Systematic Measurement Error

1
Systematic Measurement Error

• Alaa Odeh
• February 19, 2003

2
Role in Descriptive Inference

• Descriptive Inference is the process of
  understanding an unobserved phenomenon on the
  basis of a set of observations.
• One of the fundamental goals of inference is to
  distinguish the systematic component from the
  nonsystematic component of the phenomena we
  study (King 56)

3
What is Systematic Measurement Error?

• In a perfect relationship, (A) the independent
  variable (s) produce the same dependent variable
  (B). 
• The dependent variable stays the same if the
  independent variables stay the same. However
  there is always a variation in the measurement of
  the dependent variable.
• It is necessary to try to avoid any systematic
  bias (inherent in the system used) in the
  measuring the dependent variable.

4
Systematic vs. Nonsystematic

• Systematic differences are caused by some
  underlying factor, which impact one group but not
  the other.
• Usually independent variables dont capture all
  the real causes. The factor not captured by the
  equation is an error.
• Usually fundamental and predictable
  characteristics
• Constantly affects observations
• Nonsystematic differences are the factors due to
  chance.

5
Examples of Systematic DifferencesVoter Turnout

• Lets assume that voter turnout was measured.
  More of the middle-class to upper-class citizens
  showed up as opposed to the poor citizens.
• Non-systematic difference bad weather
• Systematic Difference Ideology (cant agree with
  a candidate), income difference, political
  involvement, or transportation.

6
Palestine/Israel Conflict

• If study measured the level of conflict in
  Israeli-occupied West Bank communities
• There is no way to conduct a perfect study,
  because conclusions will be affected by any
  chance event that will occur (nonsystematic).
• Systematic differences deep cultural
  differences, mutual knowledge of each other,
  geographic patterns of residential housing
  segregation. These differences will continually
  affect results no matter what week the study is
  conducted.

7
In Conclusion

• It is necessary to watch out systematic and
  nonsystematic measurement errors in descriptive
  inference.
• In a causal relationship, the independent
  variable will affect the dependent variable in
  the same way, unless systematic bias appears.
• Independent variables at times do not capture all
  the real factors. Systematic measurement
  errors are underlying factors impacting a certain
  group and constantly affecting the results.