Yves S. answered 01/16/20
Statistics made easy for undergrad, grad and MBA students
Preity,
The problem at hand here is to determine the probability of one or more successful outcomes out of a given number of trials, and given a probability of one success happening out of one individual trial. These exercises follow the binomial distribution (see more below).
So let's look at your first example:
We are looking at 5 business days (number of trials), and wondering if the store will gross more than $850 (defined as a success) for at least 3 days (number of successes), given that 50% of the time the store does (probability of an individual success).
Each day there are two possible outcomes (success/failure, or 1/0); so after two days you will have 4 possibilities (1,1); (1,0); (0,0); (0,1); after three days you will have 8 possibilities, 16 after four days, and ultimately 32 possible combinations after 5 days (0,0,0,0,0), (0,0,0,0,1) all the way to (1,1,1,1,1).
The total number of combinations possible can be written as 2^5 = 32.
Since we want to count the number of combinations where we have at least 3 successes (for example: 1,1,0,1,0), we will count the number of combinations where we have 3, 4 and 5 successes.
If you write all possible combinations, you will see there are:
- one way to have 5 successes (1,1,1,1,1)
- five ways to have 4 successes (0,1,1,1,1);(1,0,1,1,1);(1,1,0,1,1); (1,1,1,0,1); (1,1,1,1,0)
- ten ways to have 3 successes (1,1,1,0,0); (1,1,0,0,1); etc.
So the number of ways to combine three or more successes out of five trials is 1+5+10 = 16
Since the probability of an individual event is 50%, the probability of having three or more successes out of five trials is therefore the number of possible successes divided by the number of possible outcomes, i.,e. 16/32 = 0.50. In math, you would write this as [C(5,5)+C(5,4)+C(5,3)]/2^5 = 0.50
In Excel you can use the COMBIN() function:
(COMBIN(5,5)+COMBIN(5,4)+COMBIN(5,3))/2^5 = 0.5
The C(n,k) or COMBIN(n,k) works well for problems where there is equal probability of success and failure, i.e. p=0.50. For different probabilities of success, you must refer to the binomial distribution.
The probability P(k) of k successes in n events for a binomial distribution with probability of success p can be expressed as: P(k) = p^k*(1-p)^(n-k)*C(n,k)
Fortunately, you can use (hopefully) software such as Excel to solve this using the BINOM.DIST() function:
P(k) = BINOM.DIST(k,n,p,FALSE) returns the probability of k successes;
and
CDF = SUM(0 to k) of (P(k)) = BINOM.DIST(k,n,p,TRUE) returns the sum of all probabilities of up to k successes out of n trials, or the cumulative probability for k successes.
in our case:
P(>=3,5) =1-P(<=2,5) = 1-BINOM.DIST(3,5,0.5,1) = 0.5
Similarly, you get:
P(>=6,10) = 1-P(<=5,10) = 1-BINOM.DIST(5,10,0.5,1) = 0.377
P(<5,10) = P(<=4,10) = BINOM.DIST(4,10,0.5,1) = 0.377 (same as above! - check out Pascal's triangle to learn why!)
P(<6,20) = P(<=5,20) = BINOM.DIST(5,20,0.5,1) = 0.021
Now, you realize there is about 2% chance of five successes (store making more than $850) out of twenty days. If, indeed, you were to see that happening, you should suspect that the initial probability of making more than $850 per day is less than 50%; a typical threshold for suspecting foul play is 0.05 (this is called the significance level).
P(>17,20) = 1-P(<=17,20) = 1-BINOM.DIST(17,20,0.5,1) = 0.000
This means there is practically no chance you will make more than $850 for more than 17 days out of 20, if the probability of an individual success is 50%. You should suspect the probability of making more than $850 per day to be higher than 50%.
Your second set of problems can be solved in a similar fashion, adjusting the initial probabilities.
Hope this helps!
Yves