26  Income Inequality

When we talk about “income inequality,” we usually mean one of two things:

In this chapter, we’ll talk about the first of these.

Back in the second dig-deeper assignment, one of the possible tasks was to address the claim that people with more education make more money. This project is more open-ended. You don’t have to study education level; you can pick any inequality you like. Typical (or stereotypical) inequalities in our society have to do with

Those are all suitable avenues to pursue. But as you look at what data are available, you might consider others, such as

26.1 The task

As before, this is a “dig-deeper” project, with its charateristic form. In case you’ve forgotten what that is, this might help:

You’re exploring income inequality in the United States. This might be some injustice you want to probe, or just a difference.

Begin by exploring

Explore data that you assemble using the Microdata Portal (see below).

  • Check out the various attributes that are available. Choose some.
  • Download some people.
  • Make graphs, look for patterns and stories.
  • Don’t be afraid to download new datasets when you see you want different attributes, years, or States. It’s fast! It’s free!
  • Come up with a claim (or question) you want to pursue.
  • Make the simplest possible graph or calculation to address your claim.
  • Then dig deeper.

How do you know when you’re done?

You have a Google Doc (or whatever format your instructor requires). It contains:

  • Your name.
  • A simple claim or question.
  • A graph with the first-look, obvious results.
  • A description of why that simple approach might not be enough to really explain the data
  • A description of what you did to further the investigation (“dig deeper”). In this description,
    • You have grouped the data by dragging an attribute (or attributes) to the left in the table.
    • You have calculated some summary value (e.g., a mean or a sum) for each group by making a new column with a formula.
  • The results of that deeper investigation (with, e.g., a new graph of the data)
  • A conclusion: is the story any different now?
  • Possibly, ideas for additional “dig deeper” activities with this data set, and…
  • A link to a shared CODAP doc (like last time) so the instructor can see what you did.

Your Google Doc is probably no more than two pages long. Be sure to set permissions so your instructor has edit access.

26.2 The data

In this project, you should use the Microdata Portal plugin. You can find it in the Plugins menu under Getting Data. Or click this link:

When you are exploring the data, check out the various attriubutes you can get in the Attributes section of the plugin. By default, you only get Sex, Age, Year, State, and Boundaries. That’s just not enough!

Getting more attributes

Because the project is about income, you’ll need data about that. Here’s how:

  1. Open up the Attributes section of the plugin. You’ll see a number of categories, including Income.
  2. Open up the Income category. You’ll see various attributes about income, including Income-total, Income-wages, and a few others.
  3. If you like, click the Show descriptions box to see descriptions of the attributes.

The illustration below shows what the plugin looks like as you do this:

Now, Income is not the only category. Explore the other categories to see what else might be useful for your project.

Specifying place and time

Above the Attributes section, you can see sections for Place and Year. Those will let you restrict which States and years you get data for.

The default is that you get a representative selection from all States, from the 2020 Census.

26.3 Data move opportunities

This is real data, almost direct from the Census. As a consequence, it will not be formatted or coded the way you probably want, and you might get cases that are irrelevant to your project. You may also want to summarize the income of groups, for example, comparing the median incomes of people with more or less education; that median means you’re summarizing.

That is, you’ll need to do data moves as part of your exploration and data analysis. Here are some suggestions for data moves you might make:

Filtering

You’re studying income, so you probably don’t want to include 5-year-olds in your data. You also might not want to include people who aren’t working. (Or maybe you do, depending on the story you’re telling.)

This is a job for filtering.

To get rid of children, you might select them in a graph of Age and set them aside. You also might discover something strange about their incomes that appear in the data; that could help you identify which ages the Census thinks are too young to have incomes.

If you want to focus only on people with jobs, check out the Work & emplyment section, and look at the Employment_status attribute.

Grouping

If you’re looking at income inequality, that probably means you’re going to compare groups. It might be two groups (e.g., people in Minnesota and Arkansas) or more groups (e.g., White, Black, Asian, Other).

You use the grouping data move when you put that attribute on a graph, or when you drag it left in a table to “promote” it in the table’s hierarchy.

Summarizing

When you compute some quantity that characterizes a group—and probably use it to compare groups—you’re summarizing.

The obvious thing to summarize in this project is Income-total. Three things about that:

  • What’s the best way to summarize income? Two clear choices are mean() and median(). Which one should you use? Or should you use both?
  • Those two examples are measures of center. But as you dig deeper, you might want to look at other possible summaries such as measures of spread. Or you might want something other than the center for your comparison, such as a percentile. There’s a function for that.
  • Finally, you might get insight into inequalities by summarizing something other than income. There are other “income” attributes available such as Income-family_total. There might be other things to look at as well…

Calculating and recoding

When you need new values for every individual case, every person, you may need to calculate that value or “recode” it from an existing attribute.

Clear as mud? Some examples:

Recoding your grouping attribute. Suppose you’re looking at education and income, and you’ve decided that you just want to compare people who have been to college with those who have not. We did this in the “calculating” chapter as a live example. Check it out.

Who’s an immigrant? There is no direct attribute to tell you who is an immigrant. But in the Race, ancestry, origins section, there’s an attribute called Immigrate-year.

You can use that to help you make a new column (perhaps Immigration) with two values: native and immigrant. You can use a formula strategy or a grouping strategy.

Comparing incomes from different years. This is trickier!

Suppose, as you dig deeper, you want to argue that the gender difference in income is better in 2020 than it was 50 years earlier, in 1970. You get data from both years, but you see that of course, incomes in 1970 were a lot smaller.

You notice the CPI99 attribute in the Income section, and decide to include it in order to account for that inflation. The illustration in the margin shows part of your table. Here’s the deal: To convert the value of Income-total to a “standard” value in 1999 dollars, multiply the income by the value in CPI99.

Ergo: Make a new column (perhaps income99) and give it the appropriate formula.

26.4 Miscellaneous tips

Census data, and the Microdata plugin, are great, but occasionally odd. Here are some tips about things we’ve learned over the years:

Topcodes. Some values—I’m thinking of poverty measures here, but it happens in various places—have “topcodes”, that is, values that get put in the dataset to represent that value or anything higher.

Imagine you have Age and you have only two digits to represent it. You might say, we’ll let 99 stand for age 99 and anything higher. It won’t make a big difference, but if you look at a graph of Age, there might be a little blip at 99. Does that mean that there are more 99-year-olds? No, just that the centenarians have been lumped among those spring chickens.

States and boundaries. As you may have leaned in the Maps chapter, CODAP sometimes needs “boundary” attributes in order to make maps. The microdata plugin supplies those with the State attribute, and “drags them left” by default. This looks weird, but facilitates making maps by State.

Of course, to use a particular value on a map (e.g., median income) you need to calculate it. Make a new attribute at that top level and give it a suitable formula. Then you can drag that attribute into a map.