I had a similar problem in a changeblindness-experiment. Some participants did not detect the change in the first trial while others detected the change. The results files of these two types of participants had a different column order.
R, as a statistic software, does not care too much about the order of the column order. Column names must be the same.