Studied by 11 people
Animal research (replacement, refinement, reduction)
Replacement: Find alternatives to animals when possible
Refinement: Modify procedures to minimize animal distress
Reduction: fewest animal participants possible
APA ethics guidelines/standards
-Beneficence and nonmaleficence
-Justice
-Respect for people’s rights and dignity (Same three as Belmont Report)
-Fidelity and Responsibility (Establish relationship of trust, be accountable for professional activities: clinican, teacher, researcher)
-Integrity (Strive to be accurate, truthful, and honest)
-Institutional Review Board (IRB)
-Informed Consent
-Deception (Omission: withholding info and commission: actively lying)
-Debriefing
-Research misconduct (data fabrication and plagiarism)
-Animal research
Belmont Report
Provides a set of ethical guidelines for human subjects research
Three Core Principles:
-Principle of respect for persons
-Informed Consent
Principle of beneficence
-Asses potential harm to participants and benefits of study
* Principle of justice
- There must be balance between those whoo participate in research and those who benefit from it
- Sample should represent population (Diverse social backgrounds)
Deception in research
Ommission: Withholding information
Commission: actively lying
Informed consent
From Core Ethical Principles and APA Ethical standards
Is principle of respect for persons
Principle of beneficence
Part of Core Ethical Principles
Assess potential harm to participants and benefits of study.
Principle of justice
Part of Core Ethical Principles
There must be balance between those who participate in research and those who benefit from it. Sample should represent population (diverse social backgrounds)
Principle of respect for persons
Informed Consent
Content validity
Does the measure cover all/many aspects of the concept or theory?
* Better Example: test on mathematical, verbal, problem solving, and pattern recognition behaviors for intelligence
* Worse example: a test on only mathematical behavior for intelligence
Convergent validity
Does the measure correlate with other similar measures? It should.
* Example: Beck Depression Inventory (BDI) correlates well with Center for Epidemiologic Studies Depression Scale (CES-D)
Criterion validity
evaluates the statistical association of a measure with a relevant behavioral outcome (e.g., with correlation).
Cronbach's alpha
Internal reliability is often assessed with Cronbach’s alpha.
It is a combined measure of internall correlation and the number of items. Generally, we want a value of >(or equal) .80
(Don’t confuse it with alpha (a) from stats)
Difference between reliability and validity
Reliability is about consistency (does the measure correlate with itself?)
Validity is about accuracy (does the measure reflect the concept being studied?)
A measure may be reliable, but that doesn’t mean it is also valid.
A valid measure MUST be reliable.
All valid measures are reliable. Not all reliable measures are valid.
Discriminant (divergent) validity
Does the measure correlate with measures of different constructs? It should NOT (necessarily).
Example: BDI does NOT correlate as well with measures of physical health problems.
Face validity
Does the measure seem to make sense at face value?
Good Example: “head circumference” for how well a hat fits.
Bad Example: “head circumference” for intelligence
Internal reliability
Is a participant consistent with themselves?
Example: related self-report questions.
“How satisfied are you with your job?”
“To what extent do you experience job satisfaction?”
Interrater reliability
If multiple researchers are observing behavior, do they report it consistently?
Example: Two researchers observe a problem solving study. A child solves a puzzle in a certain number of steps. Do both researchers count the same number of steps?
Observational measures
Researcher logs behavioral data either on their own or through software.
Physiological measures
Biological data such as heart rate, sweat production, or brain activity.
Self-report measures
Participants answer prompts or questions about themselves.
Example: “How many alcoholic beverages do you consume per week?”
Test-retest reliability
Are scores consistent each time a measure is applied?
Example: If an IQ test is administered twice in one semester, are the scores consistent?
Acquiescence
Participants may be biased in favor of choosing the “yes” option more often than they should (or positive option like “strongly agree”).
Challenges of self-report surveys
Arguably the most important issue with self report surveys is that researchers must rely on participants to be accurate. Challenges include:
They could lie or embellish
They could be honest but still misunderstand the question or themselves
Self-reporting “more than they can know” about themselves
They could be biased about the survey
They could take shortcuts
Self-report survey “shortcuts” include response sets, acquiescence, and fence sitting.
Researchers can incorporate purposeful questions to identify when these shortcuts may be a problem.
-For example: “If I had my life to live over, I’d change almost nothing”
VS
“If I had my life to live over, I’d change almost everything.”
> the first question should correlate with positive items, but the second should not.
Participants can bias their responses in more ways than just shortcuts. Other issues include…
Social desirability - participants may respond in a way they believe paints them in a good light
“Faking Bad” - participants may respond the opposite way, inordinately emphasizing negative answers
Memory Flaws - participants may simply not remember their own behaviors accurately
Double-barreled question
The question is really two questions in disguise.
Example: “Do you agree that education is crucially important and that the core values of American Society are better taught in private schools than public?”
Fence sitting
For Likert scales or semantic differential questions, participants may inordinately favor the middle “safe” option (like “Neutral” or “I don’t know.”)
Forced-choice questions
Participants choose one of the available answers
Pros: Manageable to organize and code
Cons: Restrictive set of options
-Example: “Have you enjoyed this lecture? Yes or no?”
Likert scale
Participants are provided a prompt or a question and a scale of ordered responses to choose from
Pros: Allows for quantitative analysis (e.g, means)
Cons: relies on participants to introspect and report accurately (even when they are honest)
-Example: “I enjoy being in large crowds”
1 - Strong Agree
2 - Agree
3 - Neutral
4 - Disagree
5 - Strongly Disagree
Negatively-worded question
The question introduces double negatives that can confuse
Example: “People who do not drive with a suspended license should never be punished.”
1 - Strong Agree 2 - Agree 3 - Neutral 4 - Disagree 5 - Strongly Disagree
Masked design
One way that researchers have tried to resolve these biases is with a masked design. In these studies, observers are not researchers themselves, they are not informed of the aims of the study or the expected findings.
Without this knowledge, they cannot unwittingly move the study toward confirmation bias or influence participants toward the “desired” outcome. This does not prevent all biases, however.
Observer bias
Observational research is not immune to pitfalls of its own.
Sometimes observers may be biased to see what they expect to see (confirmation bias).
Observer effects on participants: Observers may influence participants to alter their behavior per expectations.
Open-ended questions
Participants answer however they like
Pros: Answers are rich and diverse
Cons: Can be very difficult to code and categorize responses (too many different answers)
Examples: “How do you feel about this lecture?”
Reactivity
Sometimes, the bias comes purely from the participant rather than from a researcher or a observer.
Reactivity refers to situations in which participant behavior is impacted by the mere presence of someone observing. Participants may be on best (or worst) behavior instead of truly typical behavior.
Solutions include..
Unobtrusive observations: one way mirror, one face in the crowd, etc.
Waiting until participants become comfortable to record observations.
Measure behavioral result instead of behavior itself.
Response sets
This occurs when participants response to numerous items the same way without necessarily thinking it through
Can be split into acquiescence and fence-sitting.
Semantic differential questions
Participants are provided a scale that splits two descriptors and indicate the degree of “closeness” to either side
Pros: Can provide participants with more clarity about how to answer
Cons: May restrict or bias participants answers
-Example: RateMyProfessor.Com (Professor’s overall quality)
Prof’s get F’s too: 1 2 3 4 5 A real gem
Note 
Flashcard (58)