If you take a quick look at Real Clear Politics? continuous round-up of job-approval ratings, you may get the impression that pollsters are providing substantially different numbers on President Obama:
President Obama Job Approval
| Poll | Date | Sample | Approve | Disapprove | Spread |
|---|---|---|---|---|---|
| RCP Average | 7/9 – 7/18 | — | 56.7 | 37.8 | +18.9 |
| ABC News/Wash Post | 7/15-7/18 | 1001 A | 59 | 37 | +22 |
| Gallup | 7/16-7/18 | 1547 A | 60 | 33 | +27 |
| Rasmussen Reports | 7/16-7/18 | 1500 LV | 51 | 49 | +2 |
| Diageo/Hotline | 7/9-7/13 | 800 RV | 56 | 38 | +18 |
| Ipsos/McClatchy | 7/9-7/13 | 1007 A | 57 | 38 | +19 |
| CBS News | 7/9-7/12 | 944 A | 57 | 32 | +2 |
But if you take a closer look, you?ll understand that the pollsters are not far apart at all. The issue is the definition of the sample. Scott Rasmussen (disclosure: a former chairman of the John Locke Foundation) uses a likely-voter model, just as if we were approaching an election. Others expand their samples to include all registered voters, regardless of their likelihood of voting in the next election, or even all adults, including many nonvoters. Generally speaking, tightening the screen for who gets put into the sample will tend to increase the share of GOP-leaning voters in the sample.
So my guess is that if Rasmussen loosened his screen, he’d be finding job approval for Obama in the mid- to high-50s. If the other pollsters tightened their screens, they?d find approval rates in the low- to mid-50s.
UPDATE: The Democratic firm Public Policy Polling just released its monthly national job approval number, and it?s similar to Rasmussen?s at 50 percent approve/43 percent disapprove. PPP uses a voter screen.
