Reading research · Lesson 1 of 3

Level 1 and 2: is this relevant, and what did it do?

Scan everything, structurally read what survives, and reserve deep reading for the few papers your argument rests on.

8 min read · Free — no account needed to read

Level 1, the fast scan. Two to three minutes, and its only job is a decision: read further, keep for later, or discard. Read the title, the abstract, the keywords, look at the figures and tables, and read the last paragraph of the conclusion. Figures are underrated here — a paper's main result is usually in one of them, and you can often see the finding before reading a sentence of the discussion.

Then decide explicitly. The decision is what makes this a level rather than a skim, and "keep for later" needs a note saying what for, or later-you will reopen it with no idea why it was saved.

Level 2, structural reading. Fifteen to thirty minutes, aimed at what the paper actually did. Work out: the research question, what the authors say is missing from existing work, the method and design, who or what was studied and how many, what was measured, what was found, what the authors themselves say the limitations are.

Read out of order. Introduction for the question, methods for what was done, results for what was found, then discussion — and read the discussion last and lightly, because it is where interpretation is most generous. Reading results before discussion means you form your own view of the numbers before being told what they mean.

At the end of level 2 you should be able to describe the paper to someone in four sentences: the question, the method and sample, the finding, the main limitation. If you cannot, you have not finished level 2, and no amount of highlighting will substitute.

See it happen

You have 40 saved papers and five days to make sense of a literature.

The read that arrives on its own

Start at the top and read carefully down the list, getting through as many as possible.

The same situation, with this idea

You scan all 40 first, at three minutes each — two hours. Around 12 survive as genuinely relevant. You read those at level 2, roughly 20 minutes each, which is another four hours. Three turn out to be central to your argument and get level 3. Total: under two days, with a clear view of the literature and your attention spent where it mattered — and you now know that 28 of the 40 were never going to help.

After this, you can

decide a paper's level before you start reading it rather than during.

The takeaway

Scan everything, structurally read what survives, and reserve deep reading for the few papers your argument rests on.

Easy to take the wrong way

That scanning is lazy. Deciding quickly and accurately what not to read is the skill that makes the deep reading possible.

Try this

Scan five saved papers at three minutes each and record the decision for every one. Notice how many you would otherwise have read in full.