Aim for fewer words per sentence, fewer passive sentences, higher Flesch Reading Ease, and lower Flesch-Kincaid Grade Level.
Writing is easy. Writing well is hard. In this article, I suggest principles for writing. I do not claim that these are great principles, and other people may prefer a different style. But if I am supervising your project, I would appreciate your following them.
Take this seriously. If, for example, your article has a pie chart, then you have not been paying attention, and you will have to delete it later.
1. Text. Get feedback. Be clear and concise.
2. Equations. Follow tradition. Learn Equation Editor.
3. Tables. Use clean text, no grids.
4. Excel graphs require editing.
5. Tell the truth.
6. Conclusion: Less is better.
You can get feedback on your writing several ways, but the first way is most important.
Give your first two pages to me, before you write the entire paper. This should be at least one month before your project is due. Most likely, I will have numerous comments to make on the wording and structure of just those two pages. If you get the first two pages right, then you are at a huge advantage for writing the remainder! (I recommend you discuss your milestones with me before you decide you are done and before you make plans to leave the country.)
Students want to write the whole thing first, and then go back later and put the paper in a logical order. Unfortunately, you will be too tired and too in love with your own work to want to rip it up. You are unlikely to have even your first paragraph right. Show me your work in pieces. Sections are easier for me to digest, and you will get feedback as you go.
Get feedback on your work from the University’s Academic Skills Centre (especially WASS or ELSP), who tutor students in writing.
Get feedback from Word. Use the spell checker and the grammar checker often. The grammar checker in Word 2000 is better than earlier versions, though not perfect. But when in doubt whether to accept Word’s proposed change, bet on the billionaire, especially for spell checking. Mistakes annoy your reader and create a judder bar to smooth reading, so do not have any.
One problem with Word's grammar checker: Word sometimes marks text to be ignored. To make sure the grammar checker checks everything, select all the text with Edit, Select All. Then select Tools, Language, Set Language, English (New Zealand). Be sure to clear the tick box that says, “Do not check spelling or grammar.”
Use Word’s readability statistics in the spelling and grammar checker. Aim for 5 characters per word, 12 words per sentence, under 15% passive sentences, and a Flesch-Kincaid Grade Level of 9 or below. Do not try to cheat the readability statistics, such as by making bullet points of your whole report. Write in complete sentences. Compare your document’s readability statistics with the readability statistics for your classmates’ documents. Yours should be the best. (Flesch’s article is on the Department server.) Here are two examples. Based on this alone, which document would you prefer to read?
Aim for fewer words per sentence, fewer passive sentences, higher Flesch Reading Ease, and lower Flesch-Kincaid Grade Level.
Here are the steps you should follow in writing your report:
1. Write a new section.
2. Proofread the section. Shorten sentences. Eliminate passive voice.
3. Run Word's grammar and spell checker. Make those corrections.
4. Have a friend proofread it.
5. Show the section to WASS or ELSP.
6. Correct the section.
7. Show the section to me.
8. Correct the section.
9. Go back to step 1.
If I give you comments, please do not ignore them. Make the changes or explain why not.
You may have to go through this process two or three times for each chapter. After the judder bars of bad grammar and spelling are fixed, the main problems with your writing will be clarity and conciseness.
Write your report as though you were explaining the project to an eight-year-old child.
Link linguistically. For better clarity, write so the predicate of sentence i is the subject of sentence i+ 1. “The cat in the hat came back. He came back to get his red box. The red box had two Things. The Things were named Thing One and Thing Two….” Linking your sentences this way is a challenge, but linking keeps your text organised and prevents repetition. This linking is described by Gopen, George D., Swan, Judith A., in “The Science of Scientific Writing,” American Scientist, vol. 78, Nov-Dec., 1990, pp. 550-558, on the Department server, with other material about writing.
Think of a link. Link sentences with repetitive words, like repeating beads on a string. The repeating beads are key words at the end of a sentence which appear again at the beginning of the next sentence. Just as sentences connect using key words, the key words at the end of a paragraph should appear again at the beginning of the next paragraph. By connecting sentences and paragraphs with these links, your writing will flow logically, and will be naturally well organised.
Without links, every paragraph is likely to be made up of vaguely related sentences. Vaguely related sentences mean that consecutive paragraphs will be vaguely related, and the document ends up being disordered. To bring sense to the disorder, you end up referring the reader to different sections, with phrases such as “... as stated above...,” when proper linking would have organised the whole document in the first place. Organise the document so you make a statement once.
