4.9 The normal distribution

4.9 The normal distribution

IB Maths: Analysis and Approaches HL: Revision notes

Studying this course? Keep your progress in one place.

Key facts

  • X∼N(μ,σ2)X\sim N(\mu,\sigma^2) is symmetrical about μ\mu, with total area 1.
  • About 68%, 95% and 99.7% of values lie within 1, 2 and 3 standard deviations of the mean.
  • Use normal cdf for probabilities and inverse normal (area to the left) for values.
  • P(X=a)=0P(X=a)=0, so << and ≤\le give the same probability.
  • Write probability statements, not calculator syntax, and round in the direction the context needs.

The normal curve

A symmetrical, bell-shaped curve N(μ,σ2)N(\mu,\sigma^2) for quantities affected by many small influences.

A continuous random variable XX with a normal distribution is written X∼N(μ,σ2)X\sim N(\mu,\sigma^2): μ\mu is the mean and σ2\sigma^2 the variance, so σ\sigma is the standard deviation.

The normal curve is bell-shaped and symmetrical about x=μx=\mu, so mean, median and mode are equal. The area under it is 1, and a larger σ\sigma gives a wider, flatter curve.

It occurs naturally for heights, masses and measurement errors. Because XX is continuous, P(X=a)=0P(X=a)=0.

Picture the curve and shade the area you want before using the calculator.

1001101201301401501601701801902000.010.020.030.040.050.060.07xynormal curve
The normal curve. Change the mean and the standard deviation.

For X∼N(60,102)X\sim N(60,10^2), what is P(X=60)P(X=60)?

The 68–95–99.7 rule

Roughly 68%, 95% and 99.7% of values lie within one, two and three standard deviations of the mean.

Combine this empirical rule with symmetry to estimate tail areas without a calculator. 68% lie within one standard deviation, so 32% lie outside it, and 16% lie above μ+σ\mu+\sigma. Similarly 2.5% lie above μ+2σ\mu+2\sigma.

For N(150,122)N(150,12^2), about 95% of apples weigh between 126 g and 174 g. A value more than 3 standard deviations from the mean is very unusual, a useful check on a model.

1101201301401501601701801900.0050.010.0150.020.0250.030.0350.04xy126138150162174normal curve
N(150, 12²): the shaded area within one standard deviation is about 68%
  • Within μ ± σabout 68%
  • Within μ ± 2σabout 95%
  • Within μ ± 3σabout 99.7%

For N(150,122)N(150,12^2), roughly what percentage of values lie above 174?

Normal probabilities with technology

Use normal cdf with a lower bound, an upper bound, μ\mu and σ\sigma, and write the probability statement first.

Use the GDC's normal cdf with a lower bound, an upper bound, μ\mu and σ\sigma. Write the probability statement first, for example P(140<X<165)=0.692P(140<X<165)=0.692.

Expected numbers work as for any probability: out of 400 batteries, each with probability 0.05990.0599 of failing before 550 hours, about 400×0.0599=24.0400\times0.0599=24.0 are expected to fail.

−4−3−2−112340.10.20.30.4xylower boundupper boundnormal curve
Illustrative standard normal curve: the shaded area between a lower and an upper bound is the probability that normal cdf returns.

P(X<a)P(X<a)

Lower bound:
−1099-10^{99}
Upper bound:
aa

P(X>a)P(X>a)

Lower bound:
aa
Upper bound:
109910^{99}

P(a<X<b)P(a<X<b)

Lower bound:
aa
Upper bound:
bb

Worked example

T∼N(42,52)T\sim N(42,5^2). Find P(T<35)P(T<35) and P(T>49)P(T>49).

X∼N(60,102)X\sim N(60,10^2). What is P(X<60)P(X<60)?

Inverse normal calculations

Use inverse normal with the area to the left, so convert 'greater than' statements to 1−p1-p.

When you know a probability and want the value, use inverse normal. Most calculators need the area to the left:

  • '10% of journeys take longer than kk': P(T<k)=0.9P(T<k)=0.9, so k=48.4k=48.4.
  • 'exceeded by 99% of bags': P(M<m)=0.01P(M<m)=0.01, so m=993.4m=993.4 g.
  • Quartiles: Q1Q_1 has area 0.25 to the left, Q3Q_3 has 0.75, so IQR=Q3−Q1\text{IQR}=Q_3-Q_1.

μ\mu and σ\sigma are always given. Round in the direction the context demands: a guarantee affecting at most 2% of batteries uses g=527g=527, not 528.

25303540455055600.020.040.060.08xy42k = 48.4T ~ N(42, 5²)
T ~ N(42, 5²): the shaded upper tail has area 0.1, starting at k = 48.4

For P(T>k)=0.05P(T>k)=0.05, what area to the left do you enter into inverse normal?

Normal with other probability

Use it as the pp of a binomial model, or in a conditional probability.

Exam questions often use a normal probability as an input to another calculation.

  • Binomial: if each bag is underweight with probability p=0.0668p=0.0668, the number of underweight bags in a box of 12 is B(12,p)B(12,p), so P(at least one)=1−(1−p)12=0.564P(\text{at least one})=1-(1-p)^{12}=0.564.
  • Conditional probability: P(M>1020∣M≥1000)=P(M>1020)P(M≥1000)=0.170P(M>1020\mid M\ge1000)=\frac{P(M>1020)}{P(M\ge1000)}=0.170.

The event 'more than 1020 g' lies inside 'at least 1000 g', so the intersection is just P(M>1020)P(M>1020). Keep unrounded probabilities in memory for the next stage.

  1. 1

    Normal probability

    find p from normal cdf

  2. 2

    Second model

    binomial B(n, p) or conditional probability

  3. 3

    Final answer

    round only at the end

Normal probability as an input

Why does P(M>1020∣M≥1000)P(M>1020\mid M\ge1000) have P(M>1020)P(M>1020) on the top of the fraction?

Try an exam question

The time TT minutes taken by a train journey is normally distributed, T∼N(42,52)T\sim N(42,5^2). (a) Find P(T<35)P(T<35). (b) Find kk such that P(T>k)=0.1P(T>k)=0.1. (c) Without further calculation, write down P(T>49)P(T>49).

[4 marks]

That's the notes covered.

Carry on to the next subtopic.

Try a question

Estimate the percentage of apples with a mass between 126 g and 174 g.

1 mark, marked against the mark scheme

Revision notes