How to write #4- the methods

Hopefully by the time you get to this section of the writing guides series you will have already considered the guide to writing an introduction. What I hope I conveyed in that piece was the importance of being clear and succinct, of taking a very focused approach to setting up the reason for doing your research study. That is a theme that I will continue here, as the same principal applies to the methods.
 
The importance of the methods section cannot be over-stated. The validity and quality of any piece of research is generally assessed solely through consideration of how the study was done. A paper which has interesting results, but invalid methods is relatively worthless, whereas a well-designed and executed piece with null findings is potentially valuable to the field of study. Nobody should ever take the findings of a paper at face value, and it is the methods section that acts as the focal point for questioning and quality assessment.
 
What is the purpose of the methods section?
The methods section needs to convey a variety of information to your reader, in order to meet their needs. People will be reading your methods for a number of reasons:
  1. They want to do the same measurements themselves and want to follow your technique
  2. They are assessing the quality of your study
  3. They want to understand how you generated your results
 
This means that your methods section needs to tick a lot of boxes including:
  1. A description of your study protocol and what you did to answer your research questions
  2. The reasoning behind the choice of methods used to measure endpoints
  3. A clear description of how you made your measurements and obtained your data
  4. An account of how your data was analysed.
 
What should the methods section include?
 
Starting off
The way your methods section opens up will depend upon what sort of study you have done. If it is a laboratory based study, working with animals or in vitro systems then your first paragraph will need to state what species you used, what type of cells and give details of where you purchased materials such as chemicals and reagents. If you are describing an epidemiological or clinical study, then these elements are not required and you can move on to describing the protocol.
 
The protocol
This is the most important part of your methods section and is where all of the detail and uniqueness of your study is concentrated. In the simplest terms, this is the part where you explain what you did to generate your data. I would start off by doing just that in order to build a framework for the detailed version. For example you might write this.
 
Men and women were recruited from a clinic and were randomised to receive either a placebo or a supplement of calcium 1200 mg/day. Bone mineral density was determined by dual X-ray absorptiometry (DXA) at the start of the study. Supplementation was maintained for 24 months at the end of which bone mineral density was again determined by DXA.
 
Well, that’s a good start and it delivers the bare bones of the protocol, but there is a lot more detail that must be added to make this suitable for publication:
 
Men and women (why men and women?- justify this) (how old were they?) (how many?) (was a power calculation done to determine the required number?) were recruited from a clinic (where was the clinic?) (were they attending because they had particular health conditions?) (how were they recruited?) and were randomised (how were they randomised?) (is this a double-blind study?) to receive either a placebo or a supplement of calcium 1200 mg/day (in what form?) (is this an oral supplement?). Bone mineral density was determined by dual X-ray absorptiometry (DXA) at the start of the study. Supplementation was maintained for 24 months (why 24 months) at the end of which bone mineral density was again determined by DXA. (Was there any measurement of compliance?), (did anyone drop out?) (were reasons for drop-out recorded?)
 
So, the full version of the protocol would become this,
 
Recruitment was from a GP clinic (East Midlands, UK) and included participants aged between 45 and 65. The main objective of the study was to examine the effect of calcium supplementation on bone health, but as a secondary objective was to consider the sex-specificity of the effect both men (n=150, age range 46-63 years) and women (n=150 age, range 45 to 65 years) were included in the study. All participants completed a pre-trial screening questionnaire and were excluded if they were undergoing treatment for osteoporosis or other bone conditions (Pagets disease, osteomalacia) or had suffered a fracture in the preceding 12 months. The trial was performed double-blind with randomisation being performed by allocation of a code set by a third party. 
Controls (administered placebo) were matched by sex and age (±2 years) to participants receiving supplement (calcium carbonate, 1200 mg calcium per day administered as one daily oral tablet). Participants were provided with their doses once per month (28 day supply) and unused tablets were collected to determine compliance with the protocol. The duration of the study was 24 months in order to allow for long enough follow up between baseline and final outcome measure for an effect of supplementation to develop based on the results of Brown et al (2001) and Smith et al (2005). Bone mineral density was determined by dual X-ray absorptiometry (DXA) at the start of the study and at 24 months. Prior to study the sample size required to detect a 4% increase in BMD at the femoral next was determined based on the work of Smith et al (2005). The required sample size was 132 per group and based upon the experience of Jones et al (2010) and Cooper et al (2011) we allowed for 12% loss to follow up in establishing the initial group sizes. Among the placebo group 6 men and 5 women were lost to follow up (7.3%) whilst 10% of the supplement group failed to complete the 24 month protocol with a compliance of more than 95% of doses taken (8 men and 7 women).
 
Obviously the nature of your protocol will depend upon the type of research that you have done. The key message to take on board is that you need to capture as much detail as you possibly can. Pack the section with detail but remember that you are not writing a novel. There are no prizes for florid language and lengthy description. Short, sharp, focused is the key.
 
