Half of social-science studies fail replication test in years-long project

"Of the 164 studies that they focused on, they were able to replicate only 49% with statistical significance."

But is it really only? That is, 49% could be argued to be a good result especially in social sciences.

Only half?