You’re editing a biography of Lester Pearson, and you see a line that says, “It is a fitting coincidence that Pearson was born the same year another great Liberal leader, Wilfrid Laurier, took the Canadian helm.” You’re drawn to the interesting coincidence of dates and decide to look up the facts. You discover that Pearson was born in 1897, while Laurier took up office in 1896. Naturally, you mention this apparent discrepancy to the author, who might have meant that the two events occurred within 12 months of each other (which is true).
So far, so obvious.
Let’s look at another example. You’re a medical editor finding yourself with an article on a rare disease: “The test for this disease correctly identifies 99% of genuine cases. The patient tested positive, so we concluded that she was very likely to have the disease.” You happen to have a background in statistics, and you recognize that there’s a logical flaw: it’s impossible to conclude from the information given that the patient is very likely to have the disease.1
Maybe in this second situation you’ll still say something to the author. After all, as long as you’re polite, there can be no harm in bringing any kind of error to their attention. But if the author disagrees with you, how likely are you to push the point further? Presumably less likely than if the error were simply an error of fact, like the Lester Pearson example. You may feel you’re treading on the author’s toes, questioning their very ability to lay down their argument. You know you’re on delicate ground; your relationship with the author is too young for difficult conversations about content. We’re led to:
Question 1. Why do we treat errors of fact differently from errors of reasoning?
Let’s change tack a little. Many editing jobs are of the kind where the author pays you to edit a document before they submit it to a publisher. In that case, what happens to the text after you’ve finished your edit is entirely the author’s concern. But imagine yourself on the other side, being paid by the publisher to edit that same text. What now if you spot an error of reasoning? The publisher has a stake in this text: their name is going on it, and their reputation depends to an extent on readers viewing the writing as reliable, both factually and logically. It therefore seems reasonable to ask:
Question 2. Do we have a responsibility to alert the publisher we work for to any kind of error we haven’t been able to resolve with the author, whether an error of fact or of reasoning?
I don’t have definitive answers to these questions, but I suspect each editor has to find their own answers — if the questions apply to their work in the first place. If you’ve already been on that journey, I’d be very interested to learn where you arrived, so to speak, and how you got there.
Previous post from Paul Buckingham: A Formula for Mathematical Editing.
The Editors’ Weekly is the official blog of Editors Canada. Contact us.
1 The missing piece of information is how good the test is at correctly recognizing people who don’t have the disease. Suppose this rare disease affects 1 in 100,000 people. If the test is wrong for even a small percentage of non-affected people, say 1%, then in a test of 10 million people you’d expect 99,999 non-affected people to show positive. But only about 100 of the people in the test actually have the disease! Now we see the problem: many, many people who test positive are in fact fine. For further reading, see Bayes’ theorem.