InQuizitive Answers Module 2
Internal Validity
The relationship between two variables is genuine.
True/False
True: A frequency claim does not have to have an actual number. True: Not all claims presented in the media are based on research. False: Causal claims that use tentative language are actually association claims. False: Two frequency claims in the same data set can be combined into an association claim.
True/False
True: Scientists weigh pros and cons of research to determine which validity to emphasize.; In experiments, internal validity is often the first priority. False: A study must have maximum levels of all validities in order to be viable.; Researchers should try to maximize all types of validities in every study.
Researchers are investigating whether middle school children differ significantly in their critical-thinking ability by measuring the differences between sixth, seventh, and eighth graders. Each student at Tannenbaum Middle School receives the same intelligence test, and then the students' scores are compared. Researchers also record the age of the students to see if age affects performance on the test. Only participants whose first language is English are used.
Variable: intelligence test scores; student age; student grade level Constant: Middle School Location; primary language; type of test
Monique has created an experiment to test whether napping during the day improves productivity in the evening. She has one group of participants take a half-hour nap, one group take an hour nap, and one group take no nap. She then has all participants evaluate their productivity via a questionnaire and by writing down the number of things they have accomplished.
levels of the independent variable: half-hour nap, hour nap, no nap dependent variable: productivity independent variable: how much the participant sleeps levels of the dependent variable: productivity questionnaire
In a frequency claim, using a margin of error can help improve statistical validity.
margin of error; statistical validity
Construct Validity
A variable is operationalized well.
A researcher had two separate groups of participants drink either one glass of red wine or no red wine in order to see if drinking red wine could reduce stress on a self-reported stress measure.
Amount of red wine
Pain-tolerance
Amount of time hand is held under water
Manipulated Variable
Amount of x; x intake; Time spent doing x; Hours spent doing x;
A recent study investigated if watching Internet cat videos causes online viewers to have less stress. The study recruited a diverse sample of online participants and had them watch a video with kittens. The participants were then given a measure of stress. A valid statistical test showed that after watching the video participants had less stress.
Appropriate: Internal validity Not Appropriate: External; Statistical; Construct
Tara would like to measure how social college students are with friends at lunch compared to at dinner. She times how long students make eye contact in the dining hall with one another as a measure of their sociability. Tara finds that students at dinner are significantly more sociable based on this measure.
Construct Validity
Two frequency claims are presented together: "Ice cream sales are at a record high" and "Summer heat breaks numerous records." What claim can be safely made?
Each claim by itself is true, nothing else can be said
A recent newspaper headline reads that 1 in 25 teens in the United States attempt suicide. You read further about the study and find that researchers randomly surveyed a sample of teens from the Midwest via phone, and the study operationalized attempted suicide as "made a physical attempt to end one's life."
External Validity
In an experiment, scientists will often sacrifice external validity in exchange for better internal validity.
External; Internal
Kurt is conducting an experiment on how children respond to different types of math problems. He is giving some children problems with pictures and some children problems without pictures, and he will measure which group gets more problems right. For his research, Kurt wants to be certain that his manipulation is the cause of any group differences and that his statistical conclusions will be accurate.
Higher Priority: statistical validity; internal validity Lower Priority: Construct validity; external validity
JoAnn is interested in the availability of mental health care for patients admitted to the emergency room. She is limited in how much time she is allowed to spend with each patient, as many emergency room patients move in and out rapidly. She decides to use a shorter, less reliable measure of mental health care availability in favor of getting as many patients as possible, in order to represent people of many different backgrounds, illnesses, and injuries.
In this example, JoAnn is sacrificing construct validity to strengthen her external validity.
Timothy wants to measure the effects of amount of body spray on perceived attractiveness. He divides his participants into groups and applies no body spray, a small amount of body spray, or a large amount of body spray to each group. He then has members of the opposite sex rate the attractiveness of each participant on a 1-5 scale following a three-minute interaction.
