Transcript Sept 14 - University of Alberta

WebCT Now Available
You can Save to PowerPoint.
File – Print – Print HANDOUTS – Select the
Number of Slides per Page
Notice from Psych Dept:
Have you had a problem with WebCT?
We are collecting information to help improve WebCT and would
appreciate your help.
If you experience a problem, please send the following information
to Chris Goetz ([email protected]):
• Course Number and Section:
• Brief description of the problem:
• Date and time problem occurred:
• Steps taken to resolve the problem, including contacting the
Help Desk (492-9400):
• Was problem resolved?
• How long did it take to resolve the problem?
By the end of class today, you will
be able to:
• Define correlation and identify graphical
representations of correlations
• Discuss the pros and cons of experiments
• Understand statistical significance
• Explain the factors important to Ethics
Board reviews of research
Correlations
(another type of
descriptive statistics)
Are two variables related?
Illusory Correlation
(ex: Lincoln and Kennedy)
Finding associations – the surface features
appear to be similar
But, no real association exists
Can you think of additional illusory
correlations?
– “Bad things happen in threes”
“r”
• Strength and direction of a relation between two
variables can be assessed using the correlation
coefficient.
• Correlation represented by letter r
• r ranges between -1 and + 1 (depending on direction
of relationship)
• Negative r: as A increases, B decreases (or viceversa)
• Positive r: as A increases, B increases (or vice-versa)
•Under ideal conditions r = +/- 1
•BUT: Perfect correlation unheard of in psychology
•That would mean that each variable
exactly explains any variability in the
other. When A changes by one unit, so
does B.
•The closer to -1 or +1, the stronger the r
•What does it mean when the correlation equals 0?
•No relationship
Positively correlated
an increase in one variable is associated with an
increase in another variable.
Negatively correlated
an increase in one variable coincides with a
DECREASE in another variable
# of beers on Monday night relates to lower
grades on Tuesday’s midterm
No Correlation
• Can’t find a relationship b/w variables
• The graph will look random because
everyone’s going to have a different
pattern of scores for the two variables.
Curvilinear Correlation
What’s going on here?
Interpretation of Correlations
• The X variable might influence or cause the Y
variable.
• The Y variable might influence or cause the X
variable.
• Another variable (i.e., third variable, that was not
measured) influences both X and Y.
• X and Y don't really correlate at all, and you just
happened to observe such a strong correlation by
chance.
Correlation does not necessarily
equal causation
There is a positive relationship between ice cream
consumption and the number of drownings that occur
(Neale & Liebert, 1986).
There are three possible explanations for this positive
relationship. What are they?
•
•
•
Eating more ice cream causes more drownings
When drownings increase, people eat more ice cream.
Third variable. Something else responsible: summer.
Experimental
Testing
• In an experiment, you get to
control all of the variables.
• Experiments require you to rule
out alternative explanations.
Experiments
• Advantage
• Because of the high level of control,
causality can be established.
X causes change in Y
• Disadvantage
• May be artificial – can a lab
experiment really relate to the real
world?
VARIABLES
Everything controlled in an experiment.
Some variables left alone (but still
randomly assigned to be equal), some
are manipulated, some are measured.
You can have MORE THAN ONE
independent variable and MORE THAN
ONE dependent variable.
VARIABLES
• Independent Variable: The condition that is
varied in order assess its effect upon some
other variable (the dependent variable).
• = the thing that’s manipulated
• Dependent Variable: The variable that is
believed to be dependent on (affected by)
the independent variable.
• = the thing that’s measured
Ex: Can drugs increase IQ?
• How would you establish this?
• You need at least 2 groups
– Control Group
– Experimental Group
What can you recall from Psych 104
about each of these groups?
Control Group
• IQ tested before intervention (why?)
– Establish a baseline
• This group receives a placebo
– (ex: sugar pill)
•
Inert substance, or an activity that should
not cause any change from baseline to testing
• IQ tested after receiving placebo
Experimental Group
IQ test prior to administering drug
(baseline)
This group then gets the drug (or
experimental manipulation)
IQ tested again
Note:
• Independent Variable: Drug/No Drug
• Dependent Variable: Measure of IQ
Statistical Analysis (ex: t-test)
• Are there differences between groups?
• Are there differences before testing?
if yes, sample may not be random
THERE SHOULD NOT BE ANY
DIFFERENCES BEFORE TESTING, IF
YOU WANT TO RELY ON YOUR
RESULTS.
• Are there differences after testing?
(indicates that the drug affects IQ)
140
120
100
80
Time 1
Time 2
60
40
20
0
Experimental IQ
Control IQ
Are the results significant?
• “Significant”
– Means statistically meaningful
– May not be clinically meaningful
– Usually refers to a less than 5% likelihood that
the findings occurred by random chance.
