Chapter 3 in Kellstedt and Whitten (2009)

Evaluating Causal Relationships

Four Causal Hurdles:

1. Is there a credible causal mechanism that connects X to Y?
2. Can we eliminate the possibility that Y causes X?
3. Is there covariation between X and Y?
4. Is there some Z related to both X and Y that makes the observed relationship between X and Y spurious?

Causality and Everyday language

1. Bivariate: most theories are limited to describe relationships between a single cause (independent variable) and a single effect (dependent variable) – a simplification of reality
2. Multivariate: Social reality is complex; one dependent variable can have several causes
3. Deterministic: indicative of certainty in relationship
4. Probabilistic: most human interactions are probabilistic, without certainties

Four hurdles along the route to establishing causal relationships

1. Four hurdles
   1. Is the a credible causal mechanism that connects X to Y?
      1. How & why will more/less of X result in a change in Y?
   2. Could Y cause X?
      1. Logical assessment
   3. Is there covariation between X and Y?
      1. Are they correlated/associated?
      2. Measurable association
      3. “correlation does not prove causality”
   4. Is there some confounding variable Z that is related to both X and Y and makes the observed association between X and Y spurious?
      1. Control for any possible Z
      2. Spurious: false relationship
2. Putting it all together – adding up the answers to our four questions
   1. Failure of hurdles 1 & 4 = rethink hypothesis/theory
   2. Failure of hurdles 2 & 3 does not mean that the causal claim is false
      
3. Identifying causal claims is an essential thinking skill
   1. Identify the imbedded causal claims in statements (E.g. A particular politician’s economic policies will improve the economy)
4. What are the consequences of failing to control for other possible causes? – fail to cross the fourth hurdle
   1. Validity of the conclusion is affected

Why is Studying causality so important? Three examples from political science

1. Life satisfaction and democratic stability
   1. Ronald Inglehart (1988) argued life satisfaction (X) caused democratic system stability (Y)
      1. Hurdle 1: PASS – Satisfied people won’t undermine their government
      2. Hurdle 2: MAYBE – Life satisfaction can also be caused by democratic system stability
      3. Hurdle 3: PASS – data showed covariation
      4. Hurdle 4: MAYBE – lots of other variables lead to democratic stability
2. School choice and student achievement
   1. Type of school a child attends (X) affect student performance (Y)
      1. Hd 1: PASS – Private school may have better programs that enables stronger performance
      2. Hd 2: PASS – Since the end results cannot determine the type of school (Private/public)
      3. Hd 3: PASS – data supports
      4. Hd 4: MAYBE – numerous other factors account for student performance. One Z could be Parental involvement.
   2. Selection effect: a situation in which a systematic force causes only a nonrandom subset of eligible targets to participate in a program
3. Electoral systems and the number of political parties
   
   1. The more societal divisions there are that shape a political culture, the more political parties there will be in the legislature
      1. HD: Bandwagoning of parties
   2. Assuming that political culture is kept constant (controlled), the more disproportional -ratio of translating votes into seats- the electoral system is in translating votes into seats (X), the fewer political parties will be represented in the legislature (Y).
      1. Hd 1: PASS: Bandwagoning of parties
      2. Hd 2: PASS: Parties cannot affect the electoral system
      3. Hd 3: PASS (for now): The case of Post WW2 German support the claim as number of political parties went down.
      4. Hd 4: PASS:  unable to find a Z

Why is studying causality so important? Three examples from everyday life

1. Alcohol consumption and income
2. Treatment choice and breast cancer survival
3. Explicit lyrics and teen sexual behaviour

Conclusion: critically evaluate causal statements using the four hurdles.