A recent article by sociologist Jennifer Stansfield, “Intersectionality and sociology: A meta-analytic analysis,” argues that there are three distinct approaches to analyzing social science.
In this article, she examines the first, the “meta-analytical approach,” which focuses on understanding the relationships between two or more variables to find the relationship between each.
Stansfields work examines this approach in relation to the literature on the intersection of race and class, which she finds to be a useful tool in this analysis.
Stanfields work is based on data from the Social Science Research Network, a consortium of organizations that share information on sociology, economics, and other disciplines.
This research provides the basis for Stansfords analysis.
“In order to use the meta-analyses to understand the relationships among these variables, we need to understand how they are organized in a system,” Stanswells explains.
In the context of the meta method, the first step in this method is to take a data set and then create an equation for each variable that describes the relationship that exists between those variables.
This is done by building an equation using the following two variables: the total number of black people in a given city and the total amount of white people in that city.
The equation is then run against the data to determine which variable is the most correlated with the other variables.
The first step of the analysis is to identify the variable that is most correlated between the two variables.
Stays is able to identify a variable that has a high correlation with the amount of black residents in the city and a high amount of whites in the same city.
She then creates an equation to show how this variable is related to the other variable.
“The second step is to analyze the relationship by calculating the average value of the variables across the two cities,” Stunsfield explains.
The analysis then examines how this correlation is related across all variables, as well as the correlation between the variables and other variables in the data.
Stunsfields analysis of the correlation coefficient between the amount and number of blacks in the cities of Oakland, California and New York City is quite clear.
She concludes that the correlation is approximately 0.71.
Using the same methodology, she finds that the correlations between the black population in the two city is nearly exactly 0.7, with a correlation coefficient of 0.73.
“There are many examples of correlation coefficients of 0 to 0.8 across a large number of variables,” Stonsfield says.
This suggests that the relationship found between the number of white residents and the amount black people are in a city is approximately zero.
“It is important to understand that the data used in the meta analysis is not random,” Stoutsfield notes.
The correlation coefficient does not indicate how much the variable is correlated with each other, but it does indicate that the variable has a significant correlation with each of the other two variables, and that the total correlation coefficient is close to zero.
The data from Oakland, for instance, was collected in 2005, while New York was not until the 1980s.
“To find a correlation between a variable and the other, one has to look at the data for all the variables,” she says.
In Stansfelds case, she found that the city of New York had a very high correlation, with the number one variable in the equation being black people.
However, she noted that there were other variables that were associated with the correlation, and these were not necessarily associated with each others.
The variables that are most correlated in the analysis include: black men (all males), white women (all women), and the number that are in the Black population in a particular city.
“So, it is important not to rely on the correlation coefficients,” Stainsfields points out.
Instead, one should look at how each of these variables affects the overall correlation of the data, and then use that as the basis to build an equation that tells us how the other three variables relate to each other.
Stensfield’s analysis of this analysis was done on data for the entire dataset, from 2005 to 2020, and found that all of the correlations were between zero or between 1.0 and 1.8.
“This suggests that this is a useful and effective method to look into the relationship of variables in sociology,” Stensfields says.
The next step of her analysis was to use regression to determine how the correlations are related to each variable.
This was done by taking the correlation for the variable from the equation, and subtracting the correlation from the average of the two.
This value is then compared to the correlation value found by the regression.
“With regression, it takes a variable, like the number black in a town, and divides it by the number white in that town,” Stosfield explains, “and then finds the average.
For example, if the correlation of a variable is 0.9