You do not have to use linking words monotonously, but break the rule only with caution. None of my students has ever done too much linking. The error is always too little. A worse error is to link with words without linking ideas! Make sure the ideas link, but this is most easily done by linking words. Linking keeps your document well-ordered.
Besides careful ordering, you can improve clarity by replacing the words “this” or “it” with the original subject. “This problem…” might better be stated with the original subject “Integrality constraints…” You may be concerned that it will sound repetitive (gotcha! – that repeating the subject will sound repetitive), but the text will be clearer. Try a search on “this” and “it” in your paper. Are their references clear?
Write using the positive sense. “This is not believed to be unrealistic” is harder to understand than “We believe this is realistic.”
“Substitute 'damn' every time you're inclined to write 'very'; your editor will delete it and the writing will be just as it should be.” -- Mark Twain
“Verbosity leads to unclear, inarticulate things.” - Dan Quayle
“Short words are best and the old words when short are best of all.” — Winston Churchill
Take pride in brevity! I offer a bet with any student that I supervise: Choose a sample of my writing and a sample of your writing. I will pay you five cents for each letter you can remove from my writing without changing the meaning of the text. But you must pay me five cents for each letter I can remove from your writing, without changing the meaning of the text.
Make the body of your report twenty to thirty pages long. You can put technical information such as tables in an appendix. You do not get extra credit by making the paper long.
How do we write concisely? We must learn how, and there are some tricks to it. One of the great references on writing is The Elements of Style, by Strunk and White. The book takes some time to understand - you will probably need to learn some more grammar - so it is probably best studied closely and slowly. But the skills will stay with you, like riding a bicycle. You should buy Strunk & White, but here is the book on-line. Also, Student Services has courses on-line for writing: http://www.studentservices.canterbury.ac.nz/awc, http://www.studentservices.canterbury.ac.nz/awc_2001. The Computer Science Department has a lovely page about writing. Jonathan Shewchuck wrote a lovely article, "Three Sins of Authors in Computer Science and Math". The Economist has an excellent style guide.
But where other instructions contradict mine, I recommend that you follow mine, when I am the main supervisor for your project.
Start your conciseness efforts with the title. Have a short title, four or five words at most. Can you think of any movie that has a title more than five words? Movie writers make titles short to be interesting and memorable.
Making text brief is work. Go over your article several times to eliminate words. Replace long words with short words. Replace long phrases with short phrases. Cut long sentences in two.
Passive voice is a passive vice. The passive voice should never be used. Seriously, passive voice makes text harder to understand and requires more words. Academic writing is so often and so unfortunately passive! Writing in passive voice is a bad tradition. Passive voice is like roots in the sewer, bird-doo on the windshield, and phlegm in the nostril. When removed, the text will flow, meaning will be clear, and we all will breathe a little more freely.
“It was the opinion of the administrator that employees should be dealt with sternly by the supervisor,” versus “The administrator believed the supervisor should deal sternly with employees.” The first sentence has seventeen words, the second has ten. Killing the passive voice makes the sentence shorter and clearer.
If you feel you need to use passive voice (perhaps because you are writing for academia), then use it only for your own voice, and use active voice for everything else. “No references proving that P=NP could be found.” All other agents can be active! “The optimum cannot always be calculated by the Foo-far heuristic” would be better as “The Foo-far heuristic cannot always calculate the optimum.”
You may find that the active voice requires you to make a stand. “While looking at the issues of employee scheduling and staffing, it would seem that the use of temporary employees might actually sometimes raise costs rather than lower them,” versus “Temporary employees can raise costs.” The second sentence is so much stronger that we have to support our statement with evidence. This is good. Write actively to silence passive voice.
Besides silencing passive voice, delete empty phrases such as: in order that, obviously, clearly, that are being, it is shown that, for the fact, it appears that, what this shows is that, that result, very.
Replace long phrases and words with shorter ones. Examples: replace “with regard to” with “about”; “produces” with “gives”, “the manner in which” with “how”, “associated with” with “of”.
Do not use the phrases “may or may not be” or “but then again”. They say nothing. (For a humorous way to say nothing, see “This Page Intentionally Left Blank.”) Similarly, you can shorten the structure, “It not only ... but also ...” For example, “It not only provides local ridesharing matches but long distance travel too!” could be rewritten as “It provides local and long distance ridesharing.”
If you frequently refer to a concept, name it. Give your reader convenient and understandable shorthand for otherwise long phrases that must be used repetitively. “The time at which a widget enters the polishing department” could be written, “polish entry time.” Otherwise, every time you need to refer to “polish entry time”, you will be writing, “The time at which a widget enters the polishing department,” over and over again. Give the definition early in the text.
