Part Three

1
Part Three

• Modes of Observation

2
Chapter 8

• Experiments

3
Chapter Outline

• Introduction
• Topics Appropriate to Experiments
• The Classical Experiment
• Selecting Subjects

4
Chapter Outline

• Variations on Experimental Designs
• An Illustration of Experimentation
• Web-Based Experiments
• "Natural" Experiments
• Strengths and Weaknesses of the Experimental
  Method

5
INTRODUCTION

• EXPERIMENTS The design most linked to structured
  science.
• EXPERIMENTS Involve 3 things
• 1. group of selected subjects
• 2. do something to them
• 3. observe the effect
• Hence we are getting at CAUSATION.

6
Topics Appropriate to Experiments

• Projects with limited and well-defined concepts.
• Projects that are explanatory rather than
  descriptive.
• Studies of small group interaction.

7
Components of Experiments

• Three Pairs
• Independent and dependent variables
• Pretesting and posttesting
• Experimental and control groups

8
INDEPENDENT DEPENDENT VARIABLES

• Experiments examine the effect of the i.v. on the
  d.v.
• I.V. has 2 attributespresent (give) or not
  present (dont give)
• Measure what happens to d.v. if i.v. is present
  or not.
• E.G. 1 Group watch film Other group does
  notD.V. prejudice among subjects.

9
PRE-TESTING POST-TESTING

• SIMPLEST DESIGNPre-testing occurs firstsubjects
  are measured on D.V.
• Then, subjects are exposed to the treatment
  (film)
• Post-testing measures the effect of the I.V.
  (What would a lower level of prejudice tell us?)

10
Experimental andControl Groups

• Must be as similar as possible.
• Control group represents what the experimental
  group would have been like had it not been
  exposed to the stimulus.
• In the film example---if prejudice is reduced in
  both groups (but more in one than the other) we
  know film had an effect. (E.G., Hawthorne Effect)

11
DOUBLE-BLIND EXPERIMENT

• DBEneither subjects nor experimenters know which
  is the exp. group vs. cont. group
• E.G., Medical-Drug Studyresearchers would not be
  told which subjects were receiving the drugs
• D.V. needs to be clearly operationalized and
  precisely defined.

12
Selecting Subjects

• Representative samplingstill needed (College
  Students often usedremember you cannot
  generalize to the public with your guys!)
• Probability samplingas a rule you need at least
  100 in the sample to be representative.
  Experiments rarely involve that number. Dont use
  thisbut use its logic!)

13
EXPERIMENTAL SAMPLING

• Randomizationfirst the researcher selects a
  group of subjects from the populationmay be
  random may not.
• Then---you randomly assign subjects to control
  vs. experimental groups. (The more alike these
  two groups, the better)

14
EXPERIMENTAL MATCHING

• Another way to make the 2 groups
  comparablematching.
• We simply look at the population and match the
  control and experimental groups proportionately
  to this. (50 are Whitemake sure both groups
  have 50 whites)
• See Quota Matrixusing several matching
  variablesp.235. (COULD EVEN MATCH ON THE D.V)

15
Randomization and Matching (Why R is better)

• May not know which variables will be relevant for
  matching process.
• Most statistics used to analyze results assume
  randomization.
• Randomization only makes sense if you have a
  large pool of subjects. (If small numberuse
  matching)

16
Preexperimental Research Designs

• One-shot case study - single group of subjects is
  measured on a variable following experimental
  stimulus. (W/O Pre-testcant be sure of the
  effect)
• One-group pretest-posttest design - adds a
  pre-test for the group, but lacks a control
  group. (Does not rule out other variables making
  a contribution to the change)

17
Pre-Exp Continued.

• Static-group comparison - includes experimental
  and control group, but no pre-test. (How do we
  know the exp. Group wasnt less prejudiced to
  start with????)
• See Chartp. 237

18
Sources of Internal Invalidity

• Internal Validityperhaps the conclusions of the
  experiment may not reflect what actually went on.
• Historical events may occur during the course of
  the experiment. (Af-Am leader gets shot during
  study)
• Maturation of the subjects. (Subjects change
  during time on their own)

19
INTERNAL VALIDITYcont.

• Testing and retesting can influence behavior.
  (Subjects figure out we are looking for prejudice
  reduction)
• Instrumentation (We might use different measures
  of D.V.but what if they arent comparable)

20
INTERNAL VALIDITYcont.

• Statistical regression of subjects starting out
  in extreme positions. (Subjects low in prejudice
  might just naturally move upnot even based on
  the experiment.)
• Selection biases. (Groups must be comparable at
  the start.)
• Experimental mortality - subjects drop out of the
  study before it's completed.

21
INTERNAL VALIDITYcont.

• Demoralized control group subjects. (They may
  stop tryingno treatment)
• SEE PAGE 239if we conduct an experiment like
  this, we can usually tell what is going onwe
  have internal validity.

22
Limiting External Invalidity

• Lets say did experiment correctlywould it have
  the same affect in the theater? (Cant control
  sensitizing affect of the experiment itself.)
• Solomon four-group design see page 241
• Posttest-only control group designassumes
  subjects are randomly assigned and comparable in
  exp. And cont. groupings.

23
Solomon Four-group Design

• Four groups of subjects, assigned randomly
• Groups 1 and 2 are the control and experimental
  group.
• Group 3 does not have the pre-test.
• Group 4 is only posttested.

24
Posttest-only Control Group Design

• Includes Groups 3 and 4 of the Solomon design.
• With proper randomization, only these groups are
  needed to control problems of internal invalidity
  and the interaction between testing and stimulus.
• Explain Pygmalion Effecte.g., Spurters or
  Bloomers

25
"Natural" Experiments

• Important social scientific experiments occur
  outside controlled settings and in the course of
  normal social events.
• Raise validity issues because researcher must
  take things as they occur.

26
Web-based Experiments

• Increasingly, researchers are using the World
  Wide Web to conduct experiments.
• Because representative samples are not essential
  in most experiments, researchers use volunteers
  who respond to invitations online.

27
Experimental Method

• Strengths
• Isolation of the experimental variable over time.
• Experiments can be replicated several times using
  different groups of subjects.

28
Experimental Method

• Weaknesses
• Artificiality of laboratory setting.
• Social processes that occur in a lab might not
  occur in a more natural social setting.