 
Methods for measuring the endpoints
Now of course the number of methodological points that I could make here is as great as the number of angels that could dance on the head of a pin, stars in the galaxy, or whatever other analogy for a big thing that you care to think of. Whatever you want to measure, there will be a means of writing the description of what you did. It would be foolish of me to attempt anything more than to give you some basic rules to work from here.
 
  1. Be brief and cross-reference to methods in other papers unless you really need to describe something that you have developed from scratch or modified very heavily. Most of us are simply repeating standard methods from our own papers, or from other researchers papers. It is sufficient to state that “total circulating insulin concentration was measured using ELISA following the method of Bloggs et al (2007)”. If you have used an off the shelf kit, then it can be even simpler. “total circulating insulin concentration was measured using an CrystalChem Elisa kit, according to the manufacturers instructions”. Where a method has been adapted from a published method, then cite the published method and describe what your adaptations were. If your method is completely new, then write it up in full and give all the detail that readers would need to replicate it.
  2. Break the methods up into sub-sections to make it easier to follow. Your reader may be interested only in a small element of your methods section and so headings will direct them to what they want.
  3. Use a logical sequence in listing the methods that you used. Try to do this following a chronological sequence. For example, lets say you collected blood samples and then measured total lipid content and then looked at the fatty acid distribution within the lipid, then it makes a nonsense to set out your gas chromatography method ahead of the method for collecting and preparing the blood samples. This is an obvious example here, but sometimes the more subtle distinction is just as important. In a study where patients were subject to physiological and anthropometric measures in clinic and then metabolites were measured in blood samples, the logical order is to give the methods used for the anthropometry and physiological measures and then move on to the clinical assays. This is because within the protocol the physical measures will have been recorded whilst the patients were in clinic, and the blood results will have been generated in the lab some time later.
 
Ethical matters
With very few exceptions (e.g. in vitro studies working with cell lines) your work will have required some level of ethical approval. This needs to be stated briefly and clearly. For example,
 
Ethical approval was obtained from the local medical ethical committee.
 
Or,
 
Work with animals was performed in accordance with the Home Office Animals (Scientific Procedures) Act 1986.
 
There is no need to give approval numbers or any other detail.
 
Statistical analysis
Unless your data has been collected using qualitative methodology (which is beyond my expertise and not covered within this guide), you will have performed at the very least some basic statistical analysis. Your methods section must describe what you have done and ideally how your data was presented. For example, 
Data is presented as mean ± S.E.M throughout the paper. All data was analysed using one way ANOVA with a Tukey test for post hoc testing. P<0.05 was accepted as statistically significant.
 
would be an absolute minimum, which may be sufficient for many papers. Obviously the more complicated your analysis was, the more you will need to say. You might also, within this section want to write a statement about the statistical power of your study, or what power calculation was used to define the sample size before you started work.
 
I am not personally a fan of stating which statistical package has been used to perform the analysis, unless what you have done is very complex and your software was essential for delivering the results. Generally speaking a one way ANOVA is a one way ANOVA, regardless of whether it was performed with SPSS, Minitab or Excel (note, other statistical packages are available…).
 
 
What shouldn’t the methods section include?
 
Even though many papers are not actually physically printed, journals have constraint on space and so the editor of your paper will appreciate brevity in describing what you did in performing the study. This means that, as outlined above, you cross-reference to published methods as much as possible, and where you do need to describe in detail, you do so in a bare bones format. Remember that the methods section of the paper is not the same as a laboratory or clinical protocol, so whilst your method should be clear enough for someone to pick it up and repeat the work, you don’t need to make it a completely step-by-step guide.
 
Hence, 
To measure blood pressure, the subject was seated in a chair approximately 50 cm away from a desk set at waist height. With the right arm extended and resting on the desk a velcro sealing inflation cuff was placed over the wrist, with the pressure sensor located over the brachial artery. With the patient completely relaxed the operator switched on the Omron blood pressure machine and activated the inflation cycle by pressing the start button. The cycle took approximately 60 seconds to complete after which the recorded systolic and diastolic blood pressures and heart rates were entered into an Excel spreadsheet. Three test cycles were followed for each subject and the average determined.
 
is expressed as,
 
Blood pressure was determined with subjects in a seated position, using an automated wrist cuff system, following the manufacturers instructions. The mean value of three repeats was used in analyses.
 
Clarity and simplicity is the key. Extraneous detail simply clouds the issues, bores your reader and actually makes it harder to follow your work. There is no need to provide lengthy descriptions of the equipment that you have used unless the equipment is particularly noteworthy. For example, if you have made a lot of measurements of body mass index, the maker of your scales and stadiometer is really not of any importance. It is more important that they were calibrated and used consistently by all researchers responsible for making the measurements. But, if you have been carrying out deep sequencing of DNA to look at epigenetic markers, then the system used for sequencing is actually very important and should be reported. In other words, mention the special and non-standard, but don’t report the ordinary ‘bog standard’.
 
 
 
My advice is that in preparing to write this section of your paper, you look at research published by others working in your field. Capture the level of detail that they have used and consider what they didn’t say, but you wish that they had when you came to read and follow their work.
Advertisements

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s