and it sounds like it's a fancy word right but variate just means variables okay and so by means two multivariate means more than two okay so instead of trying to explain our phenomenon our dependent variable using one variable we're going to use several independent variables and the elaboration model you know if our slides will change the elaboration model allows us to explain the relationship between two variables by controlling for the effects of a third variable so ideally if you're putting together models and we didn't really look at models in statistics right we're looking at something where we think several variables play a role we want to be able to include anything that we think might be playing a role okay so chapter one discussed some studies on morale in the military um and morale seemed to be positively correlated to combat Effectiveness so that led people to believe that if we can imp improve morale the war effort would be more successful so still for at all sought to confirm some commonly accepted propositions about this so number one promotions should affect soldiers morale right if we're getting promoted generally we feel good about our job so they thought soldiers that were serving in units that had low promotion rates should have relatively low morale given that there was racial segregation and discrimination in the South they thought that black soldiers being trained in Northern training camps should have higher morale than those being trained in the South and Soldiers with more education should be more likely to resent being drafted than those who have less education but none of these things were true why so one thing that you will learn if you get into doing research is that you're usually going to be asked to try to explain your results and so here the the mechanism at play is reference groups and relative def deprivation so people are not evaluating themselves based on like all soldiers right they're we're not evaluating ourselves on like objective standards usually but on the basis of our position relative to people around us we compare ourselves with reference groups and so people felt relative deprivation if they didn't compare favorably with their reference group now the central logic of the elaboration model begins with an observed relationship between two variables and then we want to think about is there another variable that might actually be causing the relationship so in stofer's research the two initial variables are education level and acceptance of being drafted as fair right so that was like their third um hypothesis so here it should be that your your education is your independent variable and being accepting and being drafted was your dependent variable but another variable was actually up play whether or not your friends were drafted so Soldiers with more education would be more accepting of being drafted because it's likely that their friends would also have been drafted right so it seems fair to you if other people you know are being drafted and if other people you know are not being drafted then it's going to fail unfair those with the least educ were likely to have been in occupations that often got deferments from the draft leading those drafted to feel as though they've been treated unfairly Soldiers with low education were more likely to have friends who were deferred than Soldiers with more education and having friends who were deferred made a soldier more likely to think he should have been deferred so as education level increased acceptance of One's Own induction also increased okay so the nature of this empirical relationship was interpreted through the introdu of the third variable this is why when I've said to you guys before that it's important to have expertise in the field in which you're doing your research because you have to know what does the literature tell us okay what might be playing a role here and so when a group like Economist decides to look at the death penalty and whether or not the death penalty Works they don't have the appropriate education background and knowledge to be thinking about whether or not their models are correct so here this third variable deferment of friends it doesn't negate the original relationship it's there but it's clarifying what is really happening in that relationship so remember I said to you guys in statistics um that rates of drowning and rates of eating ice cream cones right or ice cream um are correlated okay so when drownings increase eating ice cream increases okay it's pretty obvious for us to look at that and be like well one can't be causing the other right but sometimes that's not clear because sometimes the two variables that are related seem like they should be related we know from discussing this drowning and ice cream relationship that what's really at play both of these things increase in the summer when there's warm weather so to start with we have to know if our test variable occurs before or if it intervenes so does it occur between okay so the logic H there in this multivariate relationship is that our independent variable education affects the intervening test variable having friends who are deferred or not and that in turn affects the dependent variable okay so in this case it's an intervening variable it's occurring between if the test variables an antecedent okay then the independent and dependent variables are empirically related to each other but there's no causal link they're related because they're both related to this other relationship this other variable now when this partial relationship are essentially the same as the original relationship we can say that replication has occurred right that the original relationship has been replicated under test conditions so this elaboration model is used routinely in the hope of replicating findings among subsets of a sample so say we discover a relationship between education and Prejudice we might start introducing other variables to to make sure it's not something else at play is it really about age is it really about where you live in the country race religion right this helps us test whether or not that original relationship is true so a turn used to explain aerious relationship um aerious relationship is an original relationship shown to be false through the introduction of