Getting Started with Research

Some helpful tips based on prior medical student experience...

To be successful:

  • Stay organized: Keep a detailed log of your experience (both the ups and the downs).
  • Seek mentorship: Collaborate with experienced clinicians and researchers.
  • Remember ethics: Adhere to ethical guidelines and obtain necessary approvals.
  • Think future oriented: In addition to asking how you can help with a project, think about how a project can help you. Ask about opportunities to write up and present data related to your work.  

To maximize your research experience: 

  • Be proactive: Ask questions and find opportunities to build upon the existing project. Identify how the experience can benefit you by defining roles and expectations early. For example, you might ask about opportunities to write an abstract or be included in a manuscript as a co-author.
  • Be reflective: Journal about your experience: what went well, what did not, and how did this affect your development as a clinician and/or scientist? This will help you during future interviews if asked about the research experience. 

To prepare for research in advance:

  • Complete your CITI human subjects training with the UMass Chan IRB
  • Request access to necessary services (e.g., TriNetX, REDCap) 
    • Currently REDCap access requires that a principal investigator approve a new account
  • Think about a research question - curiosity is key! Use your experiences to identify a gap in a process, understanding of disease, or treatment.

The AMA has other tips for medical students getting started in research.

Basics of Designing Your Research Question

  • Importance of a Good Research Question
    • Guides your study design
    • Determines your data collection and analytic methods
    • Helps focus your research
  • Characteristics of a Good Research Question (FINER)
    • Feasible: can it be done with available time, resources, and expertise?
    • Interesting: is it compelling and engaging, adding knowledge that others want to know about?
    • Novel: is it new, a different perspective, or an uncharted area?
    • Ethical: can it be answered with integrity, respect, and responsibility?
    • Relevant: will this resonate with the real world, and have practical implications in practice or policy?
  • Steps to Develop a Research Question
    • Identify a broad topic of interest
    • Conduct a literature review
    • Narrow down to a specific issue or gap
    • Frame it into a question
  • Good Research Questions Are:
    • Clear and focused
    • Not too broad or narrow
    • Not too easy or difficult to answer
    • Researchable

PICO Questions

  • What is PICO?
    • A framework to formulate research questions in healthcare
  • Components of PICO/PECO 
    • Patient/Population/Problem: Who is your study about?
    • Intervention: What intervention or exposure are you considering?
      • Exposure = for retrospective studies 
      • Intervention = prospective studies
    • Comparison: What are you comparing an intervention against?
    • Outcome: What do you intend to measure or achieve?
  • Examples of a PICO Questions
    • In patients with type 2 diabetes (P), how does metformin (I) compared to no medication (C) affect HbA1c levels (O)? 
    • In patients with type 2 diabetes (P) did the A1c level (O) differ in patients prescribed metformin (E) vs. no medication (C)?
  • For More Information, See the Evidence Based Medicine Guide

Basics of Research Design

Common types of research designs to consider:

  • Observational Studies
    • Descriptive Studies: Case reports, case series, cross-sectional studies
    • Analytical Studies: Cohort studies, case-control studies, retrospective cohorts 
  • Experimental Studies
    • Randomized Controlled Trials (RCTs)
    • Non-Randomized Controlled Trials
  • Qualitative Research
    • Interviews, focus groups, case studies

Retrospective Studies

Many of the projects we work on as trainees are retrospective studies. A retrospective study looks backwards in time and examines exposures to suspected risk or protection factors in relation to an outcome that is established at the start of the study.

  • Types of Retrospective Studies
    • Cohort Studies: Identify a cohort that was exposed to a risk factor and a cohort that was not, then look back to see how many in each group developed the outcome of interest.
    • Case-Control Studies: Identify patients who already have a specific condition (cases) and match them with individuals who do not have the condition (controls), then look back to compare exposures to risk factors.
  •   Advantages [things you can talk about in your discussions] 
    • Useful for studying rare diseases or diseases with a long latency period.
    • Can generate hypotheses that can be tested in future prospective studies.
  •   Disadvantages [issues inherent to retrospective design]
    • Higher susceptibility to bias (recall bias, selection bias).
    • Limited control over data collection methods.
    • Dependence on the accuracy and completeness of existing records.

Here is a nice blog providing an introduction to different types of study designs. 

