There is a sample with a total size of 70 values (expressed in percentages). I need to find out if the median of this 70 values is significantly different from zero.

I am using two tests:

1. t-test: Does it make sense to use on-sample t-test for comparing medians with zero? Or should I rather compare means?

2. Non-parametric Wilcoxon test: The goal stays the same as in point 1, but in this case I want to use a non-parametric test, as specific empirical literature recommends it because it usually delivers more powerful results. Therefore I am comparing the 70 values with a column of 70 zeros (=treated as two related samples). Is this technique justifiable?

Thank you in advance for your hints!