The pro-outgroup bias of white liberals is not an artifact of 0-100 feeling thermometers. It similarly rears itself in 4-point measures of group favorability. 
The feeling thermometer differences for comparison (too few asians to graph by ideology so they were excluded)
1/2 The responses below may not be perfectly comparable to those in the first graph (below, values 2-3=*mostly* unfavorable/favorable; whereas in the first graph, values 2-3=*somewhat* unfavorable/favorable), but no pro-outgroup bias is observed among white libs in 2007 or 2009
2/2 That said, there is a small drop in their ingroup favorability scores between 2007 and 2009 that just misses conventional levels of significance (p=0.056). Whether this drop continued thereafter (eventually manifesting in the one we observe today) is unclear.
