In the research environment I'm finding dividend per share data to be incorrect (unless I'm not understanding something).

For example, 

data = qb.GetFundamental(["KMI"], "EarningReports.DividendPerShare")

Returns a dataset containing

However KMI's dividend actually went to 0.125 in January of 2016, not April.

2015-12-31 3M:0.49;6M:0.93;9M:1.42;12M:1.7
2016-01-29 3M:0.49;6M:0.93;9M:1.42;12M:1.7
2016-02-16 3M:0.51;6M:0.93;9M:1.42;12M:1.93
2016-02-29 3M:0.51;6M:0.93;9M:1.42;12M:1.93
2016-03-31 3M:0.51;6M:0.93;9M:1.42;12M:1.93
2016-04-22 3M:0.125;6M:0.93;9M:1.42;12M:1.93

dividend history
I spot checked a other symbols with known dividend changes (e.g. MCK) and found similar issues. 
Am I misunderstanding something about how the fundamendal data is supposed to work, or is the fundamendal data just not good?