Monday, April 25, 2016

Dvorsky's "Lab Mice Are Freezing Their Asses Off—and That’s Screwing Up Science"

This example can be used to explain why the smallest of details can be so important when conducting research.

This piece by Dvosrsky summarizes a recently published article that points out a (possible!) major flaw in pre-clinical cancer research using rats. Namely, lab rats aren't being kept at an ideal rat temperature. This leads to the rats behaving differently than normal to stay warm: They eat more, they burrow more, and their metabolism changes. The researchers go on to explain that there are also plenty of other seemingly innocuous factors that can vary from rat lab to rat lab, like bedding, food, exposure to light, etc. and that these factors may also effect research findings.


Why is this so important? Psychology isn't the only field dealing with a replicability crisis: Rat researchers are also experiencing difficulties. Difficulties that may be the result of all of these seemingly tiny differences in lab rats that are used during pre-clinical research.

I think this could be useful as it is an example that students can easily grasp. Rat research is used in psychology, but is used here within a medical context, thus reaching students beyond our psychology majors.

Also, it is always a good time to share this story from The Onion about lab rats...

Monday, April 18, 2016

Weinberg's "How One Study Produced a Bunch of Untrue Headlines About Tattoos Strengthening Your Immune System"

In my Honors Statistics course, we have discussion days over the course of a semester. One of the discussion topics involves instances when the media has skewered research results (for another example, see this story about fitness trackers,)

Jezebel writer Caroline Weinberg  describes a modest study that found that people who have at least one previous tattoo experience a boost in their immunity when they get subsequent tattoos, as demonstrated via saliva samples of Immunoglobulin A. This is attributed to the fact that compared to tattoo newbies, tattoo veterans don't experience a cortisol reaction following the tattoo. Small sample size but a pretty big effect.

So, as expected, the media exaggerated these effects...but mostly because the researcher's university's marketing department did so first. Various new outlets stated things like "Sorry, Mom: Getting lots of tattoos could have surprising health benefits" and "Getting multiple tattoos can help prevent colds, study says". Lots of creative extrapolation.

I like this example as it appeals to the Millennials (tattoos!) as well as the psychology/biology/pre-professional science majors (stress/immune responses/cortisol responses) in my classes AND it is a pretty easy study to follow.

One way to use this with your students is to have them identify the questions that need answered before we can make the sort of claims that were made by journalists. How long does the immune boost last (we have no idea)? Is it really an immune boost or just the fact that repeated tattoo-er people don't experience a decrease in immunity? If so, would it be more appropriate to state "Don't get your first tattoo during cold season!"? What are ways to replicate this studying with other painful, repeated experiences (child birth? allergy shots? OT/PT that hurts? distance running? cross-fit class? piercings?)? Does the decrease in immunity, while significant and a large effect size, actually translate into the recently re-tattooed having fewer colds or with the recently first-time-tattooed having more colds?

Monday, April 11, 2016

Bichell's "A Fix For Gender-Bias In Animal Research Could Help Humans"

This news story demonstrates that research methods are both federally monitored and that best practices can change over time.

For a long time, women were not used in large scale pharmaceutical trials. Why did they omit women? They didn't want to accidentally exposed pregnant women to new drugs and because of fears that fluctuations in females hormones over the course of a month would affect research results. Which always makes me think of this scene from Anchorman:



But I digress. This has been corrected for and female participants are being included in clinical trials. But many of the animal trials that occur prior to human trials still use mostly male animals. And, again, policies have changed to correct for this. This NPR story details the whole history of this sex bias in research. Part of why this bias has been so detrimental to women is because women report more side effects to drugs than do men. So, by catching such gender differences earlier with animal models, there is the hope that fewer drugs will be developed that could hurt women. Additionally, such research has already uncovered sex differences that could benefit women and lead to new medical treatments for women. One study found that the way in which pain is communicated at the cellular differs between men and women (which may change how we treat pain in men and women). Another study may have uncovered a novel way to treat MS symptoms after finding that pregnancy hormones reduce MS symptoms.







Monday, April 4, 2016

Shapiro's "New Study Links Widening Income Gap With Life Expectancy"

This story is pretty easy to follow. Life expectancy varies by income level. The story becomes a good example for a statistics class because in the interview, the researcher describes a multivariate model. One in which multiple different independent variables (drug use, medical insurance, smoking, income, etc.) could be used to explain the disparity the exists in lifespan between people with different incomes.

As such, this story could be used as an example of multivariate regression. And The Third Variable Problem. And why correlation isn't enough.

In particular, this part of the interview (between interviewer Ari Shapiro and researcher Gary Burtless) refers to the underlying data as well as the Third Variable Problem as well as the amount to variability that can be assigned to the independent variables he lists).

SHAPIRO: Why is this gap growing so quickly between life expectancy of rich and poor people?
BURTLESS: We don't know. More affluent Americans tend to engage more in systematic exercise. They are less likely to be obese. Their smoking rates are lower. Those differences can help account for why there is a difference in how long people live. However, they do not seem to account for more than about a fifth of the increase in mortality differences between affluent and less-affluent people. So something else is going on in the background. It could be that there are behaviors that we could not examine. None of our surveys that we have access to asked about people's use of illegal drugs. Possibly, it is simply the effect of growing income inequality over the course of people's lives.
 I also like this example because it can be used to open your students minds up with how to study specific hypotheses after a result is known. Alright, a correlation has demonstrated this disparity. Why? How can we study this better? The author presents a list of factors that have already been examined. What else could there be?

You could also talk about how to model this data to allow for moderators. The interviewer says that health insurance helps offset this disparity. But does it effect everyone equally? Perhaps number of cigarettes smoked per day moderates this relationship, such that health insurance narrows the disparity with non-smokers but doesn't assist smokers.