Skip to content
Find file
Fetching contributors…
Cannot retrieve contributors at this time
66 lines (63 sloc) 2.44 KB
Deception
* Two Forms:
* Active Deception - You lie
* Passive Deceoption - You just don't tell them the truth
Ethics & Deception
* Deception is lying. Lying is unethical/unacceptable
* If you are deceiving participants then how can they really give informed consent?
* May cause participants stress or anxiety for feeling foolish because they were tricked
Is Deception Really Bad?
* On average, college students do not think so…
* But, you need to address this question on a case-by-case basis
What can you do?
* Use naturally occuring behaviors in a natural occurring setting
* Role playing? (A.K.A. Simulation research
* Consent to Deception?
General Recommendations…
* It's okay to use deception. But, you should ask yourself some questions
* Is it necessary?
* How much do I need to deceive the participant?
* What kind of harm is likely if I decieve the participant?
* Can I inform them about it after I decieve them?
* Do the benefits outweigh the costs?
Debriefing
* One way to lessen the harm of deception is to use debriefing
* Occurs immediately at the end of the experiment
Components of a debriefing
* Educational
* Dehoaxing
* Desensitization
* Note: When participants have been deceived you have to explicitly asked them if they are okay with you using their data
Ethical Treatments of Animals
* YES
Reliability & Validity
Recall…
* What is a hypothetical construct?
* A term for something that isn't directly observable
* It must be defined as the operational construct or manifest variable
Reliability
* A measure of degree of consistency
Observed Data
* True Score
* Measurement error
Measurement Error
* It's what we measured but we didn't really want to
* Two kinds
* Random - Various things that can effect the measurement
* Systematic - Caused by actual study
Unreliable data is rather useless, since no meaning can be derived from the data.
Forms of Reliability
* Test-Retest Reliability - You can test again and get similar results
* Alternate Forms Reliability - Alternate forms of the test yeild similar results
* Inter-rater Reliability - Different raters all rate similarly
* Split-half Reliability - Internal consistency
Which form to choose?
* It depends
* On the construct your measuring
* On the measurement methodology
* One the length of timebetween measurements, if any
Scneitifcally, what's good reliability?
* Correlations go from -1 to +1
* Test-Retest: .5 (given a 3 month between time)
* Split-Half: .7
* Inter-Rater: .7
Something went wrong with that request. Please try again.