Writing up your Methods section

Posted: April 22, 2011 in Useful advice for students
Tags: , , ,

Your Methods section should describe the specific procedures you used to:

  1. collect the data
  2. analyze the data

This blog post provides details about these two sub-sections that I expect to see in TECH 621 final papers. The details are specific to the research tool you used, in this case, Radian 6:

Data Collection
  1. State that data was collected using online monitoring service Radian 6. Provide a brief description of Radian 6 as well as all the widgets that you use. Don’t assume readers know them. Write the paper as if it were a conference or journal publication (some of them will be) – write so an audience who is not familiar with R6 understands exactly what you did.
  2. Describe all the procedures you used to identify and collect your data. How did you select your final data set? Specify: A) search terms; B) date range; C) language; D) region.
  3. Explain what you did to ensure that irrelevant data is filtered out of your data set. How did you filter out spam? How did you filter out posts that included the key words, but were not relevant to your topic?
  4. If you narrowed down your data set, explain every single decision you made. For example, after seeing the topic trend for a month, you see that there was more conversation during a particular week. Then you decide to focus only on that particular week. Within that particular week, you see that most of the chatter was on Twitter. Then, you can decide to focus only on Twitter for that particular week. It is fine and desirable to narrow down your scope this way – but you have to have good reasons behind every single decision.
  5. Once again, make sure you explain what the individual widgets do – and even what a widget is. Do not assume readers are familiar with them!
  6. For some students, even the final, narrowed down data set is too large to work with. In this case, if you want to perform manual content analysis, you need to draw a probability sample of items and only analyze those. I recommend stratified random sampling.Make sure you use a randomizer engine or random number generator (available at random.org) to draw a probability sample from each one of your groups/strata – in your case, days.
  7. Finally, describe your final data set. How many messages, from what media types, across what period, language, and region. We need to know, in exact detail, what was the data set that you performed your analyses on.
Data Analysis
  1. Explain, step by step, what you did to the data in order to derived meaning from it. Provide our reasons for every single decision.
  2. Avoid saying “I analyzed the data.” – what kind of analysis did you perform? Name it, cite it. If content analysis, then briefly explain your procedure and cite a source. One substitute or, preferably, addition to content analysis that I found is to pull up day-by-day or even hour-by-hour conversation clouds. The change in words provides a story about how the conversation topics changed.
  3. I understand that it may make sense to include some of this information with the results, in the Results section. That is fine. In this case, provide a general overview of data analysis methods in the Methods section, and provide details for each analysis along with the results of that analysis.

So far, all the draft I have seen need to follow this advice. Please do, and let me know if you have any questions. Also, I’d appreciate seeing feedback that you read this post. Please comment, like it, or assign a star rating below.

Dr. V

Advertisements
Comments
  1. Amanda says:

    Great advice! I have some remodeling to do with my methods section and I really appreciate the pointers! 🙂

  2. scott abney says:

    Def helpful advice!!!!!

  3. Jackie Riley says:

    Thanks for the tips. I appreciate all the help!

  4. Sarah Cox says:

    Thanks for the advice!

  5. Trish Forant says:

    Love how you’re helping others learn. Excellent point about no making assumptions. I just tweeted yesterday about something similar. You never know who your readers may be so it’s best to explain things in the simplest of terms.

    Best wishes,
    Trish (@Dayngr)
    Community Manager at Radian6