In this example, the variable amount of body spray has three levels, whereas the variable of attractiveness has five levels.
Aaron is conducting an experiment in which he exposes separate groups of athletes recovering from grade 1 muscle strains to two different forms of physical therapy in order to see which has the shortest recovery time.
In this example, type of physical therapy is the independent variable. In this example, recovery time is the dependent variable. This example would likely result in a causal claim.
What type of validity is not prioritized with association or frequency claims but is prioritized with causal claims?
Internal validity
Causal Claims
MMR shot does not cause autism, large study says.; Aggressive video games lead to aggressive thoughts in players. High "normal" blood sugar could worsen brain function. Worry may make women's brains work overtime. Want a higher GPA? Go to private college. an experiment in psychological science
Absenteeism
Number of meetings missed per week
Persuasiveness
Number of strong arguments vs weak arguments
Happiness
Number of times someone smiles in a day
Education
Number of years in school
Measured Variable
Outdoor Air Temp; Age; Height; Ethinicity
In a causal claim, using random assignment can help improve internal validity.
Random assignment; internal validity
In any claim, using a representative sample can help improve external validity.
Representative sample; external validity
Determine whether or not the following qualities are required to make a causal claim.
Required: Two variables must be correlated. One variable must occur before the other. No other explanations exist. Not Required: Two variables are measured. The claim describes one variable.
Statistical Validity
Results are accurate and reasonable.
External Validity
Results generalize to the population discussed.
Association Claims
Shy people are better at reading facial expressions; Those with ADHD do one month less work a year. Facebook users are more likely to get worse grades in college. a correlational study conducted by Harvard University a correlational study done by you in your research methods class
Frequency Claims
Sixteen percent of teens have considered suicide.; Fourty-four percent of Americans struggle to stay happy. Seventy-nine percent of people think regret is acceptable. Most Americans do not exercise enough. the FBI's uniform crime report on the number of crimes in the United States television ratings released by the Nielsen Company that describe how much each show is watched
Jeremy is testing if certain levels of a new experimental drug he designed can improve immune functioning. He recruits participants for a clinical trial and randomly assigns them to either a treatment group, in which they receive the drug, or a control where they receive a placebo. After several weeks, he measures the differences between the groups' immune functioning via blood test. Jeremy announces that the drug group had better immune functioning after the trial and claims that his drug was effective. Though the immune functioning between groups was different, that difference was not statistically significant.
Statistical Validity
Zero Association in brocolli
Terrance should not make a prediction
Tony has recently started using social media to market his business, and he wants to compare that to more traditional types of marketing (TV and newspaper ads) to see which is more effective. In his traditional advertisements, he gives out the business's phone number. In his social media advertisements (on Facebook and Twitter), he tells customers to e-mail him. He then compares the total number of phone calls generated from the TV and newspaper ads to the total number of e-mail inquiries he receives from Facebook and Twitter ads.
Two levels of manipulated variable
Type I Error
Two variables are improperly claimed to be related. false positive MMR shots are said to cause autism, which is later disproven.
Type 2 Error
Two variables are improperly said to not be related. miss Climate change initially thought to be normal is shown to be caused by humans.
After conducting an experiment, Dennis concluded that eating chocolate each day leads people to feel guiltier about other things in their life. He used a random sample of adults from all over the country. Dennis measured guilt via a self-report questionnaire that had been previously validated by other researchers. Dennis's results showed that daily chocolate eaters do feel more guilty. However, because Dennis used a statistical test that was inappropriate for his data, the relationship between chocolate and guilt does not actually exist.
Type 1 error
Patrick conducted a study to see whether using social networking sites would lead college students to be more social in an in-person context. He randomly assigned college students to either use a social networking site before a group interaction or not. A week following the group interaction, he had each participant rate how social the other participants were. Patrick's results later showed no relationship between the two variables. However, this was because participants could not remember how social the other participants had been, and in actuality, there was a difference.
Type 2 Error
