Variance $sigma^2$ = sum$\frac{xi^2}{n}$ - (x-bar)^2. This "sum of squares over n minus square of mean" form is far superior to the definitional formula sum(xi - x-bar)^2/n for computation. For frequency data: $sigma^2$ = sum$\frac{fi*xi^2}{N}$ - (x-bar)^2. Remember: variance is always >= 0, equals 0 only when all observations are identical. Most JEE problems give sum(xi) and sum($xi^2$) and expect you to use this formula directly.