Hi all,
When answering biostats exercises/assignments for this module is it okay to round means to one decimal place more than that presented in the data and SD/SEM to two dp? Is that an acceptable approach? I have seen this approach used elsewhere.
Regards
Simon
Can you give an example?
I didn't have a specific example from a research paper or data analysis, just a citation I was once given by an Australian epidemiologist. I am not sure how valid it is honestly...
"Means should not normally be given to more than one decimal place more than the raw data, but standard deviations or standard errors may need to be quoted to one extra decimal place"
from Gardner and Altman (Statistics with Confidence, 1989)
Is this approach fine for our coursework? What should we round up to?
Many thanks
Example from tutorial would be Q8) For age where raw data = whole years, mean = 53.3 and SD is 10.44.
These rules of thumb for rounding in statistics would be fine for the coursework and your project as well. As you have mentioned, some suggest that you round to one more decimal point (dp) than your original data. However, if your original data were mixed, round one dp more than the least precise. For example, consider the set 10, 8.7, 9.2, 11.4, 6, 5.3. Some are whole numbers, some have one decimal place. The least precise numbers in this set are the whole numbers, so you would report your mean and SD to one decimal place. So for your Q8 example, it would have been fine to report mean 53.3 and SD 10.4. But overall, we are pretty flexible with rounding on your HW...
For probabilities, I have seen people rounding out to 3 or 4 decimal places. The style of some journals (and the AMA) suggest rounding to just 3 dp. For example, if p=0.0003, you could report it as such or p<0.001 in some journals.
