mediating vs. confounding variable

[theonlytycrane]

Research goal: To investigate how patient race may affect pain medication disbursal.

To me, having insurance or not is an entirely different reason as to why patients may be receiving different amounts of pain meds. It doesn't help to provide insight into the relationship between race -> med disbursal. Is confound inappropriate here?

---

Members don't see this ad.

 

[bobeanie95]

Insurance would be considered a mediating variable. This is because it serves to describe the relationship between ethnicity and amount of prescriptions... it offers a "because." In other words, E (ethnicity)---> A (amount of prescriptions), insurance would be the link between E and A: E (ethnicity)---> I (insurance)---> A (amount of prescriptions).

Confounding variable would be wrong because a confounding variable is a variable not taken into account during the study. In this case, the researchers note that insurance affects prescriptions amounts, so it wouldn't be confounding.

 

[theonlytycrane]

@bobeanie95

If the q-stem said something like: "Researchers failed to account for whether patients had insurance and if that determined pain med distribution."

Would this change the factor to a confound?

 

[wizzed101]

yes

 

[bobeanie95]

@bobeanie95

If the q-stem said something like: "Researchers failed to account for whether patients had insurance and if that determined pain med distribution."

Would this change the factor to a confound?

Yes, that would be correct. They could also ask you which could be a confounding variable, without mentioning it in the passage, and that would be a legitimate answer choice.

 

[victorias]

A confounding variable is one that the researchers failed to control for and it had an effect on the results. These variables lower internal validity.

Without knowing the specifics of this research - some possible confounders could be age (did the researchers have the same age group of people in the Hispanic versus Caucasian groups?), gender (were there equal numbers of males and females in each of the two groups?), types of diseases (maybe Hispanic patients had diseases that didn't require pain medication). These are all factors that the researchers could have controlled for from the beginning of the study.

Mediating variable is one that could have had an effect on the results and is something that the researchers could not control for.
Another example of mediating variable is cultural beliefs - maybe Hispanic culture doesn't look too favorably at using medications, they might prefer natural remedies or traditional practices. This is something that the authors could not control for - it just would't be possible.

In this question, researchers could not have controlled for insurance levels - it is just not possible to do that so it is a mediating variable.