Why would my the polyA selection RNAseq preparation method get much greater dispersion than the RNAseq data got from ribo-depletion library?

Question

[bibilong1111]

I have RNAseq data got from both polyA selection and ribo-depletion library preparation, but the ribo-depletion library data showed much less dispersion. Their sequencing depth are the same. Attached is the relative log expression plot and the first several sample are the polyA selection library. The data have not been normalized.

[Asaf]

There are no figures...

What do you mean by dispersion? variance between replicates?