Blog

Suppose a team of researchers investigated the effect of drinking caffeinated beverages (i.e., yes versus no) on migraines among American women aged 18-40. The researchers hypothesized that pregnancy might be a confounder of this effect, as women who are pregnant are less likely to drink caffeinated beverages, and pregnancy increases the risk of migraines. To minimize confounding by pregnancy, the researchers performed stratified analyses. The results from these stratified analyses are reported in Table 1.   Table 1. Stratified effect estimates from the hypothetical study on caffeine intake and migraines.   Estimates for the subsample of pregnant women Estimates for the subsample women who are not pregnant Risk difference 0.05 0.05 Risk ratio 1.50 2.00   The researchers would like to recombine the stratified effect estimates reported in Table 1 using Mantel-Haenszel methods. Is it appropriate to apply the Mantel-Haenszel method to recombine the stratum-specific estimates of the risk differences (yes/no)? (1 point) Describe why it is/is not appropriate to apply the Mantel-Haenszel method (max. 2 points). Is it appropriate to apply the Mantel-Haenszel method to recombine the stratum-specific estimates of the risk ratios (yes/no)? (1 point) Describe why it is/is not appropriate to apply the Mantel-Haenszel method (max. 2 points).

Suppose a team of researchers investigated the association between obesity and gallstones using a prospective cohort study design. The researchers were concerned about confounding by high cholesterol diet, so they performed a few different analyses in which they attempted to minimize confounding. The results of their analyses are reported in Table 1.   Table 1. Unadjusted, stratified, and Mantel-Haenszel effect estimates from the hypothetical study on obesity and gallstones.   Unadjusted Estimates Estimates for the subsample with a low cholesterol diet Estimates for the subsample with a high cholesterol diet Mantel-Haenszel estimates Odds ratio 3.24 3.01 2.83 2.92 Risk difference 0.20 0.16 0.17 0.17   Based on the information in Table 1 report the following information: Is it appropriate to estimate a odds ratio using Mantel-Haenszel methods (yes/no)? (2 points) Explain why this is/is not appropriate (2 points). Is it appropriate to estimate a risk difference using Mantel-Haenszel methods (yes/no)? (2 points) Explain why this is/is not appropriate (2 points).

Suppose a team of researchers is interested in investigating the effect of physical activity on atherosclerosis. Before conducting their study, they drew a directed acyclic graph (DAG) to determine what confounder variables they need to adjust for in the analyses. Figure 1 shows the DAG that the researchers drew.   For each of the variables in Figure 1 (stroke, LDL cholesterol, and socioeconomic status) describe 1) what type of “third variable” each variable is (1 point), 2) whether the researchers should or should not adjust for the respective variable in the analysis stage to minimize confounding (1 point), and 3) why the researchers should or should not adjust for the respective variable (1 point).  

This class is a full term class.

Suppose that a team of researchers developed the following research question: Are American adults who undergo a mindfulness intervention at lower risk of depression after 6 months compared to American adults who were in the control group?   What are the two average potential outcomes (i.e., E[Y(x)] and E[Y(x*)]) that can be defined based on this research question? (max. 4 points per average potential outcome)  

According to research, is there typically an overall dominant side of the brain? 

Match the terms with the correct definitions/descriptions. 

Build a term that means incision into the eardrum. 

Suppose a team of researchers performs a case-control study on the effect of childhood adversity on psoriasis (a skin condition characterized by scaly patches of skin). Cases are individuals with psoriasis and controls are individuals without psoriasis. Stress was measured in interviews by asking individuals about any adverse events experienced during childhood. During the interviews, the interviewer tended to ask more clarifying questions to individuals with psoriasis compared to individuals without psoriasis.   What source of measurement error is described in the case above? Pick one (1) source (2 points) and clarify why you chose this source (max. 2 points). Note: if you write down multiple sources, only the first one will be graded.

Build a term that means inflammation of a joint.