Example 124 in 2008 the los angeles times published a

This preview shows page 9 - 11 out of 16 pages.

EXAMPLE 1.24 In 2008, the Los Angeles Times published a headline 33 that included “Hospitals… Riskier than a Casino in Event of Cardiac Arrest.” The article, based on a study published in the New England Journal of Medicine , states that the widespread availability of defibrillators and bystanders in public places like casinos leads to a higher survival rate than hospitals in the case of cardiac arrest. (a) What are the primary variables of interest in this study? Which is the explanatory variable and which is the response variable? (b) Give an example of one potential confounding variable in this study. (c) If you are having a heart attack, would you go to a hospital or a casino? Solution (a) The two primary variables of interest are the place of cardiac arrest (explanatory) and whether or not the person survives (response). (b) A confounding variable is the health of the person at the time of cardiac arrest. Older, frailer, sicker people are more likely to be in the hospital and also less likely to survive (not because they are in a hospital, but just because they are weaker to begin with). Someone in a casino is much more likely to be in better physical shape and thus better able to survive a heart attack. Notice that the confounding variable (health of the person) influences both of the variables of interest: where the person might be and whether the person is likely to survive. (c) If you are having a heart attack, you should go to a hospital! Even though casinos have a higher survival rate, this can be explained by the confounding variable, and we cannot conclude that being in a casino causes a higher survival rate. For a person of a given health status, it is probably safer to be in a hospital under the care of professionals. Many seemingly surprising claims in the media (such as that hospitals are riskier than casinos) can be explained simply by the presence of a confounding variable. Knowing how and when to be on the lookout for confounding variables is essential for statistical literacy and for assessing any data-based claims. Observational Studies vs Experiments How can we establish (statistically) when an association represents a causal relationship? The key is in how the data are collected. If we want to study how the explanatory variable influences the response variable, we have to be able to control or specify the values of the explanatory variable to make sure it is not associated with any potential confounding variables. Note that in data such as LifeExpectancyVehicles or the study of cardiac arrest we merely collect available data after the fact. We call data collected in this way, with no effort or ability to manipulate the variables of interest, an observational study . With observational data we can never be certain that an apparent association is not due to some confounding variable, and thus the association is not evidence of a causal relationship.
The alternative is to intentionally control one or more of the explanatory variables when

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture