Causation

• One of the things that research does is explain relationships between things. A primary relationship is causal, meaning that one thing causes another.

• The three key things in proving causation are:
  
  • Temporal order – The cause must come before an effect
    
  • Association – There needs to be a pattern that shows that the two phenomena are even linked in some way. Statistical measures of correlation and comparisons between samples are ways researchers use to demonstrate association.
    
  • Eliminating alternatives – Even if there is a temporal order and an association between two phenomena there may be some other explanation for both, or a better explanation for the effect. By controlling for other variables and thus eliminating them as possible explanations, researchers strengthen their argument of a direct causal relationship between a cause and an effect. Another term for this is spuriousness which occurs when there appears to be a relationship between two things which is actually explained by an antecedent or an intervening variable. Even though you might find an association between two variables, you must be careful to look for factors which can explain both. For example, there might appear to be a relationship between the number of cars a family has and the number of baseball games they attend. Do the multiple methods of transportation cause people to go to more baseball games? Probably not. The relationship can be explained by an intervening factor, income, which can explain both.

• It is important to keep in ming common errors in logic when dealing with causal relationships:
  
  • Teleology is the error of using some sort of ultimate purpose or goal to explain a relationship. This would include saying that there is a relationship because it is "God's plan". This is not scientifically testable as there is no action from an individual. "God's plan" is too vague and distant to test.
    
  • Tautology is basically circular reasoning, meaning that a statement is true by definition. An example would be to say that all professional baseball players are good, so a baseball player is good because they are a professional. By defining professional baseball players as good, it is not a causal relationship between being professional and being good.
    
    
• Units of analysis are the objects that you want to draw conclusions about. These may be different from the units of observation which are the actual units you are studying. For example, researchers sometimes study an individual (unit of observation) in order to draw conclusions about families (units of analysis).
  
  • Ecological fallacy occurs when you mismatch units of analysis. When you have data from states as a whole and try to make assertions about the individuals in those states. This is a problem because the information that you have for the entire state does not apply to each individual. An example would be if you had information that Colorado College is a split Division 1/Division 3 NCAA school, the ecological fallacy would be concluding that a given CC varsity athlete is both Division 1 and Division 3.
    
  • Reductionism is similar to the ecological fallacy, but in reverse. Reductionism is when you apply data about an individual level of observation and try to make statements about a group level of analysis. An example of reductionism would be to say that the qualities or characteristics of group members apply to the group itself. Individual members of Colorado College are often described as "liberal"; despite this characterization, the institution itself is not necessarily "liberal"--in its finances, its political positions, its organization of the curriculum.