Selection Bias

1
Selection Bias

• By Peter Boothby

2
Selection Bias

• How should we select observations for inclusion
  in a study?
• The decision as to which observations to select
  is important for the outcome of the research and
  the degree to which it can produce reliable
  results.
• The most obvious example of selection bias is
  when we, knowing what we want to see as the
  outcome our research, select observations that
  support this desired conclusion.

3

• To avoid selection bias, selection should allow
  for the possibility of some variation on the
  dependent variable.
• For example, research that attempts to explain
  the outbreak of war with studies only of wars,
  fails to let the dependent variable vary.

4

• When observations are selected on the basis of a
  particular value of the dependent variable, we
  cannot learn anything about the causes of the
  dependent variable without taking into account
  other instances when the dependent variable takes
  on other values.

5

• We must be aware of the biases introduced by
  selection on the dependent variable and seek to
  correct for these biases.
• Avoid cases of extreme selection bias where there
  is no variation on the dependent variable
  because we cannot learn about causal effects.

6
Avoiding Selection Bias

• Selecting cases on the independent variable is
  less likely to bias results.
• Avoid bias by selecting on an explanatory
  variable that is irrelevant to the study (and has
  no effect on the dependent variable).