the test variable okay so the relationship between ice cream sales and drowning rates is spous okay it's false it's not real they're they're related but they're both related to the same variable which is summer okay so in order to be able to say we have an explanation we need the test variable has to be antecedent to both the independent and dependent variables and the partial relationships have to be zero or significantly less than those found in the original relationship so when it's an intervening variable when the variable comes in between the independent variable and the dependent variable okay that means there's a spous relationship there so babby uh found a relationship between region of the country in which Med School faculty attended medical school and their attitudes towards Medicare so among those attending Eastern schools 78% said they approved of Medicare amongst those attending Southern medical schools only 59% did now this made sense uh in they in light of the fact that the South seemed more resistant to these programs than the East did and um presumably where we get our education right would impact us now they introduced an incedent variable the region of the country in which the faculty member was raised and when you introduce this the region of the medical school training has little to no effect now interpretation is similar to explanation except for the time placement of the test variable and the implications that follow from that so interpretation represents the research outcome in which a test or control variable is discovered to be a mediating Factor right um mediating meaning an intervening variable researchers observe that children from broken homes are more likely to become delinquent than those from intact homes okay is it then something about divorce or never being married that causes your children to be more delinquent no supervision is the variable that is actually intervening okay sometimes the elaboration model show some partial relationships that are different some might be stronger than others right so this process is called specification you guys might have read some um Journal articles that are very specific about the circumstances under which a result occurs right um so Glock and colleagues did a study in 1967 they looked at Episcopal church members and they found that involvement decreased as social class increased so as they became higher social class they were less involved in the church so Church involvement provides an alternative form of gratification for people who maybe are not like feeling success or gratification in the larger society this helped to explain why women were more religious than men why old people were more religious than young people because those people had less social capital um and therefore a place that they could gain that was through the church so Glock reason that people of lower social class had fewer chances to gain self-esteem and secular society now if social class were related to church involvement only by virtue of the fact that lower class women would be denied opportunities for gratification in secular society the original original relationship should disappear amongst women who are getting gratification social class should be unrelated to church involvement amongst women who held an office position among women who have held office in secular organizations there is essentially no relationship between social class and charch involvement so Glock has specified conditions under which the original relationship holds so a lot of research in social sciences because we are dealing with people and people are unique is going to have specifications it's going to hold under certain circumstances there's very few things that sort of hold generally across the board um one fact that has been found consistently for decades in criminology is that men commit the overwhelming majority of crime they're responsible for 75% of crime women are responsible for 25% of crime there was a theory that that would change as women gained um more equality and access to the workplace and it did not so our basic Paradigm right thought process we're operating under here there's a relationship between two variables okay Rosenberg suggests we can do this even in cases where there's no original relationship right so we could say uh maybe we want to study Union membership and attitudes towards having Jewish folks in the union staff okay so analysis indicated there was no relationship between length of union membership and these attitudes people who belong to the union for four less than four years were just as willing like had the same attitudes as people who belong for more than 4 years what was really playing a role here was not how long they'd been in the union but how old the union members were younger members were more favorable towards Jewish folks than older members were and then within specific age groups those in the union the longest were most supportive so age is acting as a suppressor variable right it's concealing the relationship between the length of membership and attitudes this is again why it's really important that you H that you do that literature review and look at what do we already know so that you know what variables to include in your models so the basic Paradigm focuses on those partial relationships being the same as or weaker than the original relationships but it doesn't give us any guidelines for specifying when is that relationship significant okay so the limitations of this Paradigm to par partials that are the same or weaker neglects two other possibilities maybe the partial relationship is stronger maybe it's the opposite of the original relationship so looking at the starting salaries of men and women in the same organization deserve found that women had higher starting salaries there was a distorter variable here playing a role though it had to do with the time at first higher okay when they were first hired was impacting what their starting salaries were we also um want to think about X post facto hypothesis X postao just means after the fact so an X postao hypothesis is a hypothesis created after confirming data has already been collected you need to start your research with your hypothesis and then reject or retain it um because an X postao hypothesis cannot be disconfirmed