Hello, could someone assist me in interpreting the results of the sequential Mann-Kendall Sneyer test? Indeed, according to Dufek (2008: Precipitation variability in São Paulo State, Brazil), "In the absence of any trend, the graphical representation of the direct series (u(t)) and the backward series (u'(t)) obtained with this method yields curves that overlap several times." In my case, I observe two to three overlaps, often with sequences that exhibit significant trends. Should I also conclude that there is an absence of trends in my dataset?
The sequential Mann-Kendall test, also known as the Mann-Kendall-Sneyers (MKS) test, is a variation of the Mann-Kendall test that aims to detect trends in time series data. The test involves comparing the original time series to its reverse version to identify potential trends. The graphical representation of the direct series (u(t)) and the backward series (u'(t)) can provide insights into the presence or absence of trends. However, the interpretation can be nuanced.
Dufek (2008) suggests that if there is no trend, the curves of the direct and backward series will overlap several times. In your case, you observe two to three overlaps, often with sequences that exhibit significant trends. This situation requires careful consideration:
In conclusion, while the observation of overlaps in the graphical representation of the sequential Mann-Kendall test might suggest a lack of clear trends, the presence of significant trends in some segments complicates the interpretation. It's important to analyze the trends, magnitudes, and durations of both overlaps and significant trends while considering the broader context of your data and the subject matter you're studying. If possible, consulting with a statistician or subject-matter expert might help you make a more informed interpretation of your findings.
Ma'Mon Abu Hammad Could you advise me on some documentation about this test?