"Anti-Semites" getting worse at identifying Jews?

So claim these unbiased researchers:

Anti-Semitism and Identification of Jewish Group Membership from Photographs
Susan A Andrzejewski, Judith A Hall, Elizabeth R Salib. Journal of Nonverbal Behavior. New York: Mar 2009. Vol. 33, Iss. 1; pg. 47, 12 pgs
Abstract (Summary)

Early literature found that holding more anti-Semitic attitudes positively predicted ability to discern whether a photograph was of a Jewish or non-Jewish person. This contradicts the well established finding that interpersonal sensitivity is generally associated with healthy psychological characteristics. In five new, previously unpublished studies we found that this relation was negative, such that more prejudiced individuals were now less accurate than less prejudiced individuals at a similar task, consistent with the general finding. A meta-analysis of all the studies showed that time was a significant moderator of the relation. Possible reasons for the temporal change are discussed. [PUBLICATION ABSTRACT]

Why?

We can speculate about other mechanisms that might have produced the positive relations in the earlier studies. In an earlier era, far from being associated with social deviance or maladjustment, the possession of anti-Semitic attitudes was respectable. At prestigious universities, anti-Semitism was openly espoused by many members of the student body, students’ families, faculty, and administration (Karabel 2005). Indeed, prejudice against minorities was previously seen as being rooted in normal processes (Dovidio 2001), as the expression of the ‘‘natural’’ superiority of some groups over others.

Today, however, outgroup prejudice is not considered socially acceptable and, indeed, is seen in many segments of society (perhaps especially in college students) as representing a kind of social deviance or pathology. Therefore, because being prejudiced is now seen as abnormal rather than normal, the characteristics of the high prejudiced person and the low prejudiced person may have changed over time. The person who espouses high levels of outgroup prejudice today is likely to possess a range of problematic psychological characteristics that are inconsistent with the development of interpersonal sensitivity, whereas we can speculate that the high prejudiced person of the past used to possess more welladjusted psychological characteristics. Today, highly prejudiced individuals readily endorse cultural stereotypes (Kawakami et al. 1998), and they feel angry about being constrained by other people’s standards for them (Devine et al. 1991). More prejudiced individuals are consistently classified as being less psychologically healthy than those who are less prejudiced (Hightower 1997; Pettigrew 1981). Thus, prejudice in an earlier era might not have been as psychologically unhealthy as it is today.5

A second line of speculation stems from the functional value of having skill in identifying Jewish faces. As noted earlier, the motive to ‘‘spot the enemy’’ proposed by Allport and Kramer (1946) might have made more sense at an earlier time than it does today. In an era when the normal social order was assumed to have wealthy White Christians (indeed, Protestants) at the top, the motive to bar Jews from attaining power and privilege (or dating one’s sister, etc.) would have been well served by a finely honed ability to identify them. Today, however, it is all but impossible for the prejudiced person to exclude Jews from advancement or to avoid their presence in neighborhoods, educational institutions, and workplaces. Therefore the potential value of being able to ‘‘spot the enemy’’ has been greatly reduced. Further undermining the adaptive value of being able to ‘‘spot the enemy’’ is the fact that the enemy is no longer very difficult to spot. In the earlier era of more overt discrimination, many Jews tried to conceal their minority group membership. In such a climate, the prejudiced person would need to develop an acute ability to use expressive or physiognomic cues to pick out members of the stigmatized group. Today, in contrast, it is highly unlikely that Jewish individuals try to hide their group membership, meaning that the prejudiced person has much less need to develop sensitivity to subtle cues indicating Jewish group membership.

Another possibility not mentioned: the inclusion of non-white subjects renders comparisons with earlier studies meaningless. Participants in the new studies are only around 67-74% "Caucasian". I imagine the small negative correlation could simply be an artifact of non-white subjects being both more willing to give "anti-Semitic" answers and worse at distinguishing between Jews and Northern Europeans. Regardless:

Accuracy on the Jewish identification task was significantly above chance in all five studies (chance = 50%, or 25 items correct; Table 1). Accuracy was slightly but non-significantly lower in Studies 3–5 where there was a mixture of ethnicities among the non-Jewish faces as opposed to only Anglo non-Jewish faces. [Note: Studies 3-5 also included a higher proportion of non-white participants.]