Section outline

  • Lecture:  Multi-level and causal analyses

    Approaches to analyzing complex data to investigate influence across multiple levels and to estimate causal effects.

    Faculty:  Meghan Morris & Kelsey Holt

    Location:     Mission Hall 1406


    Guidelines for leading journal article presentations during class:

    Here is a rough guide to help you prepare. You're welcome to adapt to meet the journal article or your needs.

    o   5 minute high-level summary – one slide to introduce the key points of the article (objective/methods/key results)

    o   Some reflections/questions about the paper and the content for discussion.

    o   Then start the larger conversation for the article – this could be you answering or starting a conversation about each of the article discussion question or pose new questions related to theory, methods, or application/implications of the article.  

    • Session Slides:

    • Session Audio/Video Recording (Access restricted to registered students):

    • Required Reading:

    • Diez-Roux multilevel File
      Not available unless: Your ID number contains 02
    • Kawachi CausalInf MoneySchoolingHealth File
      Not available unless: Your ID number contains 02
    • Tomita-Multilevel modeling File
      Not available unless: Your ID number contains 02
    • Beck-Causal Inference File
      Not available unless: Your ID number contains 02

    • Resources:

    • Assignment: Below are discussion questions we will consider in class, that you should be thinking about as you do the reading. If you will not attend class, please submit written responses to each question (email to meghan.morris@ucsf.edu).

      Multilevel modeling:

      Diez-Roux article:

      1. Multi-level modeling approaches aim to analyze contextual effects beyond individual characteristics. Briefly describe the benefits of conducting a multilevel analysis for health disparities research areas. 

      2. Draw from your own research or interest area to give an example of macro-level factors or group-level properties that influence individual health.

      3. Define the ecological and atomistic fallacies and explain how these fallacies differ from psychologistic and sociologistic fallacies. 


      Tomita article:

      1. The authors research stems from social disorganization theory which emphasizes the importance of one’s neighborhood and social place on their quality of health. In your own words summarize the author’s study rationale and main hypothesis. 

      2. What are the various community, social and individual measures included in the analyses? In your opinion, what are the strengths of including these measures? What additional measures could have further strengthened the author’s analytic model?

      3. The authors included three interaction terms into the full model (model 2h). Describe the value of including these interaction terms in relation to the author’s study objective. How would you interpret these interaction terms? 


      Causal Inference:

      Kawachi article:

      1. Describe what reverse causation and omitted variable bias (“confounding”) are, and give examples of how these might be present in an area of health disparities research that you are interested in.

      2. In what ways does conducting a longitudinal study address reverse causation and confounding in a way that cross-sectional studies are unable to do?

      3. Why are experiments considered the strongest study design to allow for causal inference?


      Beck article:

      1. DAGs, like the one presented in this article, are critical tools for researchers interested in drawing causal inferences from study findings. Draw a simple DAG relevant to an area of health disparities that you are interested in.

      2. How is the concept of race conceptualized differently in this study compared to other epidemiologic studies you have read where race is included as a predictor variable?

      3. The IPTW method described in this article is complex and it is beyond the scope of this course to expect a complete grasp on this analytical approach. However, in simple terms, what is this approach trying to accomplish?