Chapter 15: The Elaboration Model Introduction • Elaboration model o Also called the interpretation method, the Lazarsfeld method, or the Columbia School • Logical model for understanding relationship between two variables by controlling for the effects of the third • Shows the logical process of scientific analysis The Origins of the Elaboration Model • Samuel Stouffer o Researcher for the army studied morale in order to make the military more effective o He thought Units with low promotion have low morale Because of racism, blacks in the south would have lower morale than in the north Men with more education would resent being drafted o None of these were true! o Relative deprivation – people compared themselves to their reference group Rapid promotions meant many soldiers were passed over In the south, soldiers were insulated from adverse cultural norms Educated draftees accepted their inductions – why? • Friends have the same educational status • Draftage men with less education are probably in semiskilled production line occupations • Productionline industries are exempt from the draft • A man with little education is more likely to have friends in draftexempt occupations • A less educated draftee – when compared to his friends who weren’t drafted – will feel discriminated against o Logic – an observed relationship between two variables and the possibility that one variable may be causing the other o Positive association between education and acceptance of induction • Kendall and Lazarsfeld o Created hypothetical tables about what the data might have looked like if Stouffer had done the research • Having observed an empirical relationship between two variables, we seek to understand the nature of that relationship through the effects produced by introducing other variables • Test variable o Control variable o Held constant in an attempt to clarify the relationship between two other variables • Partial tables o Divide sample into subsets o Then the relationship between the two variables is recomputed separately for each subsample o Partial relationships – relationship between two variables when examined in a subset of cases defined by a third variable • Zeroorder relationship – the original relationship between two variables o No test variables controlled for The Elaboration Paradigm • Test variable o Antecedent – prior in time o Intervening • Replication 1
•
•
•
•
o When the partial relationships are the same as the original relationship o Proves the initial relationship is genuine Explanation o Describes a spurious relationship o The original relationship is shown to be false Test variable must be the antecedent to both the independent and dependent variables Partial relationships must be zero or significantly less than those found in the original o The relationship disappears when the antecedent test variable is introduced Interpretation o The control variable is discovered to be the mediating factor through which an independent variable has its effect on the dependent variable o The intervening variable helps interpret the mechanism through which the relationship occurs o Doesn’t deny the validity of the original causal relationship o Simply clarifies the process Specification o Produces partial relationships that differ significantly from each other One is stronger, one is less or nonexistent o Specified the conditions under which the original relationship occurs (for men, but not women) Refinements to the Paradigm o Differentiating between positive and negative relationships could be useful You could even use the elaboration model when the original relationship is zero Suppressor variable – a test variable that prevents a genuine relationship from appearing at the zeroorder level o There is no specification for what a significant difference between the original and partial relationships o Include partial relationships that are possibly stronger, or the reverse Distorter variable – reverses the direction of a zeroorder relationship o The paradigm gets extremely complicated when using three or more subsamples
Elaboration and Ex Post Facto Hypothesizing • Ex post facto hypothesis – created after confirming data that has already been collected • It’s a meaningless construct because there is no way for it to be disconfirmed • “ex post facto” means “after the fact” • This inhibits honest hypothesizing after the fact • Researchers make a lot of poorly reasoned hypotheses before they do the research to avoid this • Or worse, they try to ignore any previously observed relationships • Data analysis is a continuing process