Most of our theories are limited to descriptions of A single cause independent variable and a single effect relationships between what dependent variable Almost all of our theories about social and political phenomena are Bivariate involving just two variables Social reality is not bivariate instead it is Multivariate any interesting dependent variable is caused by more than one factors What must we do when it comes time to design research to test our theoretical ideas We have to try to account for or control for those other causes If we don t then our causal inferences about whether X causes Y may very well be wrong The notions of causality that come to mind in the world of the physical sciences are mostly Deterministic Relationships Deterministic Relationships Relationships such that if some cause occurs then the effect will occur with certainty In contrast to the world of physical sciences the world of human interactions consists of relationships Probabilistic Relationships Probabilistic Relationships Increases in X are associated with increases or decreases in the probability of Y occurring These probabilities are not certainties When is a relationship between variables probabilistic not One case alone does not decrease our confidence in the theory deterministic Example of a probabilistic and deterministic statement taxes What procedures four hurdles must we follow before we can express our degree of confidence that a causal relationship does does not exist First Hurdle Is there a credible causal mechanism that connects X to Y Second Hurdle Can we rule out the possibility that Y could cause X Probabilistic wealthy people are more likely to prefer lower taxes whereas poorer individuals are more likely to prefer higher Deterministic wealthy people will prefer lower taxes and poorer people will prefer higher taxes 1 Is there a credible causal mechanism that connects X to Y Is there covariation between X and Y 2 Can we rule out the possibility that Y could cause X 3 4 Have we controlled for all confounding variables Z that might make the association between X and Y spurious Effort to answer the question how and why Ask yourself what is the process mechanism that logically speaking suggests that X might be a cause of Y Failure to clear this hurdle is very serious the results are that o 1 Our theory needs to be thrown out altogether or o 2 We need to revise it after some careful rethinking of the underlying mechanisms through which it works You can only proceed if you get a yes Ex when considering whether a person s gender X causes him her to have particular attitudes about abortion policy Y it is a rock solid certainty that the reverse causal scenario can be dismissed a person s attitudes about abortion does not cause them to be male or female Confounding Variable A variable that is both correlated with both the independent and dependent variable and that somehow alters the relationship between those two variables Spurious Not what it appears to be or false Which two hurdles involve a Z variable In hurdle four there are commonly several Z variables Hurdle 3 4 It is possible for a causal relationship to exist between X There is no bivariate association between X and Y and Y even when Most of the interesting dependent variables are caused by More than one often more than one independent variable What problems does the multiple independent variables pose for social science When trying to establish whether a particular X causes a particular Y we need to control for the effects of other causes of Y we call those other effects of Z What is likely to happen if we fail to control for the effects what of Z We are likely to misunderstand the relationship between X and Y and make the wrong inference about whether X causes Y This is the most serious mistake a social scientist can make If we find that X and Y are correlated but that when we control for the effects of Z on both X and Y the association between X and Y disappears then the relationship between X and Y is said to be spurious The challenge of evaluating a claim that X causes Y involves what Summing the answers to all four of these questions to determine our overall confidence about whether X causes When will our confidence in our theory be greatest When we are able to answer all four questions the right way and without reservation Causal Hurdles Scorecard A shorthand for summarizing the answers to the four questions in square brackets y y y y Which scorecard can be thrown out instantly Which scorecard would have a reasonable level of evidence in its favor Y n y y y y n y y What kind of skill is spotting and identifying causal claims Thinking Skill Why is studying causality so important Three examples Life satisfaction and democratic stability Race and political participation in the US Evaluating whether head start is effective Life Satisfaction and Democratic Stability Scorecard YES there is a credible causal mechanism that connects X to Y NO we cannot eliminate the possibility that democratic stability is what causes life satisfaction YES there is a correlation between the two variables MAYBE ingelhart has maybe done an adequate job of controlling for those other factors Race and Political Participation in the US Evaluating whether head start is effective YES there might be residual effects of barriers placed upon African Americans and political participation YES we can eliminate the possibility that varying rates of participation cause an individual s racial classification YES there is correlation between an individual s race and their level of participation in the US NO the correlation that we observe is spurious and is not a function of race but instead of the disparities in wealth between Anglos and other races YES the preschool environment that anticipates the actual school setting helps prepare children for what they will encounter beyond YES we can rule out that educational outcomes could cause participation in the program YES studies have shown there is correlation YES all other variables parental involvement have been controlled What constitutes a good theory After going through rigors of the evaluation process the theory makes a contribution to scientific knowledge In other words it is one that changes the way we think about some aspect of the political world Based on the rules of the road we want our theories to be what 1 Causal 2 Not driven by data alone 3 Empirical 4 Non normative 5 General and parsimonious How do I
View Full Document
Unlocking...