Every informative essay or research paper should include a summary at the beginning. Write your summary with the expectation that most readers will ONLY read the summary. The summary should tell most readers everything they need to know. The body of the article only exists to provide context and supporting evidence.
(That was my summary. You can probably stop reading now.)
The summary of your article should describe your central conclusions, which is ultimately the point of your article. A summary is high-context/high-trust, while the body of an article is low-context/low-trust. A summary is high-context because you write it for readers who have enough context to understand what you’re talking about. It’s high-trust because it’s for readers who trust that you know what you’re talking about, and who you trust not to deliberately misinterpret you. Then you write the body of the article to give additional context, and to justify your conclusions to skeptical or inquisitive readers.
There are basically two types of articles that should have a summary:
In a research paper, the goal is to test some hypothesis. (That hypothesis doesn’t have to be empirical, e.g., your paper could prove a mathematical theorem.) Most readers only care about what the hypothesis is, and whether it’s true or false.1 So put that in your summary.
In a persuasive essay, the central argument is usually pretty simple. Put the central argument in the summary. This might be be sufficient to persuade open-minded readers who have enough background to understand the argument. Then, in the body of the essay, explain the argument and provide supporting evidence.
If most readers only read the summary, that means you should spend much more time on the summary to make sure it conveys exactly the message you want. When I write an essay, I usually spend 5–10 times longer per word writing the summary than writing any other part.
Academic papers almost always include abstracts. An abstract is like a summary. Unfortunately, many (most?) papers don’t have good abstracts. Some common mistakes:
- The abstract uses overly convoluted language, making it hard to understand. (An abstract/summary is high-context, but that doesn’t mean it has to be hard to understand.)
- The abstract describes what the paper does, but doesn’t describe what conclusions it reaches.
Examples of bad and good summaries
I don’t want to pick on anyone, so I’ll write my own bad summary.
Luke Muehlhauser’s How Feasible Is Long-Range Forecasting? has an excellent summary, which I’ll talk about in a minute. But first, here’s my own (bad) summary of this article, written in the style of a typical academic paper’s abstract:
Long-term projects or endeavors depend on predictions of distant future events under low information and high uncertainty. This paper investigates historical evidence on long-term track records. There are significant concerns regarding historical forecasts with respect to quantification of degrees of confidence, null hypotheses, and forecaster behavior. Some difficulties are discussed for assessing the accuracy and determining the value of long-term forecasts. In conclusion, this organization’s current attitude toward long-range forecasting is provided.
This summary is bad for several reasons:
- The sentences are long and hard to read.
- It doesn’t actually explain the findings of the paper. “Some difficulties are discussed”? What difficulties, specifically? Don’t make me read the entire article to learn the article’s results.
- It tries too hard to sound objective, which contorts the language. “[T]his organization’s current attitude toward long-range forecasting is provided”? Why not, “We provide our current attitude toward long-range forecasting”? Or, even better, instead of saying that you’ll provide your attitude, simply provide your attitude: something like, “We don’t know how to make good long-term forecasts, but we plan to continue making them so we can get better at it.”
Now let’s read Muehlhauser’s actual summary:
- Long-range forecasts are often stated too imprecisely to be judged for accuracy. [More]
- Even if a forecast is stated precisely, it might be difficult to find the information needed to check the forecast for accuracy. [More]
- Degrees of confidence for long-range forecasts are rarely quantified. [More]
- In most cases, no comparison to a “baseline method” or “null model” is possible, which makes it difficult to assess how easy or difficult the original forecasts were. [More]
- Incentives for forecaster accuracy are usually unclear or weak. [More]
- Very few studies have been designed so as to allow confident inference about which factors contributed to forecasting accuracy. [More]
- It’s difficult to know how comparable past forecasting exercises are to the forecasting we do for grantmaking purposes, e.g. because the forecasts we make are of a different type, and because the forecasting training and methods we use are different. [More]
Take the first sentence of this summary:
- Long-range forecasts are often stated too imprecisely to be judged for accuracy.
This sentence forms a complete thought and tells me something that I might not have known. I don’t have any way of knowing whether this statement is true, so the article needs to justify it. But if I trust that the author knows what he’s doing, I could stop reading at this point without losing much.
For most readers, those seven conclusions are all they need to know. Every point on this list is easy to understand and tells you something interesting. There’s no need to read the rest of the article. But if you want to know the details, you can follow the links or continue reading to see how the author came to those conclusions.
How to write a good summary
I don’t really know how to write a good summary. I know one when I see one, but writing is harder than reading.
Some strategies I use:2
- Write a summary that matches the overall structure of the article. If your article includes three parts, write one sentence for each part.
- If your summary covers several independent points, write them as a bulleted list, not a big paragraph. Bulleted lists are easier to read (in my experience).
- Make each sentence as short as possible without losing any critical information.
- If you must choose between precision and clarity, choose clarity. People often end up writing overly convoluted sentences because they’re trying to make their statements impossible to misinterpret. That’s okay sometimes, but don’t overdo it, especially not in a summary. Keep the summary clear, even if that means readers have to read between the lines a little.
- Rewrite each sentence multiple times. Pick whichever version is clearest.
A lot of the advice for writing a good summary is the same as the advice for writing well in general. But even if you don’t meticulously comb through your entire article to make it as clear as possible, you should spend a lot of time time on your summary.
Obviously, a paper on an empirical subject cannot definitively prove it true or false. Rather, the paper provide evidence pointing to “true” or to “false”, and the reader wants to know which way the evidence points. ↩
I did not take all of my own advice in this essay. This essay is short, and these strategies are more relevant to long articles. ↩