Friendship Paradox Explained: Why Friends Have More Friends

Quick answer: The friendship paradox is the mathematical finding that, on average across a social network, your friends have more friends than you do. It happens because highly connected people appear on many friend lists, so they are overrepresented whenever we count friends. Sociologist Scott Feld explained it in 1991. In a 2011 study of about 721 million Facebook users, 92.7% had fewer friends than the average of their friends. It does not mean every person has fewer friends than their friends, and the percentage varies between networks.

Facts checked November 22, 2026.

Scroll through social media and it can feel like everyone else is more popular than you. Strangely, for most people that is mathematically true, and it has nothing to do with personality. The friendship paradox is a neat result about averages and networks, and it has practical uses, from tracking flu outbreaks to choosing whom to vaccinate first.

A Five-Person Example

Picture a small network: Ana is friends with Ben, Cal, Dee and Eli, and none of those four are friends with each other.

PersonNumber of friendsAverage friends of their friends
Ana41
Ben14
Cal14
Dee14
Eli14

The average person has \(\frac{4 + 1 + 1 + 1 + 1}{5} = 1.6\) friends. But four of the five people have a friend (Ana) with 4 friends. Four out of five people have fewer friends than their friends do. Ana is the only exception, and there is only one of her.

Why It Happens: Popular People Get Counted More

When you average “friends of friends,” you look at each friendship from both ends. Someone with 4 friends shows up on 4 different friend lists; someone with 1 friend shows up on only 1. Popular people are oversampled. Count it over every friendship end: in our network there are 4 friendships, so 8 friendship ends. The friend at each end has these friend counts: Ana appears 4 times (with 4 friends each time) and each of the others appears once (with 1 friend):

\[ \frac{4 \times 4 + 4 \times 1}{8} = \frac{20}{8} = 2.5. \]

So a randomly chosen friend has on average 2.5 friends, compared with 1.6 for a randomly chosen person.

Two Different Averages

There are two reasonable ways to average “friends of friends,” and they give different answers:

  • Per friendship (2.5): pick a random friendship and look at the friend at one end. This is the version in Feld’s formula.
  • Per person (3.4): for each person, average their friends’ friend counts, then average those numbers over people: \(\frac{1 + 4 + 4 + 4 + 4}{5} = 3.4\).

Both exceed 1.6. They differ because the second gives every person equal weight, while the first gives every friendship equal weight.

The Formula

Scott Feld showed that if \(K\) is a person’s number of friends, \(\mu = E[K]\) its average and \(\sigma^2\) its population variance (with \(\mu > 0\)), then the average number of friends of the friend at the end of a randomly chosen friendship is

\[ \frac{E[K^2]}{E[K]} = \mu + \frac{\sigma^2}{\mu}. \]

Check it on our network: \(\mu = 1.6\). The average of the squares is \(\frac{16 + 1 + 1 + 1 + 1}{5} = 4\), so \(\sigma^2 = 4 – 1.6^2 = 1.44\) and

\[ 1.6 + \frac{1.44}{1.6} = 1.6 + 0.9 = 2.5. \]

It matches. The formula shows the paradox disappears only when \(\sigma^2 = 0\), that is, when everyone has exactly the same number of friends. Any variation at all makes friends more popular than average people. Review averages and probability if variance is new.

What Facebook’s Data Showed

In 2011, researchers Johan Ugander, Brian Karrer, Lars Backstrom and Cameron Marlow analyzed the friendship network of 721 million active Facebook users:

MeasureValue
Median number of friends99
Expected friends of a friend at the end of a random friendship (\(\langle k^2 \rangle / \langle k \rangle\))635
Users with fewer friends than their friends’ average92.7%
Users with fewer friends than their friends’ median83.6%

The 635 is the same formula in another form: \(\frac{\langle k^2 \rangle}{\langle k \rangle} = \mu + \frac{\sigma^2}{\mu}\). Friend counts are very spread out, with a few hubs having thousands, so the variance is huge and so is the paradox.

Useful Applications

  • Early warning for epidemics: in a study published in 2010, Nicholas Christakis and James Fowler followed 744 Harvard students during the 2009 flu outbreak; the epidemic curve among friends nominated by randomly chosen students led the random group by an estimated 13.9 days, based on clinical diagnoses. Well-connected people tend to get infected sooner.
  • Smarter vaccination: “acquaintance immunization” vaccinates a random friend of a random person, which tends to reach well-connected people without needing to map the whole network.
  • Social media: the same math explains why your followers and friends often seem to have more followers, more posts and more exciting lives: active people are oversampled in your feed.

The epidemic link connects to R0 and the math of disease spread, and the “what you see is not a random sample” idea is closely related to sampling bias in statistics.

Common Misconceptions

“Everyone has fewer friends than their friends.” No. The guarantee is about averages over the network; hubs like Ana are exceptions, and the share of people who experience it depends on how the network is arranged.

“It means I am unpopular.” No. It is true for most people in almost any network, simply because averages over friends overweight popular people.

“It is a psychological illusion.” It is a real mathematical effect, not a bias in how people feel.

“It only applies to friendships.” It applies to any network: coauthors, followers, sexual partners and airports all show it.

Watch: The Friendship Paradox

If you learn better by watching, these videos are worth your time.

The Math That Proves Your Friends are More Popular Than You (Half as Interesting). A fast, fun walk-through of the friendship paradox.
The Friendship Paradox | This is in your recommended because it relates to the spread of diseases (Zach Star). How the paradox helps detect and slow epidemics.

Check Your Understanding

Five quick questions to see how much of this has stuck. Pick an answer and press Submit; you will see a short explanation either way.

Answer each question and press Submit to see your result right away.

1. In the 5-person example, what is the average number of friends?
Found a bug? Let us know!
2. Averaging per person instead of per friendship, what do you get in the example?
Found a bug? Let us know!
3. Using μ + σ²/μ with μ = 1.6 and σ² = 1.44, what do you get?
Found a bug? Let us know!
4. When does the friendship paradox disappear?
Found a bug? Let us know!
5. In the 2011 Facebook study, what share had fewer friends than their friends’ average?
Found a bug? Let us know!

Frequently Asked Questions

What is the friendship paradox?

The observation, described by Scott Feld in 1991, that on average your friends have more friends than you do, because well-connected people appear in many friend lists and are counted more often.

What is the formula for the friendship paradox?

The average number of friends of a randomly chosen friend is μ + σ²/μ, where μ is the mean number of friends and σ² the variance. It exceeds μ whenever friend counts vary.

Is the friendship paradox true on Facebook?

Yes. A 2011 study of 721 million active Facebook users found 92.7% had fewer friends than the average of their friends, and 83.6% fewer than their friends’ median.

How is the friendship paradox used?

To detect outbreaks early by monitoring friends of randomly chosen people, to target vaccination at well-connected people, and to understand social media comparisons.

Does the friendship paradox mean I am unpopular?

No. It holds for most people in any network with uneven connections; it reflects how averages are calculated, not your popularity.

Keep Exploring

Sources and Further Reading

Related to This Article

What people say about "Friendship Paradox Explained: Why Friends Have More Friends - Effortless Math"?

No one replied yet.

Leave a Reply