Friday, August 24, 2012

Electoral volatility due to different questions?

Poll volatility 

Below, you find the summed changes in parliamentary seats over all parties in consecutive opinion polls for the four main polling firms in the Netherlands in the lead up to the 2012 elections. I will update the table below in the next weeks.
This overview follows from my earlier post on Dependent Interviewing. Maurice de Hond (peil.nl) is the only survey pre-loading earlier voter preferences into survey questions. I expect this to lead to less volatility in voter preferences for Maurice de Hond, as compared to the other polling firms.

Update September 12th: With the final polls out on election day, it seems that the polls of Maurice de Hond are indeed most stable over time, and in my previous post I argues this was because of the fact that he uses Dependent Interviewing in his question on " what would you vote if there were elections today". Still, I would have expected a larger effect. Let's see tomorrow which polling firm did best. My bet: Synovate, because they are using the most sound (although still not perfect) methodology of polling people. More on that tomorrow...



 Maurice de Hond (peil.nl)
Intomart/de stemming
 Synovate
 TNS-NIPO
 week 23 (03-06)
4
-
 8
-
 10-06
8
-
-
-
 17-06
6
-
 8*
-
 24-06
4
-
-
 14**
 01-07
10
-
 12*
 8
 08-07
12
-
 12
 12
 15-07
6
 8
 10
 12
 22-07
2
 10
-
 6
 29-07
4
-
 8*
 6
 05-08
2
-
-
 10
 12-08
10
10**
 8*
 16
 19-08
8
  8
12
  8
 26-08
6
20
8
16
03-09
22
20
12
14
10-09
26
10
16
20










 average change
8,7 12,2*** 10,3*** 11,8

* two-week difference
** three-week difference
*** rounded down due to inclusion of multi-week changes