“In order to assess explicit prejudice toward Jews, we directly asked respondents “How much to blame were the Jews for the financial crisis?”

That question seems to be designed to ilicit a response. A valid comparison of data would also substitute other religions and atheists for Jews then compare the data of all of the sets of questions.

In other words if you asked “In order to assess explicit prejudice toward ___, we directly asked respondents “How much to blame were the ___ for the financial crisis?” You would be likely to get people who blamed ___ simply because they were asked.

The number of people who blamed “the Jews” still seems high to me even with a leading question.