One way to name a concept is to use an acronym. Acronyms tend to make work for your reader, so you can use a few, but do not use many. Do not use an acronym unless you need it, and only if the acronym is significantly more concise. Do not define an acronym unless you use the acronym several times.
Do not try to be elegant. Writing clearly and correctly is difficult enough. If you are trying to make the text sound good, you are working on the wrong thing. You probably do not need to print in colour and you certainly do not need an expensive binding. Just get the text clear and concise.
Research your project subject at the library. Who else has done related work? Feel free to ask me for help. You will learn a lot of background about your topic, and you will find out whether your idea is new or a previously invented wheel. Ask people in the University if they have done related work, or if they know of someone who has.
Following mathematical tradition, you should italicise parameters, variables, and subscripts. Font style and size should match the document’s text. When you use Equation Editor, use its default fonts, which automatically match the surrounding text and mathematical tradition. Do not enlarge equations to have larger text than the surrounding. Do not remove the default italic for math symbols.
Do not italicise text in a formula, and use normal case for text in a formula.
For example,
MEAN TIME BETWEEN FAILURE = (TOTAL UP TIME)/(NUMBER OF FAILURES)
is better rendered as
Mean time between failure = (total up time)/(number of failures).
Number each equation, so you can reference the equation in the text. If you put the equation number on the right, you may find it convenient to use a right-justified tab (try Format, Tabs, Alignment Right, and set the tab for about 15 centimetres). Also, consider inserting an automatic counter to count the equations for you, with Insert, Field, Numbering, Autonum.
Name each model so you can reference the model in the text. “Model IKN/SRI requires integrality constraints,” is better than “The formulation with only the capacity constraints and the end-of-horizon perpetuities requires integrality constraints.”
For a math model, use the IPDME format. IPDME format is due to Gerald Brown. See his excellent articles, “How to Write About Operations Research,” and “Top Ten Secrets to Success with Optimization”, as well as Rick Rosenthal's related presentation, “Eleven Keys to Success in Optimization Modeling”.
Where reasonably possible, avoid Equation Editor. Equation Editor causes a variety of problems.
Equation Editor objects are slow to open for re-editing. The interface is annoying and unintuitive, especially for people just learning it. If you change the font of the surrounding text (say, from Arial to Times New Roman), the equation object must be opened for the equation text to match.
When saved as HTML, Equation Editor objects are saved as gif images.
These gif images do not scale automatically with the browser window or text, though they can be rescaled with HTML code.
When placed in-line, the gif images do not vertically align with the text, producing an ugly result:
.
With effort, you can get the image to align with the text, by adding align="middle" in the HTML code for the image.
But the format is still ugly: .
In PowerPoint, Equation Editor objects cannot be placed in-line with text, making animation difficult. Finally, Equation Editor objects in PowerPoint sometimes need recolouring to match the surrounding text or to print in inverse greyscale.
Instead of Equation Editor, you can often write equations with just plain text italics, subscripts, and superscripts, especially for small formulas such as ADC power = AAC main*Arectifier. Plain text is easier to write and edit in Word or PowerPoint: å ¥ i=1 1/2 i. In Word, you can set AutoCorrect to automatically convert text to symbols, such as “lambda” to “l”. Unfortunately, not all browsers will display math symbols, so proofread files converted to HTML.
But many times, Equation Editor is the right tool. Use a separate Equation Editor object for each equation, rather than putting a long list of equations in one object. Shortcut keys help a lot. You may want to print the shortcut keys from the help file.
In Equation Editor, operators such as summation S and grouping symbols such as parentheses ( ) have fields for data. The fields are designed to hold the complete mathematical element for that operator. For example, you do not need to type an open parenthesis, term, then close parenthesis. Instead, insert a parenthesis object, and fill in the middle. Also, the summation S has a field to the right to hold the summed term. With a double summation, the first summation field contains the second summation symbol and its terms.
Above, I said that you should not italicise plain text in an equation. With Equation Editor, you have to do extra work to avoid italics: select Style, Text. Text style also allows spaces.
For tables, make all fonts consistent with the surrounding text, generally 10 or 12 point Times Roman. Avoid bold, italic and underscore.
Grids are rarely helpful. If you turn the grid off, does the table still look good? Fix the table in some other way than by adding a grid.
Before you use an Excel graph, question the need for it. Is this really the best way to get the information across, or are you just trying to make a few numbers look cool? Do not use a graph for 10 or fewer numbers. Instead, use plain text or a table.
If you feel a graph is important, use the right graph. Do not use bar graphs or pie charts. Instead, use a line or x-y chart.
