The legitimacy of an analysis is not based on the amount of significance identified in the results. Without knowing the details of your scripts and the underlying models (or equations), I can't say much about your analyses. Both 3dDeconvolve and 3dTcorr1D are supposed to work on time series data, and are typically used at individual subject level. In addition, I'm not so sure how you managed to include gender as an explanatory in 3dDeconvolve. On the other hand 3dttest++ is specifically for group analysis.
The comparison between Pearson and Spearman correlations is more about the data distribution, the number of subjects and the presence of outliers. So if you suspect that your data are not normally distributed, have low number of subjects or outliers, Spearman correlation might be more preferable.
HTH,
Gang