Critique research articles any advice

HouTx, BSN, MSN, EdD · Oct 19, 2016

The best approach is to compare & contrast the content of the article with the principles you have learned in your research class. For instance, you could point out the weaknesses of a study that claims an outcome is due to the intervention - but lacks a control group. Or you could examine the statistical measures that were used and see if they align with what you learned in your Stats class.... e.g., correlation does NOT equal causation, inadequate sample size, population is not truly representative, so results cannot be generalized, etc.

This is the fun part - when you actually get to use some of those foundational courses.

jamiewilder49 · Oct 20, 2016

Thank you

BrendanO, MSN, RN · Oct 20, 2016

If the prof is a stat-head (and you understand stats well), read up on manipulation of p-values. Lots of studies are easy to pick apart if the strength of their conclusions is based mainly on the p-values of their data distribution. Similarly for any regression study that does not properly contextualize their R-squared values (in some studies, R-squared of .3 is great, in others it's garbage). Look for high-leverage outlier data, particularly if excluded or not explained.

Look for unsupportable conclusions drawn from studies with non-placebo control groups, especially if the control is just "usual care", "waiting list", "educational material" or other non-intervention. Any intervention has an effect, so studies without some kind of sham treatment in their control group have limited power.

Look for studies that cannot be double-blinded, or where data analysis was not blinded.

Look for high or disproportional rates of subject disqualifications or drop-outs from either experimental or control groups.

Look for social biases related to subjects' lifestyle, social position, environment, or wealth that correlate with subject selection criteria or demographics of study group (e.g. an all-white group is statistically likely to have a higher income than a majority AA group, which can affect access to routine/preventive care; subjects solicited via advertisement in newspapers skew older; etc.)

Look for possible unmeasured/uncontrolled confounding variables related to the dependent variable, particularly in any out-patient study group (e.g. dietary interventions with only self-reported intakes, exercise regimens with variable scheduling, etc.)