Basics of Data Analysis

  • Types of Data
    • Continuous Data: Numeric values that can be measured (e.g., weight, blood pressure)
    • Categorical Data: Data that can be divided into groups (e.g., gender, blood type)
  • Analyzing Continuous Data
    • Descriptive Statistics: Mean, median, standard deviation
    • Inferential Statistics: t-tests, ANOVA, linear regression
  • Analyzing Categorical Data
    • Descriptive Statistics: Frequencies, proportions
    • Inferential Statistics: Chi-square tests, Fisher's exact test, logistic regression

Measures of Central Tendency: When to Use Median vs. Mean for Continuous Data

  • Mean
    • Definition: The average of all data points.
    • If data is symmetrically distributed without outliers, like a bell-shaped curve, it is likely normal data and can be compared with t-tests (2 groups) or ANOVA (>2 groups). 
    • Use Mean When:
      • Data is normally distributed.
      • Example: Average blood pressure in a group of healthy adults.
      • Example: Suppose you are studying the average weight of adults in a town. If most weights cluster around a central value and there are no extreme outliers, the mean gives a good summary of the central tendency.
  • Median
    • Definition: The middle value that separates the higher half (50% of values) from the lower half of the data set.
    • If data is not symmetric or has significant outliers, it is unlikely to be normally distributed data and should be compared differently (e.g., with Mann-Whitney U test): 
    • Use Median When:
      • Data is skewed or contains outliers.
      • Example: Median income in a population where there are a few extremely high incomes.
      • Example: If you are studying the average income in a city where a few individuals earn significantly more than the majority, the median provides a better representation of the typical income than the mean, which would be skewed by the high earners.
Normal vs. Skewed Data Distributions

Symmetrical Data:The histogram on the left shows a normal (symmetrical) distribution. Both the mean (red dashed line) and the median (green dotted line) are close to each other. In this case, either measure could be used to represent the central tendency, but the mean is often preferred due to its mathematical properties.

Skewed Data:The histogram on the right shows a skewed distribution. The mean (red dashed line) is pulled towards the skew (outliers) and is not at the center of the data. The median (green dotted line) is more representative of the central tendency as it is not affected by extreme values. In this case, the median is a better measure of central tendency than the mean.

Non parametric tests

Non-parametric tests does not require that your data be normally distributed.

Kruskal-Wallis Test

  • Purpose: Compare three or more independent groups.
  • Analogous Parametric Test: One-way ANOVA.
  • When to Use: When the data are not normally distributed or when sample sizes are small.
  • Example: Comparing the median scores of three different treatments.

Mann-Whitney U Test

  • Purpose: Compare two independent groups.
  • Analogous Parametric Test: Independent t-test.
  • When to Use: When the data are not normally distributed.
  • Example: Comparing the median recovery times of two different drugs.

Wilcoxon Signed-Rank Test

  • Purpose: Compare two related groups.
  • Analogous Parametric Test: Paired t-test.
  • When to Use: When the data are not normally distributed.
  • Example: Comparing pre-treatment and post-treatment measurements.

Approach to Data Management and Analysis

  • Data Collection: This begins with gathering raw data from different sources. This data is often stored in a database, file, or spreadsheet.
  • Data Cleaning: This involves addressing missing values, fixing errors, and ensuring the dataset is high quality. This should be reproducible (i.e., the same data cleaning process can be applied to the same data and produce the same results).
  • Data Transformation: Data may be transformed into another format for analysis. This may require normalizing or creating new variables from the raw data.
  • Data Visualization: Using tools like Tableau, Excel, R, or Python, create scatter plots, bar charts, histograms, and box plots. Check if the data are normally distributed and/or have outliers.
  • Statistical Summary: Calculate basic statistics, such as mean and standard deviation (or median and interquartile range) to summarize continuous data. Computer frequencies of categorical data. Inferential statistics may be useful in making comparisons.
  • Reporting: Summarize the research, writing up the research question, hypothesis, methods, results, discussion, and conclusion.

Present and Publish

An excellent scientific presentation first requires that you have content that is worthy of your audience’s time. Second, to motivate the audience to listen, you should show your passion for that content. Third, you need a keen sense of your audience: who they are, what they know, and why they are listening to us. (Michael Alley)

Are you creating a poster for the first time? Then check out this guide for design tips and use of Powerpoint. This guide also has some great templates.

Need some example posters for inspiration? eScholarship@UMassChan has archived posters available (all posters have a student author, but not always created by a student).

Want to publish?