The statistical methods should contain no results. In theory, this should be written before you start the analysis. It is the plan of the analysis you intend to do. So for example you might say that continuous data will be summarised as mean (SD) if normally distributed and median (IQR) if skewed. So here it is very clear how the analysis will change depending on the distribution of the data. The stats methods section and results should mirror each other. You wouldn't say 'missing data must be accounted for before carrying out the analysis' but rather you would describe how you plan to deal with missing data in your analysis. Is it okay to say a variable has a normal distribution in the statistical methods section? - no you shouldn't say this, the stats methods will say how you will assess distribution. All published papers have stats methods and result in sections so it is important you get used to writing in this way and it would be wrong for us to ask you to combine them into one as this does not reflect what happens in practice. It might be useful to look at some journal articles in journals such as BMJ, Lancet, NEJM to see how these are usually presented. 2.For the table for question 1, I would expect a bit more discussion around the points than a bullet point, so even if using bullets I would have thought these would be made up of sentences. But the format is really up to you - we are not prescriptive about this. The orientation doesn't matter, although portrait by default.3.My concern: Ethnicity is categorical data while HbA1c is continuous data, hence how can linear regression be used to adjust for variables etc in order to account for confounders? Professor replied: Age has an odd distribution here as you have been given an age-specific cohort, if you look at the mean and median they are similar and the distribution is not massively skewed so it should be fine to use parametric methods. This might be something to discuss in the discussion or to consider in a sensitivity analysis.
Copyright © 2021 | Truelancer.com