We have the n observations x1,x2,…,xn with two given relations
∑i=1n(xi−a)=nand∑i=1n(xi−a)2=na,
where n>1,a>1.
First we extract the arithmetic mean of the observations. Writing the first relation explicitly,
∑i=1n(xi−a)=∑i=1nxi−∑i=1na=∑i=1nxi−na=n.
So
∑i=1nxi=na+n=n(a+1).
Dividing by the number of observations gives the mean
x=n1∑i=1nxi=nn(a+1)=a+1.
Now we need the standard deviation. By definition, the variance is
σ2=n1∑i=1n(xi−x)2.
Because x=a+1, we must convert the given sum
∑i=1n(xi−a)2 to the required sum ∑i=1n(xi−(a+1))2.
For each term we note the algebraic identity
(xi−(a+1))2=[(xi−a)−1]2=(xi−a)2−2(xi−a)+1.
Summing this expression from i=1 to n yields
∑i=1n(xi−(a+1))2=∑i=1n(xi−a)2−2∑i=1n(xi−a)+∑i=1n1.
Substituting the given values step by step:
∑i=1n(xi−a)2=na,
∑i=1n(xi−a)=n,
∑i=1n1=n.
Hence
∑i=1n(xi−(a+1))2=na−2n+n=n(a−1).
So the variance is
σ2=n1∑i=1n(xi−x)2=n1[n(a−1)]=a−1.
Taking the positive square root (because standard deviation is non-negative) gives
σ=σ2=a−1.
Hence, the correct answer is Option D.