Everyday Research Methods

An empirical reason to say "Thanks"

Senders of gratitude letters benefit from writing them, and recipients "feel great" when receiving them. Photo credit: Rawpixel/Deposit Photos

This example comes from the subfield of positive psychology, and the study of gratitude. One technique studied by positive psychologists is the practice of writing a gratitude letter to a person who has influenced you.  According to this summary, "There are two excellent reasons for writing a gratitude letter: It will make you feel really good, and it will make the recipient feel great"

The concept is simple: Write a letter to a person whom you'd like to thank. Gratitude letters can be handwritten or typed, emailed or snail-mailed. The important thing is to send them. Gratitude letters can help both the writers and the recipients. 

First, here's the journalist's description of a study about the benefits of writing these letters:

Scientists randomly assigned the 293 participants to three groups: Those receiving psychotherapy, those receiving psychotherapy and participating in expressive writing, or those receiving psychotherapy and participating in gratitude-letter writing. Even in the small study, participants in the gratitude group reported significantly better mental health than the other two groups, even three months after the trial ended.

a) What is the independent variable (IV) in the study above?  What are its levels?

b) What is the dependent variable? 

c) Was the IV manipulated as independent groups or within-groups? 

d) What kind of experiment was this? Posttest only? Pretest-posttest? Repeated measures? or Concurrent measures? 

e) What role do you think the "expressive writing" group played in this study? Use the term "design confound" or "control variable" in your response. 

f) Can this study support the claim that "writing gratitude letters (in addition to psychotherapy) improves mental health"? Apply the three causal criteria. 

g) Ask two questions about the construct validity of the study--one about the IV and one about the DV.

h) Ask a question about the statistical validity of the study. To what populations might we be able to generalize? 

 

What about the recipients? It turns out that people aren't really sure if they should send their gratitude letters. They might feel awkward about it, perhaps surmising that the recipient may not really care about such a thing anyway. But here's a quote about research on receiving these letters: 

On the receiving end, opening a gratitude letter feels even better than you might imagine. Amit Kumar, a social scientist at the University of Texas at Austin, studies the reactions of gratitude-letter recipients.

“It’s not like it makes some people feel great and some people feel just OK and some people feel kind of weird,” he said of his research findings. “Almost everybody is saying that they feel really, really great.”

i) Most psychology studies have at least two variables (because they are testing association or causal claims). But, at least according to the journalist's description above, the study by Dr. Kumar had only one variable. What was it? 

j When Dr. Kumar said “Almost everybody is saying that they feel really, really great.”, he was saying something about the mean of this variable as well as its standard error. What can you infer about these two statistics, based on his quote? 

k) This study can support a frequency claim, because it's about the level of one variable. In your own words, write the frequency claim it can support. 

(Dr. Kumar's study on gratitude letter recipients can be previewed here. You'll see that in addition to asking recipients how they felt after receiving a gratitude letter, they also asked senders to predict how the recipients would feel. The main finding was that recipients felt much better than senders predicted they would feel.)

Make a face

Should you smile, grimace, or neither when you get your COVID jab? Photo credit: Gregory Johnston/Shutterstock

Have you gotten a COVID vaccination yet? If not,  grab a chopstick and read this article before you go. Here's a link to a journalist's summary of a study on the effect of making a face (such as a smile or a grimace) on the pain of an injection. 

The background for the research is the "facial feedback hypothesis:" If you've read Chapter 14 already, you'll recognize this line of work. 

The idea that manipulating our facial expressions can affect our emotions has a long and storied history. There are many advocates of [this] hypothesis, and many critics, too. Indeed, one of the classic findings in the field — that people find cartoons funnier if they hold a pen between their teeth, inducing a smile — recently failed to replicate. This mixed research background was well known to Sarah D. Pressman and Amanda M. Acevedo at the University of California, Irvine, who led the new work, published in Emotion.

Here is some detail about the study: 

The team randomised 231 student participants to hold either a neutral expression, a regular smile (which involves the cheek muscles), a Duchenne smile (which also recruits muscles by the eyes, resulting in creasing around the eyes) or a grimace (which recruits both these sets of muscles, plus those which allow us to wrinkle our brows.

In the study they manipulated the facial expressions unobtrusively, by asking participants to hold a chopstick in different ways:

To induce these expressions, the participants (who thought they were taking part in an investigation of multi-tasking) were given chopsticks to hold in one of four positions between their teeth, and asked to imitate an example photo. Each participant then held their assigned facial position while they received an injection of saline solution into the upper arm, designed to mimic a flu jab. Their heart rates and skin conductance were measured throughout [before, during, and after the jab], as objective markers of their stress levels, and they also reported on levels of pain (anticipated and then actual) and other emotions, such as being jittery or relaxed.

This study was an experiment. Before we get to the results, let's identify some of its features.

a) What was the independent variable (IV) here? How many levels in this IV, and what were they? 

b) Was the IV manipulated as independent groups or within groups? 

c) The journalist writes, "The team randomised 231 student participants to hold either....". Is the word "randomized" here referring to random assignment or random sampling? 

d) There are several dependent variables (DVs) reported here. Name three of them. 

e) Which type of experiment was this? posttest only? pretest-posttest? repeated measures? or concurrent measures?

Now for the results. The researchers collected multiple DVs and the patterns were not the same across all DVs. I will summarize the results for only one DV, self-rated pain:

The Duchenne smiling and grimacing groups gave lower pain ratings than the neutral group while waiting for the injection, during the injection, and immediately after it. The regular smiling group fell somewhere in between.

f) Sketch a graph of this result (placing the DV on the y-axis, as usual).

g) Now a big challenge: Apply Chapter 11's twelve internal validity threats to this example. How does the study's design help rule these threats out? I've done a few of these for you. You'll find Table 11.1 helpful here!

Design confound
Selection effect
Order effect
Maturation	This was a pretest-posttest design that took place in a single session. The single session helps rule out maturation. We can also rule out maturation because all four groups had the same opportunity to "mature."
History
Regression to the mean
Attrition
Testing	All groups were tested multiple times, which means this potential threat was held constant in all groups, ruling it out as an internal validity threat.
Instrumentation
Observer bias
Demand characteristic	The independent-groups design and cover story about multitasking can probably help disguise the true nature of the experiment.
Placebo effect