Free Study Resources and Reading Plans: 150 Books and Courses That Cost Nothing

<p>The distinction between <em>finding</em> and <em>reading</em> research papers is where most people give up. Search tools index far more than they let you read.</p> <h3>Where the papers actually are</h3> <table><thead><tr> <th>Source</th><th>Coverage</th><th>What is free</th><th>Access model</th><th>Typical lag</th> </tr></thead><tbody> <tr><td>PubMed Central</td><td>Biomedicine, life sciences</td><td>Accepted manuscripts and some final versions</td><td>Free to read</td><td>Often ahead of the journal issue</td></tr> <tr><td>PubMed</td><td>Same index, plus paid journals</td><td>Abstracts only for most</td><td>Abstract free, full text via PMC</td><td>Current</td></tr> <tr><td>arXiv</td><td>Physics, maths, CS, stats, econ</td><td>Preprints</td><td>Free to read</td><td>Days to weeks</td></tr> <tr><td>DOAJ journals</td><td>All disciplines, ~19,000+ journals</td><td>Full text</td><td>Free to read</td><td>Current</td></tr> <tr><td>Semantic Scholar / Crossref</td><td>All disciplines</td><td>Metadata and some full text</td><td>Metadata free</td><td>Current</td></tr> <tr><td>Google Scholar</td><td>All disciplines</td><td>Links, some PDFs</td><td>Links only; often paywalled</td><td>Current</td></tr> <tr><td>Zenodo</td><td>All disciplines</td><td>Accepted manuscripts and datasets</td><td>Free to read</td><td>On deposit</td></tr> <tr><td>CORE</td><td>Aggregates repositories worldwide</td><td>Full texts where deposited</td><td>Free</td><td>Varies by source</td></tr> <tr><td>Europe PMC</td><td>Biomedicine, wider than PubMed</td><td>Full text including preprints</td><td>Free</td><td>Current</td></tr> </tbody></table> <h3>The preprints problem</h3> <p>Preprints are manuscripts released before peer review. That is their value and their risk. In fields where preprints are routine (physics, computer science, biology) a preprint is close to a normal research object; in fields where they are rare, treat them as work in progress.</p> <p>The 2020 COVID-19 preprint episode taught the field a hard lesson: many high-profile preprints were wrong, and headlines amplified them. The lesson for a reader is simple — <strong>a preprint's claims are unreviewed, so reproduce or distrust, but do not cite as established</strong>.</p> <h3>A working literature-search procedure</h3> <ul> <li><strong>Start from a review article, not a primary paper.</strong> Reviews list the literature you need and are freely available in most fields. Find a recent, open-access review and read its reference list.</li> <li><strong>Find the paper the field treats as canonical.</strong> Every field has two or three papers everyone cites. Ask a tutor, or search for the phrase "X is the standard reference for" — but better, find the paper's own "further reading".</li> <li><strong>Check the open-access status of the exact DOI.</strong> A paper being indexed is not the same as being readable. Test the DOI before you put it on a list.</li> <li><strong>Use the repository route.</strong> For paywalled papers, search the title on Zenodo, CORE, PubMed Central or the author's institutional repository. A large share of paywalled papers have a free green OA copy deposited by the author.</li> <li><strong>Read the abstract, then the methods, then the results.</strong> In that order. Many people read the discussion first and end up reading the authors' opinion rather than the evidence.</li> <li><strong>Read one paper per week, properly.</strong> Ten minutes with a good paper, annotating as you go, beats an hour of skimming five.</li> </ul> <blockquote><p>The goal is not to collect papers. It is to be able to explain, from memory, what a paper found and what it could not show.</p></blockquote> <h3>Reading a paper properly: six passes</h3> <ul> <li><strong>Title and abstract.</strong> Does it answer a question you care about?</li> <li><strong>Introduction.</strong> What gap does it claim to fill?</li> <li><strong>Methods.</strong> Who was studied, how many, and how were they selected? Look for the sample size and the inclusion criteria.</li> <li><strong>Results.</strong> Read the tables and figures before the prose. Authors write around their figures.</li> <li><strong>Discussion.</strong> What limitations do they admit?</li> <li><strong>References.</strong> Which earlier papers recur? Those are your next reads.</li> </ul> <h3>The evaluation questions that matter</h3> <ul> <li>How many people, and how were they recruited?</li> <li>What was the comparison group?</li> <li>Was the outcome measured or self-reported?</li> <li>Was the paper's own limitation section naming the thing you suspected?</li> </ul> <blockquote><p>A paper is a tool, not a fact. Read it for what it measured, not for what you wanted it to say.</p></blockquote> <h3>Datasets and reproducibility</h3> <p>Open data lives in Zenodo (a DOI is minted per deposit, which makes citation easy), Dryad, and discipline-specific archives. Cite the version of a dataset, not just the site. When you use someone else's code, look for a repository with a DOI and a licence, and record the commit or release you used.</p>