How to research a nonfiction book

Where this guide touches copyright, it is general information, not legal advice.

Step 1: decide what needs research

Research the outline, not the subject. Go through it chapter by chapter and list everything a careful reader could ask "how do you know?" about. Almost all of it falls into three kinds:

Your own experience counts as evidence when labeled honestly: "In my practice, I have seen..." needs no citation; "Research shows..." needs research.

Research can also become the book's strongest material. Anthropologist Dr. Michael Hittman wrote the history of the Brooklyn Paramount theatre after, as his story on scribemedia.com puts it, he started digging in the library. The story lists Rebranded: "The restored theatre uses his research". That is one author's result, not a typical one.

Step 2: find reliable sources

The Society of Professional Journalists' code of ethics says: "Use original sources whenever possible." For a book, that usually means four places.

• Library catalogs and databases. Your public library's website lists the research databases your card opens, and a reference librarian can point you to more. The Library of Congress online catalog is free to search by subject and keyword.
• Government data. Data.gov describes itself as "The Home of the U.S. Government's Open Data" and listed 431,050 datasets when we checked. For population, business and income figures, start with the Census Bureau's data pages. Cite the agency, not the article that repeated it.
• Original research. If an article cites a study, find the study. Read the summary, the method and the limits, and report what it found, not what a headline said.
• Primary sources. Documents made by the people involved at the time: letters, court records, annual reports, minutes, transcripts, photographs. Interviews you conduct yourself belong here too; our guide to interviewing people for your book covers consent, recording and quoting.

Search engines and AI tools find leads, not sources. Open and read the original yourself.

Step 3: judge every source

Librarians at Meriam Library, California State University, Chico, publish a checklist called the CRAAP test. It asks five sets of questions, and notes that "different criteria will be more or less important depending on your situation or need."

Step 4: keep a source log

Keep one log from the first day, so you never hunt for a source you read months ago. A spreadsheet works. Record a row for every fact that reaches the manuscript:

Step 5: cite sources in a trade book

The Chicago Manual of Style, the guide both publishers quoted below point their authors to, describes two systems in its free citation guide. In notes and bibliography, sources are cited in "numbered footnotes or endnotes", usually with a bibliography as well; Chicago says it is preferred by "many working in the humanities, including literature, history, and the arts." In author-date, sources are "briefly cited in the text, usually in parentheses", a style preferred by "many in the sciences and social sciences."

Trade books usually take the first route, with notes at the back. Chicago's CMOS Shop Talk says "Publishers tend to assume that the average reader will want endnotes", because they are less distracting than footnotes, and describes a trend toward notes unmarked in the text and listed at the back by page number and key phrase. A Chicago Q&A calls this style "notes keyed to text" and says it is especially favored for books meant for a general readership.

If you have a publisher, its guidelines decide. The University Press of Mississippi's manuscript guidelines say "Endnotes are preferred", and PM Press's manuscript guide strongly prefers Chicago's notes and bibliography system, allows endnotes at the end of each chapter or at the back, and says citations "should not be in the form of hyperlinks."

Cite every direct quote, every statistic you did not produce, every research finding, and any claim that is not common knowledge. When in doubt, cite.

Step 6: quote other people's work fairly

Citing a source is not the same as having permission to reproduce it. The US Copyright Office's Fair Use Index describes fair use as "a legal doctrine that promotes freedom of expression by permitting the unlicensed use of copyright-protected works in certain circumstances", weighed on four factors:

1. the purpose and character of the use;
2. the nature of the copyrighted work;
3. the amount and substantiality of the portion used in relation to the work as a whole;
4. the effect of the use on the potential market for or value of the work.

The same page says there is no formula, no "specific number of words, lines, pages, copies", that may be used without permission, and that the index "is not a substitute for legal advice." The Copyright Office's fair use FAQ adds that "in cases of doubt, the Copyright Office recommends that permission be obtained." Facts and ideas are a different matter: the Office's FAQ on what copyright protects says copyright does not protect facts, ideas, systems or methods of operation, "although it may protect the way these things are expressed." So you can report a study's findings in your own words; long quotes are where questions start.

The University Press of Mississippi tells authors they may need to clear permissions for lyrics, poetry and epigraphs, and that "You are responsible for seeking both print and electronic permissions." Keep quotes as short as your point allows, comment on what you quote, and ask a publishing lawyer about anything you are unsure of.

Step 7: fact-check before anyone else sees it

Do not count on an editor to catch errors. The University Press of Mississippi's guidelines say plainly "The Press does not fact check", and PM Press says it does not have the staff to do extensive fact-checking and asks authors to "double-check everything for accuracy prior to submission." Assume the job is yours wherever you publish.

A workable pass, once the draft is finished:

1. Mark every checkable item: each name, date, number, quote and claim about a real person or company.
2. Match each one to a row in the log and reopen the source. No row: find the source or cut the line.
3. Check quotes word for word against the page or recording.
4. Recalculate the math, and make sure the year and definition match what the source measured.
5. Check the spelling of every name, including in the acknowledgments and notes, which PM Press singles out.

Check anything a writing tool produced with extra care; our guide to writing a nonfiction book with AI explains why invented quotes and sources are the main risk.

What most authors get wrong

• Researching before outlining. Reading with no chapter to serve produces notes, not a book.
• Citing the article, not the study. Secondhand numbers drift with each retelling.
• Overstating what the evidence says. A study that found a link has not proved a cause. Our self-help guide covers keeping claims honest.

Scribe Media, which makes this site, sells ghostwriting and publishing, including editing. Everything above you can do yourself; if you want to talk through a research-heavy book, Scribe offers a free call.