Hi,
I'm about to start a study in which I need to statistically assess if there is a monotonic trend of a variable of interest (frecuency of IVF cases) over time. After doing some research, I learned about the Mann-Kendall test.
However, I've only seen this test being used in studies about agro- and hydro-meteorological time series. Does anybody know if I can use the MK test for my study even if it's not about those topics?
Thanks,
Hi Diego,
It need not be related to applications to some specific fields. Technically, you can use MK test for any data as long as the relevant assumptions are satisfied. I think one of the key assumptions in MK test in its simplest form is that the observations are not serially correlated over time, which might not be true in case of a usual time series data. Please check these assumptions and make sure that your data does not deviate significantly from them, and then you can use the MK test.
Thanks.
As Soudeep Deb said this method is general, and not specific to any field. The only requisite is that your data be a time series.
Hi Diego,
The Mann-Kendall test can be used to analyze trends in series from any field not necessarily agro- and hydro-meteorology. To apply Mann-Kendall test, you need to have at least 8 data points to ensure validity of the computed test statistic with respect to the required approximately normal distribution. Mann-kendall test is non-parametric in nature and this allows its use to eliminate the normality assumption (required by parametric approaches) which tends to be violated by data from a number of fields. Normally in trend analyses, you need to use a number of methods to eliminate the possible influence from the choice of a particular test on results of analysis. In this line, there are a number of rank-based or parametric tests you can think of using. Some of the methods include the Spearman's rho test (Spearman, 1904), Sen's innovative test (Sen, 2012), and a statistical test I recently developed (Onyutha, 2021). I also provide you with a link where you can download a tool called CSD-VAT in which I implemented the latest trend testing method. Find the details regarding relevant articles via the links (or references) below.
Article Graphical-statistical method to explore variability of hydro...
Deleted research item The research item mentioned here has been deleted
Article Statistical Uncertainty in Hydrometeorological Trend Analyses
Article Analyses of rainfall trends in the Nile River Basin
- Sen Z (2012) An innovative trend analysis methodology. Journal of Hydrological Engineering 17: 1042–1046.
- Spearman C. 1904 The proof and measurement of association between two things. The American Journal of Psychology 15: 72–101.
- Onyutha C (2021) Graphical-statistical method to explore variability of hydrological time series. Hydrology Research (2021) 52(1): 266–283.
best wishes
Charles
Yes, you can. For autocorrelated data you may use:
- Hamed KH, Rao R (1998) A modified Mann-Kendall trend test for auto-correlated data. J Hydrol 204:182–196
- or boostrap.
The modified trend could be run with R package "fuse", but I am not sure why it was removed from the CRAN repository. Any idea?
The R Package ‘Kendall’ illustrates how to use the test and how to do the boostrap.
Best wishes!
- Badges
- Science topic
- Similar topics
- Statistics
- Time Series