Everyday Research Methods

Claim: Active sex life may lead to improved job satisfaction

Would it be possible to experimentally manipulate the variable "healthy sex life"? Photo: Helder Almeida/Shutterstock

Here are some headlines that were written about a study on job satisfaction:

    Study: Active sex life may lead to improved job satisfaction and engagement in work

    Maintaining an active sex life may lead to improved job satisfaction, engagement in work

 A healthy sex life may boost job satisfaction

You'll recognize that the verbs in these claims ("lead to" and "boost") are causal ones, and you should know by now that it takes an experiment to support causal claims. Before you read on, ask yourself:

    a) What are the two variables in the claim, a "healthy sex life boosts job satisfaction?"

    b) Would it be possible to conduct an experiment on this claim?

Now read this description of how the study was conducted (source): 

To understand the impact of sex on work, the researchers followed 159 married employees over the course of two weeks, asking them to complete two brief surveys each day. They found that employees who engaged in sex reported more positive moods the next day, and the elevated mood levels in the morning led to more sustained work engagement and job satisfaction throughout the workday.

    c)  Did the study manipulate a variable, as would be required for an experimental design?  (Oh, and did you notice the journalist's verb, "led to," when describing these results?)

Even though this wasn't an experiment, you might notice that it took place longitudinally. By testing how sex one one day was associated with mood and work engagement the following day, the study appears to have temporal precedence. 

    d) Re-read the study description above. The results are describing mediation. What are the three variables in the mediation diagram? Sketch a figure with three boxes (similar to the mediation diagrams in Chapter 9) that represents this mediation. 

Mediation or no, this is still a correlational study. The study supports an association claim (of the form, "Sex the night before was associated with better mood the following morning, and morning mood was associated with evening reports of job satisfaction"). But it cannot support a causal claim. 

 

You can read the peer-reviewed article here.  

 

 

How do you count sea turtles?

Is counting nests of eggs from sea turtle eggs a valid way to measure the population of sea turtles in the ocean? Photo: Kimberley Croy/Shutterstock

How do scientists estimate the number of sea turtles living in the world's oceans?  Counting creatures seems pretty easy--they can be over one meter long, after all--until you remember they live in the ocean, travel hundreds of miles, and are nearly impossible to track. 

That's why scientists have usually estimated the number of sea turtles in the oceans by counting something easier: The number of egg nests that female turtles lay. Here's how The Atlantic's Ed Yong tells the story:

A single female will lay her eggs at several places within the same nesting ground—a reproductive spread-bet that prevents her from losing an entire generation to, say, a storm or an industrious predator. Scientists have assumed that green turtles lay an average of 3.5 clutches each, and counting these clutches helps scientists estimate the global turtle population.

Is it valid to use the number of clutches to estimate the overall number of turtles? 

...after tagging green turtles in the Indian Ocean, Nicole Esteban from Swansea University has shown that each female lays around twice that number. And if that holds true across other nesting beaches, it means that we might have overestimated the population of this endangered and declining animal by a factor of two.

Scientists can canvas these nesting beaches and count the tracks of the females. If you divide that by the number of nests that each female makes, you get the total female population. For example, if you get 300 tracks, and you assume three nests per turtle, you get a total of 100 females. “But if you think the number of nests per individual is six, it’s a very different story,” says Esteban.

Yong's story describes an example of construct validity. Specifically, you could say that the operational definition being used to estimate the number of sea turtles has been found to have poor construct validity. You can apply some concepts from Chapter 5 to this example:

a) What conceptual variable is the focus of this story? How is the variable being operationally defined?

b) What is the scale of measurement of the "number of egg clutches" measure?  Categorical? Ordinal? Interval? or Ratio?

c) What kind of reliability would probably be most relevant for the "number of egg clutches" measure: Internal? Inter-rater? or Test-retest?

 According to the data collected by Dr. Esteban, we should update the operational definition currently used, and estimate the number of sea turtles using the factor of six nests per turtle instead.  But keep in mind that we need to establish the construct validity of the new data, too. Therefore we should also ask, How did Esteban find out that each female lays 6 nests (rather than 3)? 

She arrived at that answer not by counting tracks, but by following actual turtles. In October 2012, she and her colleagues patrolled the beaches of Diego Garcia Island, waited for the turtles to finish laying their eggs, and then accosted them. They carefully cleaned the shell and then stuck on a state-of-the-art satellite tag—a flattened, waterproof, Tupperware-like box, which they painted with black antifouling paint to stop marine microbes and larvae from growing...

..After tagging eight turtles, Esteban realized that they were laying far more nests than anyone had expected. So her team returned to Diego Garcia in July 2015, to tag ten more animals at the very start of the breeding season. And they confirmed that the females were laying an average of six clutches each, with a range of two to nine.

d) In Esteban's study, what is the conceptual variable of interest (Hint: it's not "the number of sea turtles in the ocean")? How was her conceptual variable operationally defined?

As a side note, this research provides an example of how scientists use pilot data (the first eight turtles) and then follow up by collecting more data. Yong also mentions that Esteban's finding, in which female sea turtles lay about six nests each, concurs with two other independent estimates of the same variable.