The researchers developed a new method called least trimmed squares (LTS) to better estimate circular regression models. This method is more robust and can handle outliers and unusual data points well. Their approach was tested using simulations and real data, showing that the LTS estimator effectively fits circular models even when there are problematic data points present.