Select a State

Menu

Methodology

This page gives an overview of the methodology and data sources used for the Compass.

Methodology

Quantile Regression

To understand the effect of one or multiple variables (thereafter “independent variables”) on another (thereafter “dependent variable”) one of the most powerful and ubiquitous tools in statistics is a regression. A regression tells us by how much the dependent variable changes as the values of the independent variables change on average.

For example, if we want to know the relationship between the miles per gallon of a car (the dependent variable) and the weight of a car (an independent variable), a regression may tell us that for every extra 1,000 pounds a car weighs, its miles per gallon is associated with a decrease of 10 on average. The key idea here is that a regression gives us an idea how much the dependent variable (here miles per gallon) changes when the independent variable (here weight) changes.

The Compass uses a specific form of regression analysis called a “quantile regression.” A quantile is a specific “cut point” that divides data into equally sized groups. We are concerned with the middle quantile or the median. The median is the number where half the data will be above it and half below. For example, the median weight for a group of cars is the weight for which half the cars will be heavier and half lighter.

A quantile regression involves not just finding the average relationship between the dependent and independent variables. Instead, a quantile regression finds the relationship at specific quantiles. For example, in the Compass we want to know what the relationship is between the median net worth and home equity. We do not want the average net worth because average net worth is skewed by a few individuals with enormous net worth, think Bill Gates.

Decomposition Analysis

On the landing page for each state in the Compass, under “Key Takeaways,” we analyze how much each factor of interest contributes to the ethno-racial disparities we see. For example, for Michigan, we write that “home equity is a large contributor to ethno-racial disparities in net worth.” We make these determinations based on what is called a decomposition analysis. A decomposition analysis helps us quantify which variables contribute the most to the differences (e.g., in net worth) we observe across ethno-racial groups.

Data

Ethno-Racial Groups

There are many ways to divide people into ethnic and racial groups. Due to limited data, we chose to use only five large ethno-racial groups. White, Black, Hispanic, Asian, and other. “Other” is anyone who doesn’t fall into one of the first four groups. These groups are exhaustive and mutually exclusive. In states where we did not have enough data on one of the groups they were grouped into “other.”  This is not meant to suggest that these are the only groupings that matter or that ethnicity and race is as simple as a group label. However, the data is limited. There is more data on wages, drawn from the ACS survey, than there is on net worth or home equity. Therefore, some states show data on more ethno-racial groups for wages than for net worth.

Data Sources

Survey of Income and Program Participation

Data on net worth and home equity come from the Survey of Income and Program Participation (SIPP). SIPP is a national survey conducted by the Census Bureau that measures, among other things, household wealth and home equity. SIPP is designed to track the use of welfare programs and how those are tied to income and wealth. SIPP also has data on household composition and race.

SIPP tracks survey recipients for up to four years to see how their lives have changed. However, for the Compass we focus on the most recent year of SIPP data (2022). SIPP data is collected at both the individual and household level. The Compass uses SIPP data only at the household level because wealth and home equity are usually thought of as relating to a household.

American Community Survey

Data on wages come from the Census Bureau’s American Community Survey (ACS). The ACS is an annual survey that collects demographic and other information across the US. The ACS includes data from millions of individuals. The ACS collects data that is usually only available in the decennial census (which is only conducted every 10 years). The ACS collects individual data on hours worked and wages.

Contributors to Wealth and Net Worth

Natoque auctor malesuada amet vitae molestie fusce platea metus tempor pellentesque felis pulvinar class. Molestie nisl nulla

Assets

Assets directly feed into calculation of net worth: more assets means more wealth. Assets tend to go hand-in-hand with debts (e.g., purchasing a house often involves taking out a mortgage). Assets also affect and are affected by income and expenses: some assets produce income, while a higher income-to-expenses ratio allows for greater ability to purchase assets.

Debts

Debts directly feed into calculation of net worth: more debts means less wealth. Debts sometimes (but not always) indicate an investment in an asset or future income (e.g., a mortgage for a house, or education debt for higher future income). Debts also affect and are affected by income and expenses: debts typically have ongoing expenses to pay them off, while a higher income-to-expenses ratio allows for greater ability to pay them off.

Income

Income indirectly feeds into net worth via assets and debts: typically, a higher income translates to more wealth. A higher income-to-expense ratio allows for greater ability to purchase assets and pay off debt. Income might also be affected by assets and debts: some assets produce income, while some debts indicate an investment in income (e.g., education debt).

Expenses

Expenses indirectly feed into net worth via assets and debts: typically, higher expenses translates to less wealth. A higher income-to-expense ratio allows for greater ability to purchase assets and pay off debt. Expenses are also often affected by debts via required monthly payments.

Default Title

Default Description