I abhor the >0.05 convention. As mentioned in #6, p-values should just be reported, and significance should be left to the reader to interpret.

It should also be mentioned that whether or not the data is normally distributed has a significant impact on whether or not the test is appropriately interpreted.