These can all be obtained using the area under the normal distribution (bell-shaped) curve.
Let z be the number of standard deviations (sigma) x is different from the mean
z = (x - mean)/sigma
Use a table of normal distribution values (better yet, cumulative distribution, the area under the curve):
P(z) = percentage smaller than or equal to z.
For a normal distribution ("Gaussian"), the fraction less than or equal to the mean, thus z=0,
is also the median, 50%, 0.50.
P(z=1) is a standard deviation above the mean, P(1)=0.84.
a) x=31, z=(31-28)/7=0.43. P(0.43) = 0.666 = 66.6% of the users pay less than 31
so 33.4% pay more.
b) x=39, z=(39-28)/7=1.57. P(1.57) = 0.941 = 94.1%
so 1-0.941= 0.059 = 5.9% of the users pay more than 39.
c) x=18, z=(18-28)/7=-1.43. P(-1.43) = 0.076 = 7.6% of the users pay less than 18.
d) Use z = (x - mean)/sigma, but get z from the distribution table for P(z),
P(z) = 81% is 0.86 sigma above the median, so we calculate the
x = mean - 0.90 sigma = 28+0.0.86*7 = 34.0.
Thus, 19% have interest costs of 34.0 or more.