The more times you do a test, the more accurate the results will be.
Since there are only 2 options/sides the probability should be 1/2 for both heads and tails. 
Her error was both mathematical (her experimental probability number didn't make sense) and her lack of sufficient testing.