Identify
Record the research question, publication type, sample, population, setting and original source.
Know what a study
can and cannot prove.
This guide explains how to tell a useful finding from a claim that goes further than the research allows.
What qualifies for the library
A research record must lead back to an original paper, official report, publisher, or recognised research index. Blogs, campaign statistics, and claims that cannot be checked against their original source are not treated as evidence.
Attribute
Describe findings as the authors’ reported results. Policy conclusions remain attributed to their authors.
Look for other explanations
Ask whether income, conflict, earlier caregiving, family choice, the measures used, or differences between groups could explain the result.
Update
Link later replications, critiques and syntheses without rewriting the historical record.
Several studies combined
Reviews of the wider research
These show whether a finding keeps appearing across different studies. A review is only as reliable as the studies it includes and the fairness of its search and comparison methods.
Children followed or groups compared
Individual research studies
Following children over time can show which event happened first. Comparing existing families can reveal important patterns, but family differences may still explain the result.
Research translated into guidance
Expert and consensus papers
These can explain how specialists interpret a body of research. Agreement between experts is useful context, but it is not a new test of what happened to children.
How sources are usedIncluded as evidence Included as context Not treated as evidence
Original research stays separate from commentary.
The library distinguishes original research and official reports from commentary, summaries, news articles, and advocacy claims. A factual claim is treated as evidence only when it can be checked against an original publication or authoritative data source.
Original studies, systematic reviews, meta-analyses, scholarly consensus papers, and official research reports with a traceable source.
Scholarly critiques and legal commentary, clearly labelled so readers do not mistake interpretation for new empirical data.
Blogs, advocacy statistics, unsourced summaries, and secondary claims that cannot be verified against the underlying publication.
Source checking and corrections
What a source check establishes
The collection contains 36 research records.
Checking a publication record confirms citation details. Reading an abstract can confirm its reported summary, but cannot verify every numerical result or interpretation in the full paper. The collection is selective and does not constitute an independent systematic review.
5 September 2026
Removed the claim that all current records were reviewed together on 22 July 2026. That statement used an outdated record count and did not provide an individual review history. Separated topic-wide editorial commentary from the findings attributed to each paper. Corrected the author order for De Wolff and van IJzendoorn (1997), the co-author Raziye Salari in Bergström and colleagues (2018) and the missing subtitle in Dagan and colleagues (2024). Updated Zheng and colleagues to the 2026 volume citation while preserving its existing page address.
Study results and source details
Corrected the Pruett, Ebling and Insabella (2004) comparison from 132 parents to 132 families at follow-up and linked the correct publication. Its summary now reflects the reported age and sex differences. Scoped the infant-overnight answer to the studies reviewed and replaced the undefined zero graphic with the study's qualified direct-result wording. Removed unsupported rankings and the unreferenced description of McIntosh's work across 2015 to 2017.
Clarified that Dagan's parent-group comparison did not establish equal effects. Corrected the 2014 practice framework to eight considerations and retained its authors' caution about five or more monthly overnights at ages 0 to 18 months. Clarified that Hjern's caution about children under four is not a validated separation-age cutoff. Schedule illustrations remain labelled as editorial examples.
Corrected the van IJzendoorn (1999) citation to pages 225 to 249 and the synthesis count to nearly 80 studies. Corrected the McIntosh (2010) report title and the Zheng effect-size descriptions. The Children Act 1989 link is labelled England and Wales. The 2024 attachment-cohort summary distinguishes the different samples behind its age-one and age-four percentages. Researcher and historical summaries were narrowed where their earlier wording lacked source support.
Review the corrected comparisons · Read the practice framework · Read the age-seven study
Clearer reading and evidence labels
Distinguished commentary and practice guidance from consensus reports. Each guide now shows its full reading explanations once. Publication-record links are labelled by their destination. Saved library searches remain available when readers continue to another research record.