Learning Pages

Shared Birthdays


Home | Articles | Mathematics | Shared Birthdays

If there are 30 people taking part in a 'Zoom' meeting, what are the chances of at least two participants having the same birthday?

You might estimate that there is about an 8% chance of this happening because there is a 1 in 365 chance of any one person of having a birthday on a particular day. So, in a group of 30 people, the probability would be increased by 30 or 30/365.

The actual probability is 71%! The reason for this is that in our original estimate we were only comparing two people, but in a group of 30, there are actually 435 comparisons or pairings.

Let's see how this works with a smaller group of people (Alex, Bertie, Charlie and Dee) who were all born on a weekday:

If we were to use our original method for estimating the probability that at least two of them were born on the same weekday, then we would come up with an answer of 80% (1 in 5 chance of any two people being born on the same weekday multiplied by 4).

However, the probability is greater than this because everyone should be compared to everyone else. With a group of four people, there are six possible pairings:

Alex Bertie
Alex Charlie
Alex Dee
Bertie Charlie
Bertie Dee
Charlie Dee

Table 1: Pairings for four people

While it is possible to use this method to work out the number of pairings for any group size, it does become a bit arduous when dealing with large numbers. Luckily, there is a formula you can use for calculating the number of pairings. The number of possible pairings in a group of n people is:

n(n-1)/2.

The pairings don't have to be in the order shown in Table 1. They can be in any order as long as they include all the pairings. Table 2 shows the six pairings in a different order and a slightly different format:

BertieAlex
CharlieBertie
Alex
DeeCharlie
Bertie
Alex

Table 2: A different order and format

So, what is the actual probability of at least two members of this group being born on the same weekday? A tree diagram is one way of working out the probability (see Figure 1).

Figure 1: A Tree Diagram

In this tree diagram, Bernie's birthday is compared with Alex's. There is a 1 in 5 chance of them being the same and a 4 in 5 chance of them being different. The 'Yes' branch doesn't continue because the problem is to determine the probability of at least two people having the same birthday. So, if probability takes us down a 'Yes' branch, there's no need to go any further.

Charlie's birthday is then compared with Bertie's and Alex's. Then, Dee's birthday is compared Charlie's, Bertie's and Alex's.

Tree diagrams can get a bit complicated, so I am going to look at the same question using tables. I found that using tables helped me understand tree diagrams better.

There are a few things we need to bear in mind before we start:

Each pairing has a 1 in 5 chance of both people being born on the same weekday as shown in Table 3.

Match
Bertie Alex1/5
Charlie Bertie 1/5
Alex 1/5
Dee Charlie 1/5
Bertie 1/5
Alex 1/5

Table 3: Pairings and the chances of a match

If we add the probabilities together, we find that there is:

MatchTotal
Bertie Alex 1/5 1/5
Charlie Bertie 1/5 2/5
Alex 1/5
Dee Charlie 1/5 3/5
Bertie 1/5
Alex 1/5

Table 4: Addition of probabilities

We can calculate the 'no-match' probabilities of by subtracting the 'match' probabilities from one as shown in Table 5.

Match Total No Match
Bertie Alex 1/5 1/5 4/5
Charlie Bertie 1/5 2/5 3/5
Alex 1/5
Dee Charlie 1/5 3/5 2/5
Bertie 1/5
Alex 1/5

Table 5: No-match probabilities

Notice, both in the tree diagram and Table 5, that for n people the 'no-match' probabilities start at n-1/n and that the dividend (the number being divided) is reduced by 1 for every subsequent set of comparisons.

We get the overall 'no-match' probability by multiplying 4/5 x 3/5 x 2/5. This gives an answer of 24/125 or 19%. The overall 'match' probability is 81% (100% - 19%).

Using tree diagrams or tables can get quite difficult for larger numbers like 30 people and 365 days. There are, of course, mathematical formulae for solving this problem — see the Birthday Problem on Wikepedia.

Computer programs can also find the answers by modelling the processes described above. I have written a Shared Events web app which you can use to solve the Birthday Problems and many other variations.

The following table summarises the data for groups of 2-30 people:

# People # Pairings No Match Match
2 1 99.7%0.3%
3 3 99.2%0.8%
46 98.4% 1.6%
5 10 97.3% 2.7%
6 15 96% 4%
7 21 94.4% 5.6%
8 28 92.6% 7.4%
9 36 90.5%9.5%
10 45 88.3% 11.7%
11 55 85.9% 14.1%
12 66 83.3% 16.7%
13 78 80.6% 19.4%
14 91 77.7% 22.3%
15 105 74.7% 25.3%
16 120 71.6% 28.4%
17 13668.5%31.5%
18 153 65.3% 34.7%
19 171 62.1% 37.9%
20 190 58.9% 41.1%
21 210 55.6% 44.4%
22 231 52.4% 47.6%
23 253 49.3% 50.7%
24 276 46.2% 53.8%
25 300 43.1% 56.9%
26 325 40.2% 59.8%
27 351 37.3% 62.7%
28 378 34.6% 65.4%
29 406 31.9% 68.1%
30 435 29% 71%

Table 6: Data for 2-30 people

These methods don't take leap years into account or the fact that birthdays are not spread evenly throughout the year.

Written with the help of an explanation by Elouise Wills

Become a member to get early access to new articles.

Donate using Liberapay

Support the Learning Pages project