Chapter 9: Conducting experiment.
Straightforward vs. Staged manipulations.
3 types of dependent variables: self report, behavioral, physiological.
Sensitivity of dependent variable, floor vs. ceiling effects.
Ways to control participant and experimenter expectations.
Reasons for conducting pilot studies.
Advantage of having manipulation check.
Selecting research participants
Population: all individuals of interest to researcher.
The method used to select participants affect external validity (extent to generalize result to
other populations and settings).
If it is important to accurately describe the population, use probability sampling.
If want to test hypotheses, X causes Y, relationships, test predictions from theories: use non-
probability sampling like haphazard of convenience methods. Recognize that they affect the
ability to generalize your results to some larger population.
Sample size: larger = more likely the results can be statistically significant, cause larger
sample provide accurate estimates of population values.
Usually select typical size for studies in the area, but more formal way is the power analysis.
Manipulating independent variable
Construct operational definition, turn conceptual variable into set of operations – specific
instructions, events, stimuli. Manipulation is when changes the conditions the participants are
exposed to. Also the independent and dependent variables must be within context of expt setting,
this is called setting the stage.
-Supply enough info for participants to give informed consent (rationale of study).
-may want to use deception because participant may confirm hypothesis, or behave socially
acceptable. But then need to debrief them at conclusion of experiment.
-There are no clear-cut rules for setting stage, but setting must seem plausible to participant. Also,
no clear cut rules for translating conceptual variables into specific operations. How the variable is
manipulated depends on variable and cost (especially the ethics of procedures considered).
Types of manipulations (straightforward vs. staged)
Straightforward manipulations: Most.
difficulty of material to be learned, motivation levels, way Qs asked, characteristics.
manipulate X simply with written, verbal, visual instructions, and stimulus presentation verbally,
written, videotape, computer.
Ex: Sign and towel reuse: social norm sign had 44% success, friendly ones 35%.
Ex: Jury decision more harsh on high severity condition (both ovaries removed).
Ex: Memory research, more different words list recalled more easily. Being an expert did not help on
atypical script of the show. Benefit of being expert is limited.
Ex: teacher disclosure, related to rating of caring and trustworthy, X teacher competence. Staged manipulations / event manipulation: sometimes need to stage events to manipulate X
Frequently used for 2 reasons:
a) To create psychological state in participants: frustration, anger, lowering self esteem.
Ex: Sense of entitlement – unfair treatment of losing group behaved more selfishly.
b) To simulate some situation that occurs in real world.
Ex: Obedience experiment; multitask experiment;
Frequently employ a confederate (accomplice). Appears to be participant, but really is part of the
manipulation. Useful to create particular social situation.
Ex: participant model food consumed by confederate when hungry.
Ex: line judgment conformity experiment; public sneezing& perception of unrelated risk.
demand great deal of ingenuity and acting ability. Used to involve the participants in ongoing social
situation that the individuals perceive not as an experiment but as real experience. Assume result to be
natural behavior that reflects feelings and intentions of participants. BUT: great deal of subtle
interpersonal communication hard to interpret and replicate. Because many things happened, it is hard to
tell which ―ONE‖ caused the result. Hence, straight-forward manipulation is easier to manipulate, but
sometimes the nature of variable demands complicated procedures.
Strength of manipulation
Simplest design has 2 level of X, choose the levels that will give the strongest manipulation.
Strong manipulation maximizes the differences between the 2 groups and increases the chances
that the independent variable will have a statistically significant effect on dependent Y.
Ex: X= similarity Y= liking. Level 1 on X = least similarity. Level 10 = greatest similarity.
To get strongest manipulation, one group would encounter confederate of level 1, other group
with level 10. This result in greatest difference in liking means (9 point difference). Weaker
would use level 4 and 7. Strong manipulation is important in early stage of research, need to demonstrate a relationship
Subsequent research can manipulate other levels of X to get detailed picture.
The principle of using the strongest manipulation possible should be tempered by 2
considerations: a) External validity – strongest possible manipulation may entail a situation rarely
occurs in real world.
b) Ethics: manipulation should be as strong as possible within bounds of ethics.
Ex: fear or anxiety manipulation should be within physical and psy harm.
Cost of the manipulation.
Straightforward manipulation is cheaper.
Running manipulation individually is more costly.
Public support from government and private agencies is important.
Measuring the dependent variable
Y is usually one of 3 general types: self report; behavioral; or physiological.
Self-report measures: measure attitudes, liking, judgments, intended behaviors, emotional states,
attributions, confidence in judgment, and other aspects of human thought and behavior.
-Rating scales with descriptive anchors (endpoints) are commonly used.
Behavioral measures: direct observations of behaviors. Like self-reports, almost endless
measurement of behaviors are possible. May record whether a behavior occurs. Choose to record
the rate of a behavior (# times occurs in given time period); reaction time (how quick a response
occurs after a stimulus); duration (how long a behavior lasts).
Ex: couples separating vs not separating in air port. Crying coded as sadness, holding hands
coded as contact seeking.
Physiological measures: record responses of body.
- Galvanic skin response(GSR) – general emotional arousal and anxiety. Electrical conductance
of skin, that changes when sweating occurs.
-Electromyogram (EMG): muscle tension, used as measure of tension or stress.
-Electroencephalogram (EEG): electrical activity of brain cells, record general brain arousal as
response to difrerent situations (activity in parts of brain as learning occurs, or sleep stages).
-others: temperature, Hrate, blood, urine. Magnetic resonance imaging (MRI)- in behavioral
neuroscience. Provide image of brain structure. Allow to compare brain structure of ppl with a
condition with normal/ without the condition.
-fMRI: scan areas of brain while performs physical or cog task. Provide evidence for brain
processes involved. Ex: adults with higher levels of education do better on memory tasks, and
used areas of frontal cortex not used by other individuals.
Multiple measures A variable can be measured in a variety of concrete ways (operational definitions).
If X has same effect on several measures of same Y, confidence in results increased.
It is also useful to know if X affects some measures but not others. Ex: X designed to affect liking
might have an effect on some measures of liking (want to work with) but not other (want to date).
Effects of X on several different behaviors.
When more than one Y measure, the question of order arises
-responding to first measure affect responses on later measure; participants attend more closely to
2 ways of responding to this issue: if problem is serious – counterbalance.
-If not serious, present the most important measures first, so order won’t be a problem in
interpreting results on the most important Y, but may be for other. But overall impact minimized.
May be necessary to conduct separate series of experiments to explore the effects of an
independent variable on various behaviors.
Sensitivity of the dependent variable.
Y should be sensitive enough to detect differences between groups.
Ex: like the person (Y/N vs. scale of 5-7). Second one being better to detect difference of liking.
Sensitivity is important when measuring human performance.
memory—recall, recognition, reaction time.
cognitive task—speed/# of errors in proofreading task.
Physical performance—motor task.
*Such tasks vary in difficulty. Some easy so everyone do well regardless of conditions ceiling.
Ceiling effect: X appears to have no effect on Y because participants quickly reach maximum
Floor effect: task is so difficult that hardly anyone can perform well.
Ex: crowding has no effect on performance. The task was too easy, when increased in
complexity, crowding did result in lower performance.
Cost of measures
Paper-pencil self report measures are cheap.
If required trained observers, elaborate equipment, c