The Scientific Method

1
The Scientific Method
A Theory an explanation of why an event or
outcome occurs it identifies the underlying
causes of an event. A Hypothesis A specific
prediction about what should occur if a theory is
valid. It provides the means by which a theory
could be tested.
If our theory is true, then our hypothesis should
be true. Test the hypothesis.
If the hypothesis turns out to be true, then our
theory gains support. However, that doesnt
mean the theory is necessarily true. There could
be some other reason why the hypothesis is true.
If the hypothesis turns out to be false, then
(more or less) that means that our theory is
false or at least needs to be changed somewhat.
2
The Scientific Method.

• Theory It is cold out.
• Hypothesis People will wear long pants. (If the
  theory is true, then the hypothesis will be true).

People are wearing long pants. That makes us
think we are more likely to be right that it is
cold. But, is that the only reason people might
be wearing long pants??? What to do?
No one is wearing long pants. Conclusion It is
probably not cold out. Theory is disproven.
3
The Scientific Method.

• Theory People blame their failures on external
  factors because they dont want other people to
  think they are losers. (Self-Presentation)
• Hypothesis People will only do this if they
  think others will find out. (If the theory is
  true, then the hypothesis will be true).

People dont blame failure on external factors
when they think no one will know. That makes us
more confident that self-presentation is
right. But, is that the only reason people might
not blame external factors in private?
People DO blame external factors, even in
private. Conclusion Self-presentation doesnt
explain why people blame external factors for
their failures. Theory is disproven.
4
Research Methods An Outline

• Qualitative Methods
• Quantitative Methods

Surveys
Correlational Methods
Archival Research
Observational Studies
Between Subjects
Experimental Methods
Within Subjects
5
Quantitiative versus Qualitative Methods

• Qualitative Methods Using observation,
  interviews, etc to develop detailed, descriptive
  information about the topic being studied.
• Quantitative Methods Collecting data using
  operational definitions of concepts that provide
  numeric measures of the concepts.

6
Qualitative Methods

• Strengths
• In-depth, detailed description of area of study.
• Can look at how the research participants see the
  situation.
• Can provide information that the researcher
  wasnt looking for.

Weaknesses Doesnt allow for easy comparison
between people or situations. Doesnt lend itself
well to statistical analysis. Greater potential
for researcher biases to affect the
interpretation of the data.
7
Quantitative Methods Collecting data using
operational definitions of concepts that provide
numeric measures of the concepts.
8
Operational and Conceptual Definitions

• CONCEPTUAL DEFINITION Like a dictionary
  definition of a concept. What we think we are
  talking about.
• OPERATIONAL DEFINITION A specific, observable
  response that is used to measure a concept.
  This is something we are actually using in
  research.

9
Operational and Conceptual Definitions Examples
Concept Conceptual Definition Operational Definition
Love A feeling of intense positive emotion toward another individual. A desire to be with and protect that person Number of times someone glances toward their loved one.
Aggression Behavior that is intended to injure someone physically or psychologically. Intensity of shock someone (supposedly) gives to a partner.
Self-Esteem The degree to which someone feels that they are worthwhile. Answers to questions on a well-designed self-esteem questionnaire.
10
Quantitative Methods

• Strengths
• Provides objective measures of concepts.
• Allows for direct comparison 4 is always less
  than 5.
• Data can be analyzed statistically.
• Is usually easier, quicker, cheaper.

Weaknesses Only find out what you think to
ask. Can your concepts really be boiled down just
to numbers. Forces the researchers perspective
on the participants.
11
Correlational Research versus Experimental
Research

• Correlational Two or more variables are
  measured. We look at the relationship(s)
  between/among the variables. (e.g., Which
  variables are correlated with each other?).
• Experimental Manipulate one or more variables
  (the independent variable(s)), then measure one
  or more other variables (the dependent
  variable(s)). See if the DV is different at
  different levels of the IV.

12
Correlational Research

• Any time we do a study where we measure all of
  the variables, that is correlational research.
  If the researcher does not control the level of
  at least one variable for participants, then it
  is correlational research.

13
Examples of Correlational Research

• As a test of whether people make internal
  attributions for success, and external
  attributions for failure, ask people what they
  got on their most recent test, and ask them to
  rate how much their grade depended on something
  about themselves versus something external to
  themselves.

14
Examples of Correlational Research

• To test a theory that says that intelligence
  comes from the amount of intellectual stimulation
  one gets as a child Observe a childs
  environment, measuring the amount of intellectual
  stimulation. Then, give an IQ test. See if they
  correlate.

15
Examples of Correlational Research

• To test the evolutionary theory of parental
  investment, ask men and women how important
  physical attraction is in their choice of a mate.
  Also, ask them how important wealth, status, or
  earning potential is.

16
Correlation Coefficients.

• Correlational studies often (but not always)
  involve the use of a correlation coefficient.
• A correlation coefficient, usually symbolized by
  r, is a statistic that tells us how much two
  variables are linearly related.
• In other words, the higher r is, the more a graph
  of the two variables looks like a straight line.

17
The Correlation Coefficient
r 1.0
18
The Correlation Coefficient
r -1.0
19
The Correlation Coefficient
r 0.51
20
The Correlation Coefficient
r -0.51
21
The Correlation Coefficient
r -0.03
22
The Correlation Coefficient

• We should remember that a zero correlation does
  not mean that there is no relationship between
  our two variables. It means that there is no
  linear relationship. In other words, it means
  that the graph doesnt look anything like a
  straight line.

