Here’s a comparison of state Democratic primary polls by Public Policy Polling with the actual primary or caucus results so far. All polls were conducted within a few weeks of the state’s primary.
State |
ppp poll |
Actual results |
Difference |
Alabama |
59% Clinton
31% Sanders
|
78% Clinton
19% Sanders
|
+ 19% Clinton |
Arkansas |
57% Clinton
32% Sanders
|
66% Clinton
30% Sanders
|
+ 9% Clinton |
Georgia |
60% Clinton
26% Sanders
|
71% Clinton
28% Sanders
|
+ 11% Clinton |
Louisiana |
60% Clinton
29% Sanders
|
71% Clinton
23% Sanders
|
+ 11 Clinton |
Mississippi |
60% Clinton
26% Sanders
|
83% Clinton
17% Sanders
|
+ 23% Clinton |
South Carolina |
55% Clinton
34% Sanders
|
74% Clinton
26% Sanders
|
+ 19 Clinton |
Tennessee |
58% Clinton
32% Sanders
|
66% Clinton
32% Sanders
|
+ 8% Clinton |
Texas |
57% Clinton
34% Sanders
|
65% Clinton
33% Sanders
|
+ 8% Clinton |
Virginia |
56% Clinton
34% Sanders
|
64% Clinton
35% Sanders
|
+ 8% Clinton |
Vermont |
10% Clinton
86% Sanders
|
14% Clinton
86% Sanders
|
+ 4% Clinton |
Massachusetts |
49% Clinton
46% Sanders
|
50% Clinton
49% Sanders
|
close! |
Iowa |
48% Clinton
40% Sanders
|
50% Clinton
50% Sanders
|
+ 10% Sanders
|
Michigan |
50% Clinton
40% Sanders
|
48% Clinton
50% Sanders
|
+ 10% Sanders
|
Minnesota |
50% Clinton
32% Sanders
|
38% Clinton
61% Sanders
|
+ 29% Sanders
|
Oklahoma |
46% Clinton
44% Sanders
|
42% Clinton
52% Sanders
|
+ 8% Sanders
|
Conclusion: Why does anybody care what PPP says?
I’m only picking on PPP because they’ve done a lot of state primary polls, and they’re a fairly well regarded pollster (with a B+ rating from 538). Other pollsters have also been wildly wrong.
It seems like there is very little overlap between people included in PPP’s polling samples and people who actually vote in the primaries. The primary voters they haven’t counted haven’t consistently voted one way or another. It just means their polls have little predictive value. Other pollsters are having the same problem.