Given the area between two values on a normal distribution, how would you use the inverse normal?

177 Views Asked by At

Given that $X$~$N(10, 0.5^2)$, find $c$ such that: $$P(12.6-2c < x < 11.9+c)=0.2$$

I can solve simpler problems of this sort, for example I can solve:

$P(10-c < x < 10+c)=0.2$ but only because the two end areas are equal. I wanted to know if there was an easy way to calculate $c$ in my first example.

I am aware that you can integrate the equation of the normal curve using the error function, but that is way beyond my understanding at the moment.

1

There are 1 best solutions below

3
On BEST ANSWER

This problem is easier than it appears at first glance.

With a little algebra, $$12.6-2c < x < 11.9+c$$ if and only if $$5.2-4c < \frac{x-10}{0.5} < 3.8+2c$$ and $(x-10)/0.5$ has a $N(0,1)$ distribution. So we want to solve $$\Phi(3.8+2c)-\Phi(5.2-4c) = 0.2$$ for $c$, where $\Phi(x)$ is the cumulative standard normal distribution.

At $c=0$, the left-hand side of the equation is essentially $1-1=0$, while for large $c$ it becomes $1-0=1$, so we expect a positive root. But for $c>0$, $\Phi(3.8+2c) \approx 1$, so we can replace our equation with $$1-\Phi(5.2-4c) = 0.2$$ I'm guessing you can probably take it from here.