23
The Correlation Coefficient
r -0.00
24

• In social psychology, correlational data can be
  collected by
• Surveys
• Archival Research
• Observational Research

25
Strengths of Correlational Research

• Usually fairly cheap and easy to do.
• You can measure things realistically (especially
  with observational or archival research).
• Some things can not be manipulated (controlled by
  the researcher), so correlation is your only
  choice.

26
Weaknesses of Correlational Research

• The big weakness of correlational research is
  that you cannot tell how your variables are
  causally related by correlational research.
• Correlation does not equal causation.

27
Correlation and Causation.

• If two variables are correlated, then one of
  three possible causal relationships exists

1. Variable A is the cause and Variable B is the
effect. 2. Variable B is the cause and Variable A
is the effect. 3. Some third, extraneous variable
(C) is the cause of changes in both A and B, but
A and B are not directly related.
28
Correlation and Causation

• For example, suppose we do a study and find that
  the amount of time people have been dating is
  positively correlated with how much they want to
  get married.
• What are the possible causal relationships?

29
Correlation and Causation

• We know that one of the three patterns is true (A
  causes B B causes A or C causes A and B). But,
  we dont know which one is true. Therefore, we
  cannot conclude anything about the causal
  relationship between the variables.

30
The Experiment

• An experiment is the method most commonly used by
  social psychologists. Social psychologists like
  it because it lets us tell what are causes and
  what are effects.
• With correlations, a big problem is that some
  extraneous variable could be causing the
  relationship between the two things that
  correlate. With an experiment, we control the
  extraneous variables so that we know that isnt a
  problem.

31
The Independent VariableThe Hypothesized Cause

• In an experiment, we have three kinds of
  variables that we are concerned with
• Independent Variables (IV) These are the
  variables that we think are the cause of
  something. We manipulate the independent
  variable. In other words, this is the variable
  that we control what level everyone in the
  experiment has.

32
The Dependent VariableThe Hypothesized Effect

• Dependent Variables (DV) These are the variables
  that we think are the effect. We believe that
  differences in the independent variable cause
  changes in the dependent variable. We measure
  the dependent variable.

33
Extraneous VariablesPotential Sources of Error

• Extraneous Variables These are the variables
  that might cause a relationship between our
  independent and dependent variables. We want to
  control these so we know that IF we find a
  relationship between the IV and the DV, we know
  its because the IV is causing a change in the DV
  because all of the extraneous variables are
  controlled.

34
How to Control Extraneous Variables
The goal is to make all of the procedures the
same across the different conditions of the
study, except the independent variable. If the
only thing different between two situations is
the independent variable, then that HAS to be the
cause of any difference in how people act.
35
An Example

• Suppose I have the hypothesis that people will be
  more persuaded by an argument made by an expert,
  than an argument made by a non-expert.
• To test this, I want people to read a newspaper
  article about term limits for politicians. Some
  people will read the article as written by an
  expert political scientist, and other people will
  read an article written by someone without any
  particular expertise in politics.

36
Persuasion by Experts

• What is my independent variable?
• What should my dependent variable be?
• What are some possible extraneous variables that
  I should be worried about?

37
Controlling Extraneous Variables

• How can we make sure that everything is the same
  in all the conditions of the experiment except
  for the independent variable (level of
  expertise)?
• Standardized Procedures Make sure that you are
  following a fixed set of procedures for everyone
  in the study (except, of course, the IV).
• Same Experimenter.
• Say the same things.
• Same stimulus materials.

38
Controlling the Extraneous Variables.

• But what about things like the fact that people
  have different personalities, different initial
  positions, etc?
• One Solution The within-subjects experiment.
• Problem Practice effects.

39
Controlling the Extraneous Variables.

• Another solution Between subjects experiment

Use random assignment to balance out individual
differences. By deciding who is in each
condition at random (flip of a coin, for example)
we are likely to find that the individual
characteristics of people are, on average, equal
in the two conditions.
40
Strengths and Weaknesses of Experiments

• Strength Causality!!!
• Weakness Artificiality.

41
Research Methods An Outline

• Qualitative Methods
• Quantitative Methods

Surveys
Correlational Methods
Archival Research
Observational Studies
Between Subjects
Experimental Methods
Within Subjects
42
Some Other Issues in Research

• Generalizability, Representativness and Random
  Sampling.

43
Some Other Issues in Research

• Internal Validity The extent to which an
  experiment can really tell about cause and
  effect. How well you have made everything the
  same except the independent variable.
• External Validity The extent to which research
  results really tell us about the real world, and
  not just about the particular situation we
  studied.

44
Some Other Issues in Research

• One way to increase external validity is to
  increase mundane realism the extent to which
  the studys setting looks and feels like the
  outside world.
• You also need to worry about experimental
  realism the extent to which the studys setting
  feels realistic and involving to participants

45
Issues in Ethics for Social Psychological Research

• The Risk/Benefit Tradeoff
• Risks
• Emotional
• Physical
• Benefits
• For Society
• For the Individual

46
Issues in Ethics for Social Psychological Research

• Informed Consent.
• Deception.