Hey, how much money do you make?

What an obnoxious question for a stranger to ask you, right? But I’m going to up the jerk ante. I don’t even care about how much money you make, but rather I want to know what you take into account when you answer the question.

In case you haven’t noticed, people on the internet often argue about inequality and incomes. But disagreements that seems to be about income stagnation, poverty or inequality are often really about what we are counting. For instance: When conservatives argue that average workers are doing fine, they're basing that on a measure that includes income from the very same government programs they're trying to cut.

Think of data like a flashlight: what you can describe depends on where you are pointing it.

So how could you answer the question? There are five broad categories.

1. Cash wages. This was likely your answer to the question. You make $50,000 a year. Or you make $8 an hour and struggle to get enough hours to make ends meet.

According to the CBO, for the median quintile, or those between 40 and 60 percent on the market income distribution, this increased a total $1,728 from 1979 to 2007. (This quintile does include some people who aren’t working, such as retirees, but it is the best source available for breaking out this data by income type.)

2. Employer fringe benefits, health insurance and taxes. This is important from your employer’s point of view. It pays a share of payroll taxes, contributes to deferred compensation plans and, crucially, contributes to your health insurance. Add these benefits to your cash wages and you get your total labor income. For the median quintile, this increased $3,104 between 1979 and 2007.

3. Other non-labor income. This includes all the other market income, such as capital income, capital gains, business income and “other income.” Add this to labor income to find your total market income. Sometimes researchers separate out capital gains into its own category, sometimes not. For the median quintile, this was up roughly $4,468 from 1979 to 2007.

4. Government transfers. This is usually separated into programs that pay cash, like Social Security or unemployment insurance, and those that pay with vouchers (food stamps) and in-kind benefits (Medicare, Medicaid). Researchers are constantly debating how to best answer the difficult question of estimating the value of services, especially health care benefits like Medicaid. For the median quintile, this was up added up to $5,100 from 1979 to 2007. Adding government transfers to market income finds your total income.

5. Total paid in taxes. If you have to pay very high taxes, you obviously take home less income. Federal taxes paid from 1979 to 2009 for the median quintile brings income down $500. This is lower for people further down the income distribution due to expansions in tax credits, particularly the child tax credit and the Earned Income Tax Credit (EITC). The median quintile paid $100 less in federal taxes from 1979 to 2007.

Here’s a graph of these changes for a few of the sources over time:

(Note that market income dropped from 2007 to 2009, the last year we have data for, in the Great Recession, offset by less paid in taxes and higher government transfers.)

Clearly when a researcher sits down to study changes in income, these are a lot of variables to take into account -- but there are even more to consider. The above is for an individual. What if we want to generalize about income for more than one person, as researchers have to do? They have to figure out who to include.

Are they including non-workers, like retirees and students? Another question is the measurement size. Are they looking at households, which can include groups that don’t really share economic resources, like friends who are roommates? Or are they sticking to tax units used by the IRS?

If they are looking at rates of change over time, sometimes researchers debate which inflation rate to use to think about how the value of money has grown over time. A few economists also focus on consumption data, instead of income data, to get a sense of what people are actually buying instead of the cash they have.

So let’s try to moderate some debates on the economics blogosphere. James Pethokoukis of the American Enterprise Institute argues that President Obama doesn’t have a case for inequality and median household stagnation. Why? Citing the work of Richard Burkhauser, he says “median household income – properly measured – rose 36.7%, not 3.2% like Piketty and Saez argue.“ (Thomas Edsall has a backgrounder on this disagreement here.)

But what does properly measured mean? The Piketty and Saez data looks at cash wages and other income without the capital gains (so number one, and part of number three). Burkhauser adds in taxes, transfers and healthcare (so everything, or numbers one through five above), which isn’t in the Piketty and Saez data, and adjusts for household size.

Which is more accurate? It depends on what you are trying to answer. If you are looking at how the distribution of market income itself is evolving, the first approach is important. If you are looking at the money in people’s pockets, total income after taxes and transfers is more important.

Here’s another example regarding poverty. Michael Tanner of the Cato Institute argues that “the poverty rate has remained relatively constant since 1965, despite rising welfare spending.” Bruce Meyer and James Sullivan, in a paper for the Brookings Institute, argue that the poverty rate has collapsed. Who is right? It depends on where you look.

Tanner is referencing the normal poverty measure, which is total market income, plus the cash portion of government transfers (numbers one to three, and part of four, from above). The Brookings paper takes that and includes taxes and full transfers (so all of four and five), which reduces poverty. They also use a lower rate of inflation and consumption to show an even further decline. Whatever you think of the later changes, it would be a problem to look at a poverty rate that didn’t include the earned income tax credit.

No matter where you shine the data flashlight, the top 1% have done very well. Top-end inequality has skyrocketed, no matter how you approach median income. Meanwhile, as healthcare becomes more important to people’s compensation, efficiency in health care services becomes more important for people. Medicare overpaying for prescription drugs doesn’t give real resources to the elderly, even though some measurements would treat this as cash income.

Politically, conservatives are in a double-bind when it comes to the policy solutions for inequality. Many conservatives use the all-inclusive definition of income -- one that includes, and in fact heavily relies on, government benefits -- to argue that income hasn’t stagnated. But many conservatives would also like to see government programs cut significantly.

These programs are doing serious work to keep median wages from stagnating and poverty from becoming an epidemic. They are the front line defense against inequality, and it also the front line conservatives are secretly relying on data-wise to argue that inequality is no big deal.

Mike Konczal is a fellow at the Roosevelt Institute , where he focuses on financial regulation, inequality and unemployment. He writes a weekly column for Wonkblog. Follow him on twitter here.