Poll: Americans Don’t Trust Polls

If you’re like 75 per­cent of Amer­ic­ans, you think the polling data presen­ted in this art­icle is biased.

Ac­cord­ing to a new poll from the “data in­vest­ment man­age­ment” or­gan­iz­a­tion Kantar on Wed­nes­day, three in four Amer­ic­ans “be­lieve that most polls you hear about … are biased to­ward a par­tic­u­lar point of view,” while only 19 per­cent think they are un­biased.

That find­ing, re­leased as part of a lar­ger event on the fu­ture of pub­lic-opin­ion polling sponsored by Kantar, paints a dark pic­ture of the sur­vey re­search in­dustry, which is try­ing to ad­just to changes in tech­no­logy and pri­vacy pref­er­ences that are mak­ing it harder for the in­dustry to reach many Amer­ic­ans.

But the event also comes less than two weeks since Kantar ac­quired the lead polling firm on Pres­id­ent Obama’s reelec­tion cam­paign last year, Ben­en­son Strategy Group.

The poll is de­scribed as a “bench­mark sur­vey” that aims “to start map­ping the path to pub­lic opin­ion.” Re­spond­ents in­dic­ated that they have more trust in sur­veys con­duc­ted by non­par­tis­an groups, such as found­a­tions or aca­dem­ic groups, than those con­duc­ted for polit­ic­al cli­ents or by the news me­dia. Sixty-four per­cent have at least “some trust” in polls con­duc­ted by “aca­dem­ic cen­ters,” and 54 per­cent trust sur­veys from “non­par­tis­an found­a­tions.” But only 46 per­cent have a modic­um of trust in polls from polling com­pan­ies, and even few­er, 43 per­cent, trust those sur­veys from news me­dia or­gan­iz­a­tion.

That’s still more trust than Amer­ic­ans put in sur­vey res­ults from poll ag­greg­at­ors who av­er­age or oth­er­wise meld polling data (30 per­cent); polit­ic­al parties or can­did­ates (27 per­cent); and auto­mated voice-re­cord­ing firms (16 per­cent).

(In an ex­per­i­ment, half of re­spond­ents were asked about spe­cif­ic firms, which in­dic­ated more trust in me­dia polls. The two news or­gan­iz­a­tions chosen — NBC News and The Wall Street Journ­al — might mol­li­fy dis­trust from both ends of the polit­ic­al spec­trum.)

Re­spond­ents were split fairly evenly on the ques­tion of wheth­er journ­al­ists and ed­it­ors “use a com­mon set of stand­ards to eval­u­ate the qual­ity of polls to de­term­ine which polls they will re­port and which ones they will not” — 51 per­cent said yes, and 41 per­cent said no.

Na­tion­al Journ­al and The Hot­line only re­port res­ults for polls con­duc­ted util­iz­ing prob­ab­il­ity samples and gen­er­ally avoid auto­mated-phone sur­veys.

The Kantar poll was con­duc­ted Ju­ly 24-Aug. 4 and sur­veyed 1,011 adults by land­line and cell phone. The mar­gin of er­ror is plus-or-minus 3.1 per­cent­age points — as­sum­ing you be­lieve that.