Provide a substantive contribution that advances the discussion in a meaningful way by identifying strengths of the posting, challenging assumptions, and asking clarifying questions. Your response is expected to reference the assigned readings, as well as other theoretical, empirical, or professional literature to support your views and writings. Reference your sources using standard APA guidelines. Review the Participation Guidelines section of the Discussion Participation Scoring Guide to gain an understanding of what is required in a substantive response.
Peer 1 Response :Teddrick
If correlation does not imply causation, what does it imply?
Correlation can imply a relationship between two variables or more but not necessarily imply causation (Warner, 2013). Before indicating a causation because of the correlation all of the variables need to be considered. Warner (2013) describes that casual interference could be determined by the research design that could or not describe a relationship between the variables. Causation is only established when there is a study utilizing randomization, it is error free and it is well designed; even in with this description there will always be a level of uncertainty. Although two variables are related, does not mean that one causes the other. Also, a variable can be change and not be a direct cause from the change of other variables (Warner, 2013).
Are there ever any circumstances when a correlation can be interpreted as evidence for a causal connection between two variables? If yes, what circumstances?
Yes, there could be circumstances when a researcher could interpret a causal connection between two variables. An example could be a study on identical twins and grades, a set of twins that consistently get the same grades. If one twin for purpose of research does no study and goes to the mall and the other twin stays studying for a determine time, if their grades suddenly change with a large variance the researcher could determine that studying and going to the mall had a casual effect on their scores. This could present a case of causation.
Warner, R. M. (2013). Applied Statistics: From Bivariate Through Multivariate Techniques (2nd ed.). Sage Publications.
Peer 2 Response: Cait
Warner (2009) continuously stresses through Chapter 7 that correlation does not imply causation and never will. There are many types of correlation ranging from positive correlation, negative correlation, near zero correlation and Pearson correlations. Positive correlations clearly state that the scores on both the x and the y will increase, whereas negative correlations indicate that the scores on the x will increase while the y will decrease (Warner, 2009). Depending on the different types of correlations that are available, it is not always or ever possible to make casual interpretations of the meanings of the correlations. This can specifically be seen in the Pearson correlation. In the Pearson correlation, since the data is normally conducted, monitored and evaluated in non-experimental studies, researchers can not imply causation with the results (Warner, 2009). In situations such as this, “correlational design does not imply causation” (Warner, 2009, p.303). Correlation design will not be able to imply causation because the x and the y variables cannot be determined accurately because the research cannot be controlled in all research designs and/or experiments(Warner, 2009). However, Warner (2009) states that, “it is possible to make the case for a casual interpretation of a correlation somewhat stronger by statically controlling for some of the Z variables that you know are likely to be confounded with X, but it is never possible to identify and control for all the possible confounds” (p.303). If the data that is obtained can control the results of the z variable, as well as the x variable, there is more potential to use casual interpretation. Having statistic control over the x and y variables can make casual interpretation somewhat possible, however, it is extremely important to remember that it is never fully possible to be able to control all of the confounds in a research experiment (Warner, 2009).
Warner, R. M. (2013). Applied Statistics From Bivariate Through Multivariate Techniques (2nd ed.). Thousand Oakes, CA: SAGE Publications.