Usually in an ordinary test of hypothesis, if the calculated statistics is very large (>>) when compared with the tabulated statistics, we say that the difference in the "pertinent" parameters, we are comparing is highly significant and hence we reject the null hypothesis (which is usually: No difference between parameters). Now with respect to your ARDL Bond test or any other "irregular" parametric or non-parametric test, you need to know where the critical region lie, as well as where your null (guess) lie, a sketch of the problem may also be quite useful in determining whether to accept or reject the null.