Lec14 - Confounding A problem with extraneous variables...

Info iconThis preview shows pages 1–4. Sign up to view the full content.

View Full Document Right Arrow Icon
Confounding: A problem with extraneous variables Association is not causation Just because changes in your explanatory variable are associated with changes in your response variable, doesn’t mean that the first causes the second. There may be some other factor (an extraneous) that is actually causing the change in the response variable. When results can be explained in more than one way, the results are said to be confounded Example : Smoking and birthweight Pregnant women who smoke give birth to significantly smaller babies than non-smokers ( P < 10 -5 ). Smoking may be the cause, but it may not. Smokers differ from nonsmokers in several ways. For example, smokers drink alcohol more often than nonsmokers. Perhaps alcohol consumption causes the smaller birthweight. If so, we say that the effect that smoking has on birthweight is confounded with the effect that drinking has on birthweight.
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Confounding: A problem with extraneous variables X = explanatory variable (smoking) Y = response variable (birthweight) Z = extraneous variable (drinking) Z X Y The effect of X on Y is confounded with the effect of Z on Y Nicotine birthweight Alcohol Observational studies can provide information about causality but you must be cautious. Causal interpretations often require extra support like : • several observational studies that consider various external factors (e.g. match smokers who stopped smoking with second pregnancy with smokers that continued to smoke through second pregnancy,matched for birthweight of first child) • additional experimental evidence (e.g. higher blood/urine levels of cotinine in infants; poor circulation of blood in placenta within 3 hours of smoking) X Y X causes Y?
Background image of page 2
Designing experiments An experiment is a study in which the researcher intervenes and imposes treatment conditions An experimental unit is a unit to which the treatments are assigned Randomization: Assigning treatments at random Example: Imagine that you are each a plant and this classroom is a field, and I would like to do an experiment to look at the effects of nitrogen on plant growth. I am going to fertilize half of you at random and measure growth after two weeks. 1. Each plant is assigned a number. 2. I choose one half of the plants to be treated using a random number generator or a table.
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
Image of page 4
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}

Page1 / 18

Lec14 - Confounding A problem with extraneous variables...

This preview shows document pages 1 - 4. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online