# What percentage of systematic data starts to be considered random

There is a set of 100 different objects in a box. I pull out 99 of them randomly and place them on a table and ask a person to take a look, then I put all objects back into the box. After a short (5 minutes) interval I pull out another 99 random objects from the same box. The person is asked to say whenever an object is the same as it was before. If the missing object did not stand out from all others on the first go, then the person will answer that the objects are all the same.

How many objects should there be in the box and how many should be placed on a table, assuming there is an equal chance for every object to be pulled out, in order to trick the average person into thinking that all or almost all of the objects are different? What if the process is repeated 6 times, every 5 minutes, would it have any noticeable effect on the results?

• In case you don't get any good answers here (wait for it some time) let me know. I could merge this to Cross Validated. However, I think this is on topic here as well. Jan 19, 2014 at 13:15