How Do You Calculate Probability Between Z Scores Without Tables?

5 replies · 2K views · last reply
g.lemaitre
Messages
267
Reaction score
2

Homework Statement



I want to calculate the probability of a random sample falling between 2 z scores using the way real mathematicians do it not the fake way by resorting to tables. Ok, so the book outlines the equation below but says that it requires calculus which is beyond the scope of this course. I know calculus so let's do it.

Homework Equations



Two formulas for the normal probability density function: y = f(x) = e^(-(1/2)((x-μ)/σ)^2) / (σ√(2π)) for all real x, and P(a ≤ x ≤ b) = integral from a to b of f(x) dx.


z = (x - μ)/σ

The Attempt at a Solution



Let's say x = 21, μ = 14 and σ = 6

thus

(21 - 14)/6 = 1.16, according to the tables the probability of a random sample falling between the z scores 0 and 1.16 is .3770

Now, let's use the calc equation:

[tex]\frac{\exp\frac{1}{2}(\frac{21-14}{6})^2}{6 \sqrt{2\pi}}[/tex]

= .0336, not .3770, so I'm doing something wrong
 
Last edited by a moderator:
hold on, I'm forgetting what tags are used to enclose latex.
 
well [tex]is the way to enclose latex, it's not working on my computer.[/tex]
 
g.lemaitre said:

Homework Statement



I want to calculate the probability of a random sample falling between 2 z scores using the way real mathematicians do it not the fake way by resorting to tables. Ok, so the book outlines the equation below but says that it requires calculus which is beyond the scope of this course. I know calculus so let's do it.

Homework Equations



Two formulas for the normal probability density, y = f(x) = e^(-(1/2)((x-μ)/σ)^2) / (σ√(2π)) for all real x, and the probability P(a ≤ x ≤ b) = integral from a to b of f(x) dx.


z = (x - μ)/σ

The Attempt at a Solution



Let's say x = 21, μ = 14 and σ = 6

thus

(21 - 14)/6 = 1.16, according to the tables the probability of a random sample falling between the z scores 0 and 1.16 is .3770

Now, let's use the calc equation:

[tex]\frac{\exp\frac{1}{2}(\frac{21-14}{6})^2}{6 \sqrt{2\pi}[/tex]

= .0336, not .3770, so I'm doing something wrong

You need a second bracket } after the 2\pi, because without it you have \frac{}{ (no closure).

So you have written
[tex]\frac{\exp\frac{1}{2}(\frac{21-14}{6})^2}{6 \sqrt{2\pi}}[/tex]
and called this the calc equation. Well, it is nonsense! The actual probability that an N(0,1) random variable falls between 0 and 1.16 is
[tex]\int_0^{1.16} \frac{1}{\sqrt{2\pi}} \exp(-z^2/2)\, dz,[/tex] which is a non-elementary integral; that is, there is no finite formula to express an integral of the form
[tex]G(a) = \int_0^a \frac{1}{\sqrt{2\pi}} \exp(-z^2/2)\, dz,[/tex] for general values of 'a' (although exact values are available for some special values of 'a'). This is a *theorem*: it is impossible to express G(a) in finitely many elementary terms. It not that nobody has been smart enough to figure out a formula; it is a rigorously-proven fact that it is impossible to write such a formula. Basically, even if you write down a complicated formula taking 1 million pages to write out, it still won't represent G(a)! Not even 1 billion pages are enough. So, when we want to compute G(a) we must resort to the use of tables or numerical integration methods or approximate formulas. Don't scorn tables---they (or their modern equivalents) are necessary.

Anyway, using tables (or Maple, which I prefer) the answer for that integral is 0.3769755969, approximately.

RGV
 
when I did the calculations I got .204. Without the negative sign, since z is squared, i got .78

Screenshot of an online scientific calculator showing the expression (1/sqrt(2*(pi)))*e^(-1.16^2/2) with results 0.20357138829076 and 0.78181391023662, plus the formula (1/√(2π)) × e^(−1.16²/2) rendered above.
 
g.lemaitre said:
when I did the calculations I got .204. Without the negative sign, since z is squared, i got .78

Screenshot of an online scientific calculator evaluating (1/sqrt(2*pi))*e^(-1.16^2/2), with the result 0.20357138829076 shown below, and the standard normal density formula above it.

You are NOT computing the integral, which is what needs to be done. Of course we can calculate f(z) for any z, but that is not the point. We need to calculate the INTEGRAL of f(z) for z from 0 to 1.16, and that is a much different problem. It needs tables or equivalent.

RGV