Shishikura (1991) proved that the Hausdorff Dimension of the boundary of the Mandelbrot set equals 2, in this paper, but I can't figure out one thing : can we say all open subsets of this boundary has dimension 2 ? Are there any references ? Thanks.

Yes  and it is indeed in Shishikura's paper, as in your comment. The "open subsets" of the boundary with respect to which topology? That of the ambient space, or the relative topology of the set itself? Since the latter topology is extremely wild, it is much clearer to use the ambient topology, as Shishikura does. 

