Send the link below via email or IMCopy
Present to your audienceStart remote presentation
- Invited audience members will follow you as you navigate and present
- People invited to a presentation do not need a Prezi account
- This link expires 10 minutes after you close the presentation
- A maximum of 30 users can follow your presentation
- Learn more about this feature in our knowledge base article
Transcript of Research Methods
Form a Hypothesis
Analyze the Results
What's the Purpose?
Define Your Variables
The concepts or ideas being measured
Each variable will have multiple levels:
Choose a Sample
Choose a Method
A method of comparing two sets of data
The Scientific Method
Observe a phenomenon and record it.
Look for statistical relationships between variables.
Can't prove cause and effect, but can definitely provide clues
Can help identify risk factors
Goal is to determine cause and effect
Differs from other methods because researchers actually manipulate (control) one or more variables
How is the variable being defined for purposes of the present study?
Example: Defining "happiness"
Asking a person to rate his own happiness
Asking a person's friends to rate his happiness
Measuring how frequently a person smiles
What's a Hypothesis?
Can be based on "common sense," observations, and/or the results of prior research
Occasionally researchers will raise general questions rather than making specific predictions
A specific prediction
Studying one person in great depth
Can study people in unusual circumstances
People may lie, forget, misremember
Hard to generalize to larger population
Often quick and cheap
Can be anonymous
Includes written/web/oral surveys, questionnaires, tests, etc.
Data is only as good as the answers you get
Certain groups may systematically distort answers
Ratings of others are subject to
halo and horns effects
It's real behavior
You don't know much about the subjects
Works best if they don't know they're being observed
Can only observe behavior, not thoughts
Ethics limit this method to public behavior
The ONLY method that allows determination of cause and effect
May sometimes be too artificial
Some variables cannot be manipulated
The researcher manipulates one variable to see if it leads to any changes in a second variable
1. Identify the IV and DV
assign participants to groups
compare the DV for both groups
This is the one the researchers are manipulating
Chandon, P., & Wansink, B. (2007). The biasing health halos of fast-food restaurant health claims: Lower calorie estimates and higher side-dish consumption intentions. Journal of Consumer Research, 34, 301-314.
Which is healthier: McDonald’s or Subway?
The McSubway Study
Participants were given a coupon for either a McDonald’s Big Mac (600 calories) or a Subway 12-inch Italian BMT (900 calories)
Subway eaters were more likely to get bigger drinks, non-diet drinks, and cookies.
McDonald’s eaters underestimated their calorie consumption by 7%, Subway eaters by 52%.
Gets the new or different treatment
Gets the status quo
Goal is to avoid
between the two groups
We are looking to see if the groups'
scores are different
Are the Results Significant?
p value tells us the odds that the results are merely a fluke
Generally considered significant at p < .05
What are the implications of the findings? Do they really mean anything?
Watch Out For "Balanced" Reporting!
Compile the Results of Multiple Studies
Replication is key
Reporters will give equal coverage to both sides of "controversial" topics, even when there's no real controversy
Are the Results Reported Accurately?
Confusing Correlation With Causation?
Fox Business graphic showing what the wealthiest Americans in the top income tax bracket would pay if the Bush tax cuts were allowed to expire:
Same data, with a non-distorted Y axis:
Reliability and Validity
Reliability: Are the results consistent?
Validity: Does the test measure what it claims to measure?
Watching real-world behavior
Experimenter Expectancy Effects
Representative (unbiased) sample:
A sample in which there are no
between the sample and the larger population of interest
People must be told in advance what they getting into
Personal information cannot be revealed without permission
Freedom from Coercion
People cannot be forced to participate
Only when necessary, and when the benefits outweigh the costs
3. Avoid confounds
Make sure the groups are treated the same in every possible way, except for the IV.
Placebo and nocebo effects
We can (sometimes) eliminate these by keeping participants and/or researchers "blind"
In the case of placebo effects, we can never eliminate them. But we can account for them by having a control group, which gives us a baseline for comparison.
Positive correlation: As values of X go up, so do values of Y
Negative correlation: As values of X go up, values of Y go down
Correlations can range from -1.0 to +1.0
Farther from zero = stronger relationship
Sign (+ or -) indicates the direction of the relationship
Three ways to explain a correlation:
Correlation Does Not Necessarily Mean Causation!
Years in prison
1. A causes B
2. B causes A
3. There's a third variable that explains the relationship between A and B.
Measures of Central Tendency
The mathematical average
The middle score
The most frequent number
Sum of all scores
Total number of scores
1, 1, 1, 2, 3, 5, 8
All three of these can be called "averages"
Employee salaries (in thousands)
There's always a fourth possibility:
The correlation might simply be a coincidence.
Watch Out for Illusory Correlations!
Did I dream about Aunt Mildred?
Did Aunt Mildred die?
Black, brown, red, etc.
Older vs. younger
It only counts as a "variable" if there's variation within the study!
The measurement tool is not the same as the concept being measured.
If I give a survey about political beliefs, my variable is "political beliefs," not "a survey."
Variables should always be named using no more than a couple of words.
1. Identify the independent and dependent variables
This is the one that the researchers are manipulating.
The people who volunteer to be in a study may not represent the population
Researchers are people too! Biases may affect how they collect or interpret data.
Participants alter behavior based on what they think the researcher is looking for
Participants improve (or experience harm) due solely to expectations