Writing About and Interpreting Results

Author

NURS 60N

1 Objectives and Overview

This project is intended to help you:

  1. Synthesize parts of results that you have practiced in the previous assignments
  2. Present and explain those pieces separately and in totality
  3. Interpret potential implications of your results to practice and/or research

The Data section presents a couple different sets of data you may choose to complete this project; you may also use your own data (e.g., that you have permission to use from your hospital, etc.) if you prefer; please ask me to vet it first if you feel it may not pertain or be sufficient.

Please analyse one of those datasets (or your own). From your analyses, choose a collection of results that together you feel creates an interesting and useful message. This “message” should be more or less cohesive, presenting a single story with a main theme; it may have additional, supporting points to it, but try to keep your core message clear and to have all your results pertain to it1.

1.1 Data

You may use your own data. If you are unsure if it will work for this project, please ask I’ll let you know what I think.

Instead, you may use one of these sets of data. Links to the data themselves are in bold:

  1. The Midlife Development in the U.S. (MIDUS) data we used for the correlation assignment.
  2. Other/additional variables from the MIDUS study. We used a subset of these data for the correlation and partial correlation assignment. There are lots more. However, you will need to create an account with them to access these data. It’s not hard or intrusive—just to make sure the data are being used by those who understand the ethical and privacy issues—but is worth noting.
  3. The Agency for Healthcare Research and Quality’s Social Determinants of Health Updated Database. As noted in the documentation about the data, “to make it easier to access a range of well documented, readily linkable SDOH variables across domains without having to access multiple source files, facilitating SDOH research and analysis to improve health outcome.”
    These data are grouped into “tracts,” which are geographical areas created by the US Census similar to ZIP codes. Like the CDC data on cancer incidents, most variables here are counts of people in a give category within that census tract. Note too that these are the latest data available, related to the 2020 census. Keep in mind that Trump tried to interfere with the census, so these data aren’t as unbiased as earlier versions would be.
    Since these are essentially 2020 census data for the US, this is a large set of data, about 85,000 rows and 330 variables. It is rich, primary-source data about as individualized SDOH data as one can get without conducting their own study2—at least as much as these data aren’t biased. The data are available as:
  4. The effects of an empathy role-playing program for operating room nursing students. This is an MS Excel file of data from a study conducted by Larti et al. (2019). These data are more like those for research studies conducted by the likes of us.
    As per the site, “[t]his study was carried out among 77 operating room nursing students from the first to the fourth years studying at the School of Nursing and Midwifery of Isfahan University of Medical Sciences [in Iran] in the academic year 2017–2018. The intervention administered to the experimental group included a 12-hour training program on expressing empathy to patients that incorporated role-playing. The Jefferson Scale of Empathy-Health Profession Student version was completed by the participants before, immediately after, and 1 month after the intervention.”

1.2 Instructions

  1. Using Excel/Calc or SPSS, conduct descriptive and inferential statistics on variables in your chosen dataset. You needn’t analyze all variables or all combinations of them. I would, however, like to you explore them sufficiently that the final set of results you submit are interesting and well chosen.

    The types of inferential statistics—the tests you run—will depend on the types of data you’re investigating. In addition to what we’ve covered in this course (and the concomittant assignments & materials), I have collected a small set of online guides to choosing the right statistic. For example, Howell (2008) presents a simple-to-follow decision tree to choose most of the analyses you will likely use here.

  2. Choose a subset of your analyses that presents a single message. This message can—and likely will—have different pieces of evidence that all collectively support aspect of it, but try to create a single story with your data. One way to do this is to first posit a research question, and then use the rest of your deliverable to answer that question (and interpret it for research/practice).

    • For example, for the MIDUS data, a possible research question could be “do financial burdens affect the health of Blacks and that of Whites?”

  3. Create a Results section. In this section, present both descriptive and inferential statistics related to your message. It is typical to present descriptives first to help put subsequent inferential statistics into perspective. Of course, I also suggest using tables and figures to present details and patterns, respectively, while using the text to focus more on your message itself, using the results in the tables/figures and from your analyses to support and conditional the message. Some more of my thoughts on writing results are here.

    Please follow the convention of only describing the results and not making any subjective conclusions about them. “Just the facts.” You may use APA 7th style in your writing, formatting, and even conventions & strategies for reporting results. I think Chapter 4 in the Manual is actually very useful for guidelines.

    This Results section will likely cover 1 – 3 pages (250 – 750 words) of text. In addition to that, this section will likely include 3 – 5 tables/figures; these can go within the text or as their own pages, whichever you prefer.

  4. Create a Discussion section after the Results. It is here that you may—should—give your opinions about the results you presented above. The general structure of this section should:

    1. Begin with a concise summary of the main message given in your Results section. This is often a single sentence and it commonly presented as the topic sentence for the first paragraph of the Discussion section.

    2. Complete the rest of that first paragraph with summaries of evidence that support/conditional what you reported in that concise summary.

    3. Provide a few (i.e., not many) additional paragraphs that give additional supporting evidence for any suggestions that you feel need it. (E.g., that couldn’t be effectively summarized in the first paragraph of your “Discussion” section.)

    4. Create a Limitations subsection of your Discussion that presents any important limitations on the conclusions you drew from these data. It’s likely you’ll have at least one, and it may well be because not everything you needed (to make your decision well) was available in the data3.

      Please also use these limitations throughout what you write in both the Results and main section of the Discussion.

  5. Create an Implications section that provides concrete ways your message should guide practice and/or actionable ways your message could advance research4. You needn’t address both, and—indeed—one good example should suffice if it demonstrates real insight into your results. (Some publications ask authors to write a Conclusion section that works a lot like this.)

  6. In addition to these sections, you may include a Reference section that informs your main points and serves as practice creating one, say with a reference manager.

Your deliverable then will include:

  1. Sure, a title page
  2. Statement of research question (optional)
  3. Results (1 – 3 pages of text plus graphics)
  4. Discussion (1 – 2 pages)
    1. Limitations (1 – 2 pages)
  5. Implications (1 – 2 pages)
  6. References (optional)

Thanks!

Footnotes

  1. Note that descriptive statistics help set the stage, letting the reader know whom these data are actually about. So, it’s nearly always pertinent to present the mean, SDs (or SEMS), N, etc. Even there, though, you only need to present the descriptives that matter.↩︎

  2. Or joining something like All of Us↩︎

  3. Believe me, that’s often even the case with our own data. Let alone others’.↩︎

  4. Please, try never to type “larger samples and more research are needed.” That much we already know; what else can you tell me?↩︎