Once you have chosen a graph, sort data in the most interesting order. For example, if you have a list of teams and profits, sort the data by profit, not team number. This way, the reader can see each team’s rank by profit. If you really want to show data sorted two ways, make two separate graphs, and display them side-by-side.
After arranging the data, engage the best tool for improving Excel graphs - the delete button. Minimise the ink needed to show the data.
You can tell when you have too much ink when it is obviously "made in Excel." The February 6, 2001, edition of PC Magazine ran an article about Excel graphs. Here is what they wrote in the first paragraph of that article:
For all its power and versatility, the charting engine in Microsoft Excel has a frustrating tendency to make “graphy-looking” charts - the kind that would impress your high school algebra teacher more than your colleagues or clients. In many cases, the axis lines, legends, centered titles, and square data-point markers that Excel's chart wizard gives you by default distract the viewer's attention from the message you want to convey. To get your point across, you often need much less visual paraphernalia than Excel wants to give you.
Delete boxes and pointless shading. Use no boxes, not around the data, nor around the entire graph. Instead, stretch the graph (the real graph, the part containing the data) to fill up the space, so the data area is bigger.
Delete Excel’s depressing default grey background. The grey background lowers contrast to the data.
With only one series, delete the legend, and stretch the graph to fill the space. If you have two or more series, delete the legend and use carefully placed textboxes (with no outline and no fill) to label each series.
Reduce the size of data markers to 2 or 3 points, or remove them.
Colour causes trouble. Use a grey scale or texture instead. Grey scale more naturally displays magnitude. If you have to denote levels of magnitude, e.g. “hail,” “heavy rain,” “light rain,” and “fog,” use a grey scale from dark (heaviest, strongest, most expensive) to light grey (lightest, weakest, cheapest). If you use colour, e.g. “red is for hail,” “blue is for heavy rain,” etc., your audience must memorise the colour coding just to interpret the chart.
Excel’s default chart text is Arial, sometimes bold, and sometimes rotated. Make all fonts consistent with your document’s text, generally 10 or 12 point Times Roman. (To import an Excel chart to PowerPoint, use 14 point text in Excel.) Turn off Autoscale. Avoid bold, italic, and underscore. Avoid coloured text. Use sentence case. Reset the text orientation to 0 degrees, so all text reads from left to write, including on label for the y-axis.
Format axes properly. In the graphs below, the vertical axis is sales, so the numbers on the vertical axis should be formatted with a dollar sign. Format to show only the important digits. Show commas to separate thousands.
Bar graphs require more ink than line graphs. Remove unneeded boxes and shading.
Ideally, a graphic would be in-line, so the text around the graphic explains it. In-line graphics are hard to do well. Most publications want conventional captions. So put the punch line in the caption, in a complete sentence. “Figure 13: Most deaths by chocolate occur on Friday nights.”
If your project requires many graphs or relies on visual displays of data, discuss your display with me. You should also read The Visual Display of Quantitative Data, by Edward Tufte, on reserve at the library. A fairly good web page on the subject is Informative Presentation of Tables, Graphs and Statistics, though Tufte's book is the authority.
When you write, be sure to tell your client the truth, even if it hurts.
The consulting firm, McKinsey & Company, has the following firm code:
- Place client interests first.
- Accept only those assignments that we are fully qualified to perform.
- Maintain independence and objectivity, even if it means declining or withdrawing from an assignment.
As a student, you will be accepting assignments that you are not fully qualified to perform. That is understandable. But do not write a false conclusion that your client wants you to write. Tell your client the truth, even if it hurts. I expect you to hold to high ethical standards. If you make a mistake, I expect you to apologise to the client.
Related to telling the truth, speak wholeheartedly. Sometimes presenters or writers give material that they know is too advanced or too complicated for the audience. Present only information that you actually mean for your audience to understand. Otherwise, you frustrate your audience and waste everyone's time.
Simple and plain are usually best, but getting there takes a surprising amount of work. Do not try to make your work or presentation look cool. Instead, get the information across as cleanly and simply as possible. In the end, this will be right.
“No passion in the world is equal to the passion to alter someone else’s draft.” — HG Wells.
“We have read your manuscript with boundless delight. If we were to publish your paper, it would be impossible for us to publish any work of lower standard. And as it is unthinkable that in the next thousand years we shall see its equal, we are, to our regret, compelled to return your divine composition, and to beg you a thousand times to overlook our short sight and timidity.” — rejection slips from a Chinese economic journal.
17 Mar 2006 jfr
Readability statistics for this document:
