Probability as a share: how many cases fit

34 min · 5 drills

In the textbook: Ch. 5.1 Grundbegriffe

A scratch card has eight covered fields, with a star under two of them. You scratch off two fields. It seems the chance of uncovering both stars is 2 in 8, but in fact it is seven times smaller. To avoid that mistake you count all equally likely cases and the ones that fit; with thousands of cases you do not list them but count the ways of choosing.

A probability is a share: which part of all equally likely cases fits the question. Counting cases is where the exam tasks on lotteries, faulty parts and matching days begin.

The task. A scratch card has 8 fields, with a star under two of them. A buyer scratches off two fields and wins if both stars appear. The organiser has sold 560 such cards.

Trusting the 2-in-8 chance, a quarter of the cards would win, about 140.

How many winners should the organiser really expect?

Count the pairs of fields: 8 choices for the first, 7 for the second, and each pair counted twice, once per order:

8⋅72=28 pairs

One pair of 28 uncovers both stars, so expect about 560/28=20 winners, seven times fewer than 140.

Check yourself. A task at exam level: 3 balls are drawn from an urn with 10 numbered balls. Whoever named exactly these 3 numbers wins, the order does not matter. 600 people each make one random guess. How many winners are there on average?

Outcome, event and probability as a share

Number the fields of the card from 1 to 8; the stars lie under fields 3 and 6. A buyer may scratch fields 1 and 2, or 1 and 3, or 3 and 6, and so on. Each such possibility is an outcome (German Ergebnis), like a face of the die in the first article of the chapter, only here an outcome is a pair of fields. The buyer does not know where the stars are, so no pair of fields comes up more often than another. Such outcomes are called equally likely.

A question about the card fits some outcomes and not others:

  • “the card wins” fits one outcome, the pair 3 and 6;
  • “at least one star is uncovered” fits all pairs that contain field 3 or field 6.

The part of the outcomes that the question asks about is called an event (German Ereignis). Events are written with capital letters: call the win G.

Imagine very many cards sold. Every pair of fields is scratched equally often, so the share of winning cards equals the share that the pair 3 and 6 has among all pairs. This share is called the probability of the event. It is written P(G), from probability. For any event E with equally likely outcomes:

P(E)=number of outcomes in Enumber of all outcomes

A probability lies between 0 and 1. Zero means the event holds no outcome at all, 1 means it holds all outcomes.

The formula holds only if the outcomes are equally likely. The 2 in 8 from the start is the chance that a star lies under one chosen field. Winning needs both stars at once, so an outcome here is not a field but a pair of fields. The pairs have to be counted.

Counting selections: with and without order

To count the pairs, write out every way of scratching off two fields one after the other. In the table the row is the field scratched first, the column the field scratched second. The same field cannot be scratched twice, so the cells on the diagonal stay empty.

56 ordered pairs

Pairs of fields on the card: row the first field scratched, column the second; stars under fields 3 and 61234567812345678–????????–????????–????????–????????–????????–????????–????????–
Step 1 of 4Ordered pairs

Any of the 8 fields can be scratched first. Any of the 7 remaining can be scratched second.

Each first field goes with 7 second ones, so there are 8⋅7=56 pairs. These are all cells of the table except the diagonal.

The same idea works for any number of fields. Let k items be chosen from n.

  • With order (a combination lock, places 1, 2, 3 in a ranking): n options for the first place, n−1 for the second, and so on, k factors: n⋅(n−1)⋯(n−k+1).
  • Without order (scratching fields, choosing a committee, a lottery guess): each selection appears in the ordered count as often as its k elements can be arranged.

Anna, Ben and Vera can be lined up in 6 ways: ABV, AVB, BAV, BVA, VAB, VBA. There are 3 options for the first place, 2 for the second, 1 for the third: 3⋅2⋅1=6. In the same way k elements can be arranged in k⋅(k−1)⋯2⋅1 ways. This number is called the factorial (German Fakultät) and is written k!, here 3!=6. A pair of fields can be arranged in 2!=2 ways, which is why we divided by 2 on the card.

The number of selections without order is called the binomial coefficient, read “n choose k” (German “n über k”):

(nk)=n⋅(n−1)⋯(n−k+1)k!,(103)=10⋅9⋅83⋅2⋅1=7206=120

The top three of 10 contestants can be formed in 720 ways, a committee of three people in only 120. In a committee “Anna, Ben, Vera” and “Vera, Anna, Ben” are the same.

On the formula sheet the formula is written differently: (nk)=n!k!(n−k)!. It is the same number. In 10!7! all factors from 7 down to 1 cancel, and 10⋅9⋅8 is left. On a calculator it is the nCr key.

The complement: together always 1

Back to the card. Let X be the number of stars among the two scratched fields. Such a number has two words of its own:

  • a number that depends on chance is called a random variable (German: Zufallsgröße);
  • the table of its values with their probabilities is called its distribution (German: Verteilung), more in the section “Random variable and its distribution”.

Sort the 28 pairs by the value of X:

  • X=0: both fields from the 6 empty ones, (62)=6⋅52=15 pairs;
  • X=2: both stars, 1 pair;
  • X=1: one of the 2 star fields and one of the 6 empty ones, 2⋅6=12 pairs.
x012
P(X=x)15281228128

Every pair landed in exactly one column; none is missing and none is counted twice. So 15+12+1=28, and the probabilities of all values add up to 1: 2828=1.

Take the event A “at least one star is uncovered”, that is X=1 or X=2. All other outcomes form the complement A′ (German Gegenereignis, “not A”, as in the article on events): “no star”, X=0. A and A′ share no outcomes and together give all outcomes, so their probabilities add up to 1:

P(A′)=1−P(A)

So P(A)=1−1528=1328=0.4643. Check with the table of the number of stars above: 1228+128=1328. You go through the complement when it is easier to count: here it is one value, X=0, against two, X=1 and X=2. Pairs that come up in tasks:

EventComplement
at least one (“mindestens einer”)none (“keiner”)
all on different daysat least two on the same day
at most 2 (“höchstens 2”)at least 3 (“mindestens 3”)

How to compute “A and B” and “A or B” for two different events is shown in the section “And, or, not” of the chapter's first article.

Your turn1 of 5

A card has 10 fields, with a star under 3 of them. The buyer scratches 2 fields. What is the probability of uncovering at least one star?

A probability table with one gap

That the probabilities of all values add up to 1 was a check for the scratch card. In the first exam task it is the only way to find the missing number:

Exam-style task

Bei der Endkontrolle einer Charge werden zufällig vier Teile geprüft; N ist die Anzahl der fehlerhaften Teile unter ihnen. Die Verteilung von N ist teilweise bekannt:

N01234P(N=⋅)0.210?0.2400.3150.170

Bestimmen Sie den Erwartungswert E(N).

  1. a)2.772
  2. b)2.611
  3. c)2.170
  4. d)6.580
  5. e)2.380

In the probability table of this task one number is replaced by a question mark. The expected value that is asked for needs every number in the row, the missing one included. Such a table is built like the star table of the card: each outcome gives exactly one value of N (the number of faulty parts), so the probabilities of all values add up to 1. The missing value is 1 minus the sum of the others. We first practise filling in the gap on questions about a probability, and then solve the task from the paper in example 3.

When the task asks for a probability such as P(N≤1), write down which values belong to the event before adding. The wording decides about the boundary value:

WordingEventValues for N from 0 to 4
“höchstens 1” (at most 1)N≤10, 1
“kleiner als 2”, “weniger als 2” (less than 2)N<20, 1
“mindestens 2” (at least 2)N≥22, 3, 4
“größer als 2”, “mehr als 2” (more than 2)N>23, 4
“von 1 bis einschließlich 3” (from 1 to 3 inclusive)1≤N≤31, 2, 3

An earlier batch, the same check of four parts: P(N=0) is missing, P(N=1)=0.250, P(N=2)=0.135, P(N=3)=0.205, P(N=4)=0.220. What is the probability that at most one part is faulty?

N ≤ 1: values 0 and 1

Distribution of the number N of faulty parts with a gap at N = 001234probability?0.2500.1350.2050.220
Step 1 of 4Choose the values

“At most 1” means N≤1: the values 0 and 1. The 1 itself belongs to it.

X is the number of complaints an online shop receives in a day: P(X=0)=0.30, P(X=1) unknown, P(X=2)=0.20, P(X=3)=0.15, P(X=4)=0.10.

What is P(X<3)?

“Less than 3” means the values 0, 1 and 2. The gap at X=1 belongs to the event.

The complement X≥3 covers the values 3 and 4, which have no gap. 0.15+0.10=0.25.

P(X<3)=1−0.25=0.75.

The same answer through the gap: P(X=1)=1−0.75=0.25 and 0.30+0.25+0.20=0.75. The complement is quicker when it has fewer values.

This is the task from the sheet above. The gap is at N=1, and the question asks for the expected value E(N): the mean that N would have over very many batches. Each value counts in it as strongly as it is likely. So each value is multiplied by its probability and the products are added, E(N)=∑n⋅P(N=n); the sign ∑ (sigma) means “sum over all values”. More in the section “Expected value and variance: centre and spread”.

The known values: 0.210+0.240+0.315+0.170=0.935, so P(N=1)=1−0.935=0.065.

E(N)=0⋅0.210+1⋅0.065+2⋅0.240+3⋅0.315+4⋅0.170

=0+0.065+0.480+0.945+0.680=2.170. Option (c).

The completed row adds up to 0.210+0.065+0.240+0.315+0.170=1. And 2.170 lies between 0 and 4, as every mean of a quantity that only takes values from 0 to 4 must.

A “more than” question: P(X > 1). X is the number of sick days of an employee in a month: P(X=0)=0.35, P(X=1)=0.25, P(X=2) unknown, P(X=3)=0.10, P(X=4)=0.05. P(X>1) is wanted. The gap: 1−(0.35+0.25+0.10+0.05)=1−0.75=0.25. Write down the values “more than 1” and add them.

Your turn2 of 5

A taxi company records the number Y of passengers per ride: P(Y=1)=0.38, P(Y=2) unknown, P(Y=3)=0.32, P(Y=4)=0.16, P(Y=5)=0.06. What is the probability that at most 3 passengers ride?

Traps. This is how the wrong options of the exam task come about, together with the typical mistakes in questions about a probability:

  • Gap filled in wrongly, option (e), 2.380: when filling in, P(N=0)=0.210 was not subtracted, and the gap became 1−0.725=0.275 instead of 0.065. Check: after filling in, the whole row must add up to exactly 1.
  • Computing with the squares, option (d), 6.580: this is ∑n2⋅P(N=n), a different quantity that is needed only for the variance. Check: an expected value lies between the smallest and the largest value, here between 0 and 4, so 6.580 drops out at once.
  • Boundary shifted, in questions like example 1: “at most 1” read as “less than 1” and only N=0 taken, 0.190 there. Check: ask whether the boundary value itself belongs to the event; with “at most” and “at least” it does.
  • Gap taken as 0, also in example 1: only the printed numbers of the event added, that is P(N=1)=0.250 alone. The gap lies inside the event and is found first.

The formula sheet has P(A′)=1−P(A). That the probabilities of all values of a random variable add up to 1 you have to remember yourself.

How many winners in a draw: selections without order

The task with a gap in the probability table needed no counting of selections. The second task goes back to the scratch card from the start: one player's chance again comes from the number of selections, and the question is how many winners there will be:

Exam-style task

Beim Schulfest steckt in 14 Kuverts je ein Zettel; auf genau 4 davon steht «Hauptpreis». Wer bezahlt, darf 4 Kuverts öffnen und erhält den Hauptpreis nur, wenn er genau diese 4 erwischt. Insgesamt öffnen 12012 Gäste ihre Kuverts. Wie viele Hauptpreise gehen im Mittel weg?

  1. a)9
  2. b)12
  3. c)33
  4. d)11
  5. e)10

The words “exactly these” and “on average” mark this task. One selection out of all wins, and the order within it does not matter. What is asked is not one participant's chance but the number of winners. This is the scratch card: one participant's chance is 1 divided by the number of selections, and the expected number of winners is the number of participants times this chance.

four envelopes in order: 24024

Envelopes: open 4 of 14; number of selections, one guest's chance and expected winnershownumberwith orderorders of one selectionselectionschance of a guestwinners14 · 13 · 12 · 1124024
Step 1 of 4Count with order

Of n=14 envelopes, k=4 are opened. With order: 14⋅13⋅12⋅11=24024. There are four factors, as many as envelopes are opened.

The door of an escape room opens when exactly the 4 right ones of 11 levers are pulled, the order does not matter. Every group pulls 4 levers at random. The operator wants the door to open on the first try no more than 3 times a month on average.

How many groups can he accept at most?

(114)=11⋅10⋅9⋅84⋅3⋅2⋅1=792024=330. One selection opens the door: P=1330.

Let there be N groups. On average the door opens N⋅1330 times, and that may be at most 3: N330≤3.

Multiply both sides by 330: N≤3⋅330=990 groups.

The question “with what probability does at least one participant win” is solved through independence: it is covered in the method “1 − (1 − p)ⁿ: at least once in n tries”.

Your turn3 of 5

A league has 11 teams, and exactly 3 are relegated at the end of the season. 1980 newspaper readers each name 3 teams at random, the order does not matter. How many readers guess exactly right on average?

Traps. The wrong options of the task above come about like this:

  • Wrong index in (nk), option (c), 33: (143)=364 taken, one factor short, and 12012/364=33. Check: the numerator has as many factors as envelopes are opened, here 4.
  • Wrong number of items, option (a), 9: 15 envelopes taken instead of 14, (154)=1365 and 12012/1365=8.8, about 9. Check: n is the number of all envelopes in the task, k the number opened.

The formula sheet has n! and (nk)=n!k!(n−k)!. That the winners on average are the number of participants times one participant's chance you have to remember yourself.

Parcels over days: all on different days or at least two on the same one

In the draw the order did not matter, and the selections were counted without it. In the third task the parcels can be told apart, and the counting has to keep the order:

Exam-style task

Eine Kundin hat bei einem Lieferdienst 5 Pakete bestellt. Jedes wird an einem der sieben Tage der nächsten Woche zugestellt; der Zustelltag eines Pakets ändert nichts an dem der übrigen, und kein Tag wird bevorzugt. Mit welcher Wahrscheinlichkeit kommen alle Pakete an unterschiedlichen Tagen an? Runden Sie auf zwei Nachkommastellen.

  1. a)0.97
  2. b)0.28
  3. c)0.85
  4. d)0.15
  5. e)0.25

Several parcels, people or appointments fall independently on one of a few days. The question is “all on different days” or “at least two on the same day”. The days play the part of the card's fields, with one difference:

  • all cases: two parcels can come on the same day, so each of the n parcels has all d days, dn equally likely lists “parcel 1 on Wednesday, parcel 2 on Friday, …”;
  • all different: like the ordered pairs on the card, the first parcel has d days, the second d−1, and so on.

The parcels are numbered, like the first and the second scratched field. If you only write down how many parcels come on each day, these cases are not equally likely, and the share comes out wrong. With two parcels and two days you see it at once: “both on Monday” happens in one way out of four, “one on each day” in two, the first on Monday or the second on Monday.

all lists of days: 16807

Five parcels and seven days: how many days each parcel can choose from, in all cases and when all days differall casesall differentparcel 1parcel 2parcel 3parcel 4parcel 5product7777716807
Step 1 of 3All cases

There are d=7 days and n=5 parcels. Each parcel has 7 options, and a day may repeat.

75=16807 equally likely lists.

Suppose four friends were born in random months, independently of each other, all months equally likely.

What is the probability that at least two have their birthday in the same month?

The “days” here are d=12 months, people n=4. 124=20736 lists.

All in different months: 12⋅11⋅10⋅9=11880. P(all different)=1188020736=0.5729.

“At least two in the same month” happens exactly when not all are different. 1−0.5729=0.4271.

