David, this is a statistics term, so perhaps my explanation in English won't be very thorough or professional:
The "tendency analysis" is a method to interpretate statistic data, for example the date collected through a survey. This method is used to compare various series of data which are expressed in different measure units, in different scales. The series of data are usually time series, for instance the percentage of vote intention in each month of the campaigne.
In order to homogenize the different series, you transform each one as follows:
Vote intention: January 1,235.4 (I invented all the figures)
February 1,938.2
March 1,750.0
...
You take the first figure as the "basis" (we call it "basis number" in Spanish) and consider it as 100 (100%). Then you transform the following figures into percentages by comparison with the first one, like this:
Jan: 100
Feb: (1,938.2*100)/1,235.4= 157
Mar: (1,750.0*100)/1,235.4= 142
This means that the vote intention in February was 57% higher than in January, and that in March it was a 42% higher than in January.
Finally, if you are comparing this series with others expressing different concepts (e. g. personal income of the voters, which is expressed in pounds whereas the vote intention is expressed in persons), you use the transformed series:
Vote intention: Jan 100
Feb 157
Personal income: Jan 100
Feb 125
You can say that the vote intention has risen in a greater amuount that the voters' personal income from January to February.
Of course, I didn't choose a sensible comparison in the rush of the moment, but when you manage really long time series and complicated statistic equations this analysis comes in useful.
Hopefully I didn't make it messier.
