SCF Debt Figures: Check Who Is in the Median Before Comparing Yourself

A debt median is meaningful only after you know who is included. “Among families with this debt” answers a different question from “among all families.” Removing zero balances can change the middle value even though nobody’s debt changes.

Before comparing a headline with your own balance, identify its group, statistic and date. This guide shows why that check matters, using an invented exercise followed by a separate official table example. A survey comparison is not a recommended debt level.

Start with ten invented families

Imagine a complete population of ten families, each counted equally. These are fictional balances for one debt category at one snapshot, in a common dollar basis. They are not SCF observations. A zero means no balance in this category, not freedom from every kind of debt.

Invented balances, sorted from smallest to largest; positions show order, not dollar distance

Family position	Balance
1	$0
2	$0
3	$0
4	$0
5	$0
6	$0
7	$1,000
8	$3,000
9	$8,000
10	$28,000

The balances add to $40,000. Four of the ten families have positive balances, so the holding share is 40%.

Calculate the middle twice

For all ten families, the middle positions are fifth and sixth. Both are zero, so the median is $0. The mean, or arithmetic average, is $40,000 divided by ten: $4,000.

Now ask only about families holding this debt. Their sorted balances are $1,000, $3,000, $8,000 and $28,000. The middle two are $3,000 and $8,000; their midpoint is $5,500. The same $40,000 divided by four gives a $10,000 mean.

Results from the fictional exercise only

Group	Families	Median	Mean
All families	10	$0	$4,000
Positive-balance holders	4	$5,500	$10,000

No family in this exercise owes exactly $5,500. With an even number of observations, the median can fall between the two middle amounts. Nor does a $0 all-family median mean nobody has debt: four families do.

Two shortcuts that fail

First, “average” does not tell you the median. Change only the largest invented balance from $28,000 to $68,000. The all-family mean rises to $8,000 and the holder mean to $20,000. Both medians stay unchanged because the middle positions have not moved.

Second, a holding share cannot convert one median into another. In the original example, 40% multiplied by the $10,000 holder mean equals the $4,000 all-family mean. But 40% multiplied by the $5,500 holder median gives $2,200, not the actual $0 all-family median.

That mean relationship works here because every nonholder has a zero and the groups use consistent counting. It is not permission to reconstruct precise survey estimates from rounded published figures. If nobody holds the debt, the holder mean and median are undefined. Missing information must not be silently turned into a zero.

Read a real SCF row separately

The latest Survey of Consumer Finances (SCF) is the 2025 survey, released October 9, 2026. Historical tables have nominal and inflation-adjusted versions; summary extracts use 2025 dollars. Federal Reserve SCF release and data.

In Table 4 of the Federal Reserve’s report, printed page 25, the 2025 credit-card-balances row reads as follows. Dollar entries are in thousands of 2025 dollars.

Official 2025 SCF credit-card-balances row, with dollar entries converted from thousands

Measure	Value	Comparison group
Share holding	44.7%	All families
Median balance	$3,100	Families holding this debt
Mean balance	$8,000	Families holding this debt

Here, a family is the primary economic unit: an economically dominant person or couple and financially interdependent household members, not necessarily every resident. Credit-card balances exclude purchases since the latest bill payment. Debts generally describe interview-time positions; income refers to the preceding calendar year. The report generally adjusts dollars to 2025 using CPI-U-RS. Table 4 and definitions, pages 25 and 39–45.

Keep this official row separate from the invented exercise. The share and balance statistics answer different questions. Neither balance statistic tells you how much debt is affordable for a particular family, or establishes a repayment target.

Why the classroom arithmetic is not a survey estimator

Our exercise has ten equally counted families and no missing values. Working with actual survey records requires more than sorting a spreadsheet.

The 2025 codebook identifies X42001 as the estimation weight. Weighting addresses unequal selection probabilities and nonresponse; unweighted record counts are not population shares. 2025 SCF codebook.

Five imputed records represent each family, not five independent families. Uncertainty calculations must account for sampling and imputation, using the applicable replicate-weight and implicate-combining guidance. SCF standard-error guidance.

This guide calculates no SCF microdata estimates and makes no statistical-significance claims. Its arithmetic checks validate only the fictional exercise.

Repair the headline before using it

Consider this deliberately fictional headline: “The typical family owes $10,000.” For our exercise, a precise replacement is: “Families with a positive balance average $10,000; the middle balance across all ten families is $0.”

When saving a real figure, use this extraction checklist:

• Which debt category and definition?
• Which family or household unit?
• All families, or only holders?
• Mean, median or holding share?
• Which survey year, release date and observation period?
• Which dollar multiplier and inflation basis?
• Which weighting and uncertainty limitations?

If a headline omits those labels, find the underlying table before treating its number as a useful comparison.