Causality and Controls

							Causality and Controls


        Laurin Ariail
     January 23, 2003
         POL 221
Causality
• The notion that A  B
• For example: tax rate  disposable personal
  income
• Can show causation if:
  – Changes in the independent variable precede
    predicted changes in the dependent variable.
  – Changes in the dependent variable are
    nonrandomly related to changes in the
    independent variable.
  – *most difficult* No other independent variable is
    responsible for changes in the dependent variable.
Research Design
• “plan of attack”
• Two methods:
  – Experimental Design
     • Two groups:
        – EXPERIMENTAL GROUP - group to which the change
          is made.
        – CONTROL GROUP - group identical to experimental
          except for variable being tested.
     • CONTROL GROUP - discounts the effects of other
       variables on the outcome.
Experimental design: an example
• An election in Kenya
• The predicted causal relationship is:
  Radio broadcasting  voter turnout and
                       election winner
• Control Group – group that hears no
  broadcasted advertisements.
• Experimental Group – group that hears the
  radio ads.
Method of Research design two:

  – Quasi-experimental design
    • Most common.
    • No control group can be established.
    • Simulates experimental design by collection
      of significant data and/or employing
      statistical methods.
Quasi-experimental design: an example
• Union Café prices
• The predicted causal relationship is:
  yearly income  attitude on prices
• Data:
  Survey students to obtain data and
  analyze to yield conclusion.
Conclusion
• Causality: A  B

• Two methods of determining causality:
  – Experimental design
    • Control v. Experimental groups
  – Quasi-experimental design
    • Extensive data collection