supervisor asks which cohort studies measured green space by satellite and which by distance to a park, and you open six PDFs to answer. Your assistant drafts a section per paper, you check each against the source, and an evidence table answers that question at a glance. The synthesis at the top of each theme is yours.

  1. Hand over the papers, theme by theme

    Give your assistant ten PDFs and ask for an Evidence section per paper, titled author, year and design, with its DOI as the source: methods, sample, findings, limitations, and My reading left empty.

  2. Read each paper and correct its section

    With the paper open, check every number against the source, rewrite the section in your own words, and fill in My reading: what this paper means for your question.

  3. Line the studies up in a table

    Ask for a table element per theme: a row per study with its design, sample and finding, naming its section, and the paper your argument leans on marked as the pick.

  4. Write the synthesis, then test yourself

    Write each theme’s overview yourself: where the studies agree, where they differ, what is still unknown. Before the panel, ask your assistant for a quiz on each theme.

Try this prompt

Your assistant, connected to know.sh (Claude, ChatGPT or a local model):

I am attaching the PDFs of ten papers. Using know.sh, add one Evidence section per paper to “Measuring green space exposure” on my shelf “Green space and adolescent mental health”, titled “First author year: design, exposure and outcome”, with the DOI as its source and four parts: Methods, Sample, Findings, Limitations. Take every number from the paper; write “not reported” rather than guess. End each with an empty “My reading”.

Your assistant, connected to know.sh (Claude, ChatGPT or a local model):

Using know.sh, read every Evidence section in “Outcomes and instruments” on my shelf “Green space and adolescent mental health” and add a new Insight section holding a table element: a row per study, with first author and year with the section number, design, sample size and finding in a few words. Mark the Key paper as the pick. Write “not reported” rather than guess, and do not change the sections.

Made with

DocumentSectionYour AI assistantElementsThe A–Z indexQuiz

See also