NUMBERS AND STATISTICS

Use tilde (~) to mean approximately equal to.

Numbers beginning a sentence must be spelled. Rewrite a sentence so you don't start it with numbers greater than ninetynine.

Note: one, two, three… nine, 10, 11, 12… Exceptions: a 2m tape measure; 3 million.

Put a space between numbers and units: for example, 75 kg. Exception: 75%.

Note: 0.32, not .32.

Note: 143, 2,461 or 2461, 21,278, 1,409,000…

When you quote numbers, make sure you use the minimum number of significant digits or decimal places. For example, 23 ± 7 years is easier to read than 23.4 ± 6.6 years, and the loss of accuracy is not important in most situations.

Use the appropriate number of digits: two significant digits for standard deviations (one digit if the standard deviation is for a descriptive statistic like height or weight, or if precision is not important); two decimal places for correlations, two significant digits for percentages. Examples: 73 ± 5; r = 0.45; r = 0.08; 16%; 1.3%; 0.013%.

If it is more convenient to show p values than confidence limits, show the exact p value to one significant digit (for p < 0.1) or two decimal places (for p > 0.10). Do not use p < 0.05 or p > 0.05. Examples: p = 0.03; p = 0.007; p = 0.09; p = 0.74. (The exact p value is important for anyone using your data to calculate confidence limits or using your data in a metaanalysis.)

Make sure the significant digits of the mean and standard deviation are consistent. Examples: 20 ± 13; 0.020 ± 0.013; 156 ± 7; 1.56 ± 0.07; 15600 ± 700.

Use the standard deviation as a measure of spread. Do not use the standard error of the mean.

Avoid test statistics like t, F and c^{2}, but if the journal insists on them, show only two significant digits.

Show 95% confidence intervals for effect statistics like a correlation coefficient or the difference between means.

Interpret the magnitudes of outcomes in a qualitative way, using both your experience of the magnitudes that matter in this area of human endeavor and also any published scales of magnitudes (e.g., Cohen, 1988; Hopkins, 1998). You must interpret the observed effects and the confidence limits. For example, you might have to say that you observed a moderate effect, but that the true value of the effect could be anything between trivial and very strong.
TABLES

Create tables with the Table pulldown in Word. Do not use tabs.

Examples of tables in Sportscience style are shown in Tables 1 and 2.
Table 1: A simple generic table for articles at the Sportscience website^{a}.



heading

heading

heading

item

item^{b}

item

item

item

item

item

item

item

item

item

item

item

item

item

^{a}Put any footnotes here. Note that the caption and footnotes are in cells of the table. ^{b}Number footnotes as shown.



Table 2: A complex table^{a}.





heading

heading

heading

Subheading1





item

item^{b}

item


item

item

item

Subheading2





item

item

item


item

item

item

^{a}Put any footnotes here. Note that the caption and footnotes are in cells of the table. ^{b}Number footnotes as shown.




FIGURES

Note these rules for choice of figure format:

line diagrams or scattergrams if independent and dependent variables are numeric;

bar graphs if only the dependent variable is numeric;

bar graphs or pie charts for proportions.

Do not use scanned images of graphs or diagrams, because the lines and symbols become too "pixelly." Draw the figures directly in a computer, using preferably PowerPoint, Excel, or the drawing window of Microsoft Word.

Make sure the fonts and any symbols are big enough.

Do not make figures any wider than ~14 cm, because they need to be viewable in a Webbrowser window without the reader having to scroll sideways.

When using Word, paste each figure directly into the text using Paste Special…, unselect Float Over Text, and paste them in as bitmaps or drawings. Also, make sure the figure is displayed at 100% size and that it looks OK when the document is displayed at 100%

Put the figure into the cell of a table, as shown. Place the title and any footnotes for the figure in cells above and below the figure. The style for this text is 11pt Arial.

Place each figure or table immediately after the paragraph that first refers to it.

See the examples (Figures 14).
Figure 1: Informative title for a time series^{a}.


Data are means. Bars are standard deviations (shown only for Groups B and C). ^{a}Use letters to label footnotes, if necessary.

Figure 2: Informative title for a scattergram.


Leastsquares lines are shown for each variable.

Figure 3: Informative title for a bar graph.


Data are means. Bars are standard deviations.

Figure 4: Informative title for an outcomes figure.


Data are means. Bars are 95% confidence intervals.
 
Connect the points in a line diagram with line segments. Show curves only if you are modeling a curve to the data.

Change the color and shape of symbols for different groups of points: . This strategy helps colorblind readers.

Show scattergrams only for a good reason (e.g. to call attention to outliers, a nonzero intercept, heteroscedasticity, or a nonlinear trend); otherwise state the correlation coefficient and/or standard error of the estimate without a figure.

Hierarchical diagrams summarizing the relationships between concepts or variables can be confusing. Make them as simple as possible.
Share with your friends: 