I recently came across the following paper:
Keller, A., Litzelman, K., Wisk, L. E., Maddox, T., Cheng, E. R., Creswell, P. D., & Witt, W. P. (2012). Does the perception that stress affects health matter? The association with health and mortality. Health Psychology, 31(5), 677.
This study tracked about thirty thousand adults in the US for 8 years and they started by asking people "How much stress have you experienced in the last year?" They also asked, "Do you believe that stress is harmful to your health?" And then they used public death records to figure out who died. They found that people who experienced "a lot of" stress over the past year had a 43% increased risk of dying, but that was only true for the people who also believed that stress is harmful to their health. People who experienced a lot of stress but did not view stress as harmful were no more likely to die. In fact, they had the lowest risk of dying of anyone in the study, including people who had relatively little stress. The researchers estimated that over the eight years they were tracking deaths, 182,00 Americans died prematurely, not from stress, but from the belief that stress is bad for them, which is over 20,000 deaths a year. If that estimate is correct, that would make believing stress is bad for you, the 15th largest cause of death in the United States last year, ...
Now, my questions is: "Is this study confounded?" I think the way they have measured "How much stress have you experienced in the last year?" highly depends on the reference point defined individually for the subject. I may see an event as very stressful, and someone else may see the same event as not stressful at all. [For more information about refernce point, please look at: Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica: Journal of the econometric society, 263-291.] On the other hand, they have asked: "Do you believe that stress is harmful to your health?" Again answering this question is very subjective depending on the same reference point.
Counter example: imagine subjects A has experienced so much disaster in their life that they have seen very negative health effects of stress. On the other hand, subject B has not experienced that much stress in their life. To both subjects, last year was highly stressful. Obviously, highly stressful for subject A means a complete disaster, and for subject B means a little worse than other years. When you ask them "Do you believe that stress is harmful to your health?", obviously, subject A responds yes because they have experienced very negative health effects, but subject B who has not experienced that much stress in their life would probably say no. Now, subject A will probably die earlier than subject B. Could you argue that dying earlier is because of believing stress is bad for you?
I believe this study is loosely measuring a correlation, and they cannot conclude a causal effect in this way. Please let me know if I am missing any point or I do not understand what they have done.