When taking your EPSO numerical reasoning tests you need to be quick and accurate, well… accurate within reason. You need to be accurate in the method that you choose to solve the problem and you also need to be accurate in the arithmetical procedure that you use to obtain the desired answer.

But how accurate do you need to be?

What could this magical method be? Well, we are referring to the problem of rounding. This is where an answer is given to some level of precision but is not necessarily the exact answer. So how do you know how precise you need to be?

It all depends on the question and what is being calculated. OK, so that isn’t particularly helpful but in general, three significant digits of accuracy would be sufficient but, better still, take a look at the possible answers that are listed. If there are five possible answers which are:

A – 1 B – 2 C – 3 D – 4 E – 5

then there is obviously no need to work to 5 decimal places of accuracy as all you are looking for it a single integer answer. If, however, the possible answers are:

A – 1.37% B – 2.91% C – 1.88% D – 4.5% E – 0.66%

then you obviously need to be more precise during your working.

In general, you should try not to round partial results at all during your calculation but work to the maximum precision of your calculator. The final result can then be rounded to the precision of the possible answers. This can often be achieved by doing the calculation in a particular order so that each partial answer in the calculator is the required start of the next step of the calculation. If the calculation requires the calculation of two partial results then usually you can use the calculator’s memory to store one of the partial results until the other one is done.

In some cases this is not easy to achieve and you need to write down a partial result and re-input it for a later step. You could write down a partial result to 8, 9 or 10 significant digits (the Windows calculator in scientific mode uses 32 digits!) and type it back in but this is probably unnecessary and it does take up valuable time. You do, however, need to make sure though that you keep at least one more digit of accuracy than the final result requires. This should ensure that your final answer, when rounded to the required level, is close enough to the given answer.

So, for example, suppose we have a question which tells us that the population of a country rises by 40000 to 4.312 million while its wheat consumption rises from 42.8 to 51.7 million tonnes. What is the percentage per capita increase over the given period?

The population is in millions and the increase is given as a full number so we need to mentally convert to millions before we can adjust the population for the start of the period which will be 4.312-0.04=4.272. Essentially, the problem now boils down to:

*100 * ((51.7/4.312m) - (42.8/4.272)) / (42.8/4.272)*

At this point we can use our calculator to do:

*51.7 / 4.312 = 11.98979592*

*42.8 / 4.272 = 10.01872659*

And then calculate the final answer of 19.7% from these figures. You can use these long numbers to get the exact figure or, assuming the possible answers are given to three significant digits, you can round each of them to four significant figures which would give:

*100 * (11.99 – 10.02) / 10.02 = 19.7%*

Need more practice? Try out our free Numerical Reasoning demos!

And you can of course ask us any free advice, any time!