Last updated: 14 June, 2018
Inconsistency in writing has a major impact on readers. Reports, proposals and articles that suffer from inconsistencies are more likely to be rejected. If authors and editors are to improve consistency in their documents, they need to know what sort of errors to look for. So we used PerfectIt – Intelligent Editing's add-in for MS Word – to check 2400 documents for consistency errors. PerfectIt provides a powerful way to compare the frequency of inconsistencies because it automatically checks for the worst errors in several different categories, including:
Each of the 2400 documents we tested was over 1000 words and was downloaded from the internet using the search term "final report".
There is no established way to count the frequency of consistency errors. For example, if the phrase 'decision-making' appears in three locations and 'decision making' (without the hyphen) appears in ten locations, should that count as one, three or ten errors? To overcome this problem, we simply measured the proportion of documents that contained at least one consistency mistake in each category. This approach allowed us to compare the results for different categories and identify the most common errors without favouring any particular kind of error.
Table 1 shows the ten most frequent errors and the proportion of documents in which they appeared.
|RANK||TYPE OF ERROR||EXAMPLE||FREQUENCY|
|1||Phrases in capitals||The word 'Government' in one location but 'government' elsewhere.||79.7%|
|2||Hyphenated phrases||The phrase 'decision-making' in one location but 'decision making' elsewhere.||62.5%|
|3||Heading case inconsistencies||Two headings at the same level but one in title case and the other in sentence case.||40.3%|
|4||Numbers in sentences||Numbers spelled out in one location but in numerals elsewhere.||39.4%|
|5||List / bullet punctuation||Within a single list, some items ending in a semi-colon but others having no punctuation.||37.8%|
|6||Table / figure labels||A figure called 'Graphic 1' being followed by another called 'Image 1'.||22.2%|
|7||Spelling||The word 'colour' in one location but 'color' elsewhere.||22.1%|
|8||Punctuation in tables||Some entries in a column being followed by a full stop but others having no punctuation.||16.6%|
|9||Capitalization in tables||Some entries in a column starting in lowercase but others in uppercase.||13.0%|
|10||Hyphenation of compound modifiers||The word 'anti' being followed by a hyphen in some cases but not in others.||12.4%|
The results illustrate how much room there is for improvement in almost all published documents. Every single error tracked in Table 1 appeared in more than 10% of the documents tested. That's an astounding proportion of documents with errors.
The clear winner (or should that be loser?) is inconsistency of capitalization. Almost 80% of documents over 1000 words suffer from inconsistent capitalization. Lots of those documents actually have far more than one phrase that is inconsistently capitalized. In the sample we tested, some 300 documents had more than ten capitalization inconsistencies.
The results also show that 6 out of 10 documents published online have inconsistent hyphenation. Perhaps even worse is that 4 out of 10 documents have bullets/lists punctuated inconsistently. If you bear in mind that not all documents contain lists, the proportion that have mistakes in punctuation is staggering. Authors need to do better.
Of course, not all inconsistencies are errors – some will be 'false positives'. For example, in terms of capitalization, there are reasons why a phrase might be capitalized in one place but not in others. Some authors put the word 'university' in lowercase when it's describing universities in general, but write 'University' when it's replacing a name. However, a quick manual check suggests that relatively few of the inconsistencies highlighted by PerfectIt in Table 1 are false positives.
Dr Hilary Cadman, a leading science editor, says "PerfectIt doesn't replace the editor's judgement. It does pick up occasional inconsistencies that are not actually errors, but the proportion is low. And that's not a problem anyway because the program gives me control over the changes I make".
The first line of defence against consistency errors is simply being aware of them. The Top 10 errors listed in Table 1 give authors and editors a reminder of important things to check before documents are published.
However, the best way to improve consistency in documents is to run PerfectIt. The program quickly locates consistency mistakes throughout the text. Instead of painstakingly looking for each error, PerfectIt scans an entire document in seconds, leaving you more time to focus on the things that matter most: your words and their meaning.
PerfectIt has a free trial and can be downloaded from the Intelligent Editing website.