Find: a shared day. Three friends each choose one day in June independently, all 30 days equally likely. All cases: 303=27000, all on different days: 30⋅29⋅28=24360. Find the probability that at least two chose the same day.

Your turn4 of 5

Four clients book a consultation independently of each other, each on one of the five working days of the week, all days equally likely. What is the probability that two or more bookings fall on at least one day?

Traps. The wrong options of the task above come about like this:

  • 365 days instead of 7, option (a), 0.97: this is the birthday problem over a year, 365⋅364⋅363⋅362⋅3613655=0.97. Check: d is the number of days in the task: a week, a month or a year.
  • The wrong event, option (c), 0.85: this is 1−0.15, the probability of “at least two on the same day”. The question was “all on different days”, so subtracting from 1 is wrong here. Check: reread the question before the last step.

This formula is not on the formula sheet. Remember: all cases dn, all different d⋅(d−1)⋯(d−n+1).

Arrangements with repeats and drawing from two groups without replacement

The three task types above cover everything about counting cases that has been asked so far. Two more forms grow out of the same rules. Neither of the two forms below was set in any of the 13 exams 2023–2026, but both are possible.

Arrangements with repeated letters

The six cards of the word BANANE are shuffled and laid in a row.

How many different words can come out?

  1. If all cards were different, there would be 6!=720 orders.
  2. If the two A swap places, the word is the same: each word is counted 2!=2 times. Because of the two N, twice as often again.
  3. There are 7202!⋅2!=180 words, and the word BANANE itself comes out with probability 1180=0.0056.

In general, for n items with groups of identical items of k1, k2, … each: n!k1!⋅k2!⋯.

Your turn5 of 5

A flag consists of 8 horizontal stripes: 4 red, 2 green and 2 white. How many different flags can be made?

Drawing from two groups without replacement

Say a batch has 12 parts, 4 of them faulty. An inspector picks 3 parts at random.

What is the probability that exactly one of them is faulty?

  1. All selections of 3 parts are equally likely: (123)=12⋅11⋅103⋅2⋅1=220.
  2. A favourable selection has two parts: one faulty of 4, (41)=4 ways, and two good ones of 8, (82)=28 ways.
  3. Any first part goes with any second part, so the numbers of ways are multiplied: 4⋅28=112.
  4. P=112220=0.5091.

In short: share, selections and traps

The topic in a minute

  • Probability with equally likely outcomes: the number of outcomes in the event divided by the number of all outcomes. Trap: an outcome must be what is chosen as a whole; on the card a pair of fields, not a field.
  • Selections of k from n: with order n⋅(n−1)⋯(n−k+1), without order (nk), the same divided by k!. Trap: counting with order where it does not matter.
  • Complement: P(A′)=1−P(A); the probabilities of all values of a random variable add up to 1. Trap: taking “none” when “at least one” is asked, and the other way round.
  • One missing value in the table

    A probability table with a question mark; the question says 'at most', 'more than' or 'from … to … inclusive'.
    P(gap)=1−sum of the others

    Trap: shifting the boundary or taking the gap as 0.

  • Prize draws and combinations

    k are chosen from n and one selection wins; asked are the winners on average, the largest number of participants or 'at least one'.
    winners=N⋅1(nk)

    Trap: wrong index in (nk) or wrong n; counting with order.

  • Same days, all different

    n parcels or people fall independently on one of d days; asked is 'all different' or 'at least two on the same day'.
    P(all different)=d⋅(d−1)⋯(d−n+1)dn

    Trap: counting with 365 days or giving the complement.

Sometimes the question is not about everyone but only about a part: those who have already bought, or those with a positive test. That is conditional probability, the next article: “Conditional probability”.

Cards to memorise

Tap a card to see the formula. All topic cards

Formula

Each trial succeeds with probability p, q=1−p. What is the probability that the first success comes by the k-th trial at the latest?

Formula

How likely is exactly one of two events A and B?

Now practise it

Probability as a share: how many cases fit

Trainer · Probability

Practise
Next: Conditional probability: the share within a group
0/5

Was this article helpful?