I have a file containing the times taken to do something. The data is fits a normal distribution. I am trying to find the percent of people that take longer than 1800 seconds. So I find the z-score By symmetry I take -1.67 and from a table I look up -1.67 and find 0.0475 which gives me 4.75%

Am I good up to this point?

If I haven't made a mistake yet, what's confusing me is that my statistical analysis program (StatsGraphics Centurion) makes me think it should be about 5.26%. This is because the program created a class with the interval (1815,1908] which has the relative frequency of 0.0526. Shouldn't this number be smaller than 0.0475 because the interval (1815,1908] is smaller than [1800,1900]. Important: the largest value is 1869 so that's why I took the one class interval (1815, 1908]