Hypothesis Testing

1
Hypothesis Testing Validity

• October 4, 2004
• Sociology 487
• http//sociology.osu.edu/classes/soc487/sayer/

2
Voter Registration Deadline TODAY

• Visit http//www.osuvotes.com/blog.php for more
  information on registration and issues
• Ask Jill for voter registration card

3
Lecture Topics

• What are research hypotheses?
• How do we test hypotheses?
• Dependent and independent variables
• Assessing validity

4
Relationship between Theory Research
5
What is a Hypothesis?

• Explicit statement about the empirical
  relationship between two or more variables
• An Hypothesis is a predictive statement
• Test to see whether prediction is right or wrong

6
The Process
HYPOTHESIS Y F(X) y f(x) ? y f(x)
Theoretical expectation
Operationalization
Testable Hypothesis
Hypothesis Testing (Observation)
7
Variables Relationships

• Operationalization connects abstract concepts
  with concrete observations
• Variables are actual measures of concepts
• Two types of variables
• Dependent
• Independent

8
Variables and Attributes

• Variables are logical groupings of attributes
• Attributes are characteristics of persons or
  things
• Example
• Variable gender
• Attributes female and male

9
Relationship of Types of Variables

• Independent variables lead to variation in
  dependent variables
• Y F(X)
• Theory guides determination of dependent and
  independent variables

10
Stating Hypotheses

• Specify relationship between dependent variable
  and independent variable
• If this then that
• State the direction of association
• Positive
• Negative (inverse)
• Curvilinear

11
Qualitative Independent Variables

• Qualitative independent variables
• Gender
• Race / ethnicity
• Political party identification
• Sexual orientation
• Direction of association with dependent variables
  stated differently for qualitative and
  quantitative variables

12
Research vs. Null Hypothesis

• Research hypothesis is actual prediction we want
  to test
• Y F(X)
• Null hypothesis
• X does not lead to variation in Y
• Use probability to reject or fail to reject null
  hypothesis

13
Are our results an accurate reflection of reality?
Generalizability
Measurement Validity
Validity
Causal Validity
14
Causal Validity

• Truth of hypothesis that X causes Y
• Establishing causal validity is difficult
• Association between two variables is not
  sufficient for establishing a causal relationship
• correlation is not causation

15
Key Points

• Hypotheses are explicit statements about the
  relationship between variables
• Testing hypotheses involves examining whether
  independent variable causes variation in
  dependent variable
• Correlation does not prove causation