In the study, participants were asked to count black squares on a white screen. The number of squares shown ranged from one to nine and participants were given as much time as they wanted before answering. When the number of squares was in the subitizing range (one to four), both math-anxious and non-math-anxious participants performed equally well, but when the number of squares was in the counting range (five to nine), the math-anxious group took longer and were less accurate.
The University of Waterloo's news release about the study includes this interesting note: "Previous studies have shown that a weakness in basic math abilities has a greater negative effect on employment opportunities than reading difficulties [do].""
Link to Original Source