Over the years, statisticians have always used 0.05 as a standard criterion for testing for a guess statement whether you are using the spss software, the minitab or the E-views to run an analysis for your guess statement.
And it was also believed that when the p-value is greater than 0.05, we accept the guess statement and reject the alternate hypothesis. However, when the p-value is less than 0.05, we reject the null hypothesis and accept the alternate hypothesis and conclude on it.
There are many theories to justify the above statement. For example, take a look at Fisher publication on the statistical methods for research workers (1925) he added tables with value of the random variables for specially selected value of p.
Fisher made a suggestion that for sample sizes from 1 to 20 having a p- value of 0.05 is approximately 2. Which is very convenient to judge whether a guess statement or null hypothesis is significant or not. If the deviation is like two times more than the standard deviation then we say they are statistically significant.
For example in preparing this table we have to understand that in practice we do not want to know the actual value of P for any observed variable, however, whether or not it is observed that the value is open to questioning. If for instance P falls between 0.1 and 0.9 (0.1≤p≤0.9) we can never accept or believe that the hypothesis was actually being tested. But if it falls between 0.01 to 0.05 (0.01≤p≤0.05) If we have something like the above value we can strongly say that the hypothesis was actually being tested for significance.
If though the value is exactly 0.05 it is still a level for significance as far as statistics is concern.