Quasi-Experiments
• Many important variables are controlled, but not
everything is under the researcher’s control. EX:
Gender, Age  you cannot randomly assign
either.
• Common in Psychology.
• Cannot determine causality (like experiments
can), but can still determine relationships and
make inferences about causality.
Brief Review
• Which correlation is the strongest:
.13 (b) .35 (c) -.47 (d) 0
 -.47
• What is an advantage of the experimental
design?
•
Establish causality
• What is a placebo?
•
Sugar pill: not going to cause any change in
whatever you’re measuring
• What does a t-test assess?
•
Whether there are differences b/w groups
What does it mean to be
ethical?
“do the right
thing”
Psychologists are committed to increasing scientific
and professional knowledge of behaviour and
people’s understanding of themselves and others
and to the use of such knowledge to improve the
condition of individuals, organizations, and society.
Psychologists respect and protect civil and human
rights and the central importance of freedom of
inquiry and expression in research, teaching, and
publication. They strive to help the public in
developing informed judgments and choices
concerning human behaviour.
-Ethical Principles of Psychologists and Code of
Conduct, American Psychological Association (APA,
2002).
What are Research Ethics?
• The principles by which researchers decide the best
way to balance the contribution of their research to
human knowledge against potential damage to
human welfare (Haslam & McGarty, 2003)
• ethics are standards regarding the treatment of
others
• these standards apply to both research and clinical
practice
The Nuremberg Code
•
•
•
•
1947.
First international code of ethics
WW2 medical abuses
First attempt to provide standards for
evaluating clinical research (ex: weigh benefits
and costs before deciding the research should
be conducted).
Government Policy
• In Canada we have the Tri-Council Policy
Statement: Ethical Conduct for Research
with Humans
• Three federal funding agencies follow the
same guidelines: SSHRC, CIHR, NSERC
University Research Ethics Boards
(REBs)
• Different boards for different faculties/
departments
• In Canada, all fall under Tri-Council policy
• If one researcher on campus violates the
Tri-Council policy, federal funding to ALL
researchers can be revoked
Major Issues to Consider
•
•
•
•
•
•
Informed consent
Voluntariness and Inducements
Right to privacy
Discomfort and harm
Deception
Debriefing
Informed Consent Procedures
• Free and informed consent lies at the heart of
ethical research involving human subjects. It
encompasses a process that begins with the initial
contact and carries through to the end of the
involvement of research subjects in the project. As
used in this Policy, the process of free and informed
consent refers to the dialogue, information sharing
and general process through which prospective
subjects choose to participate in research involving
themselves.
(Tri-Council Policy Statement Section 2)
Participants Must be Informed
About:
•
•
•
•
Purpose of the research
Expected duration
Procedures
Right to decline to participate and/or
withdraw at any point
• Continuing and meaningful opportunities for
deciding whether or not to continue to
participate
…
Participants Must be Informed
About (II):
• Consequences of withdrawing
• Incentives for participation
• Potential risks, discomfort, and/or adverse
effects
• Limits of confidentiality
• Research benefits
• How results will be disseminated
• Contact information
Voluntary Participation and
Inducements
• Consent must be freely given
• Undue influence includes inducement,
deprivation, exercise of control or authority
over prospective subjects
• Noncoercive compensation is allowed
– What do this imply to you?
– Compensation for effort, time, and
inconvenience
Participants’ Right to Privacy
• Dignity and autonomy of human subjects is the
ethical basis of respect for the privacy of research
subjects. Privacy is a fundamental value, perceived
by many as essential for the protection and
promotion of human dignity. Hence, the access,
control and dissemination of personal information
are essential to ethical research. (Tri-Council Policy
Statement Section 3)
Confidentiality Issues to
Consider
•
•
•
•
Who has access to the data?
Who will be granted access if requested?
How long will data be kept?
How will confidentiality be safeguarded?
– Passwords for computer files, assigning
meaningless ID numbers, anonymity, consent
forms locked in filing cabinets, etc.
Discomfort and harm
• Always aim to minimize discomfort and
harm
• Strive to ensure that stress does not
persist outside the experimental session
• If researchers become aware that
procedures are causing harm take steps to
minimize that harm
Deception
• Withholding information prevents fully
informed consent to be given
• Use of deception must be justified by the
study’s significant value (nondeceptive
alternatives are not possible)
• Do not deceive if physical pain or severe
emotional distress is expected
• Deception is explained to participants as
soon as possible
Debriefing
• Provide participants with information about
the nature, results, and conclusions of the
research
• Correct any misconceptions
Tutorials
http://www.mcmaster.ca/ors/ethics/tutorial/
codes_tutorials.htm