Unit 2: Accuracy Filter

Some of the errors in this category are likely to be considered the most serious by reviewers. The core aim of academic and scientific writing is to convey data, information and/or knowledge with zero distortion of meaning, and so errors that cause degradation of meaning need to be avoided.

When checking your draft using the accuracy filter, there are six error types to look out for:

1. Factual errors in world knowledge
2. Factual errors within the article
3. Overly bold claims
4. Overgeneralization errors
5. Statistical and numerical errors
6. Spelling and grammatical errors

Introduction to the accuracy filter

Corpus example: The population of Japan is 12,734,100.

In this example, the writer omitted the final zero when copying the figure, making the population ten times smaller than the cited source. This type of careless error can cause severe confusion. Readers may either believe the incorrect figure or work out that it is wrong — either way, the writer's credibility suffers.

Copying errors are easy to avoid by being systematic. When using copy-and-paste, double-check that the first and last characters have been included. It is more difficult to make selection errors when using arrow keys than a mouse, which relies on fine motor skills. When copying onto paper, be aware that sixes and zeros may be mistaken for each other, particularly when zeros are not exact ovals.

Corpus example: There are two types of.... First, .... Second, .... Third, ....

When writing a research article, the content of the draft may change significantly due to changes decided by the writer, co-authors, reviewers, editors or proofreaders. Mistakes in lists are commonplace. The number of items in the introductory sentence should reflect the number of items that follow.

In the example above, the introductory sentence lists two items, but three items follow. This appears obvious, but is easily missed by writers who are familiar with earlier versions of the text. People who read the document for the first time will, however, notice such errors immediately.

Corpus example: XXX will play a key factor in the near future.

The modal verb will tends to be used to show certainty about a future action or state. However, even simple events in the future may not happen due to unforeseen circumstances. Claims that are overly bold can be hedged in three ways:

• Reduce certainty
• Limit scope
• Add a condition

Corpus example: All women...

Claims about a characteristic of all members of a particular group are problematic. When making claims about groups, it is necessary to play devil's advocate and identify whether it is possible to argue against the claim. If it is, then consider adding modality, limiting the scope, or adding a conditional — the same three strategies used to hedge overly bold claims.

Corpus example: (p > 0.5)

The probability value p shows the probability of finding equivalent or more extreme results while assuming the null hypothesis is true. The smaller the p value, the higher the statistical significance. Reported p values typically need to be either p < 0.05 or p < 0.01 to claim statistical significance.

In the corpus example, a typographic error was not caught in proofreading. A probability of 50% means the result is no better than chance — no significance can be claimed, and the article would almost certainly be rejected.

Corpus example: There are three form the first experiment that XXX.

Spell-checkers can miss errors when the misspelled word happens to be a different valid word (e.g. form instead of from). A key problem for LaTeX users is not using a spell-checker at all, since with many free LaTeX editors, spellcheck is not enabled by default.

Checking spelling and grammar in one's own writing is also particularly difficult due to screen-memory interference — writers may not necessarily read what is currently written but may read what was previously written.

The following corpus examples each contain a spelling or grammatical error. Click each item to reveal the correction.