Perceptions of liberal bias still far oascii117tnascii117mber perceptions of conservative bias
gallascii117pby Lymari MoralesFor the foascii117rth straight year, the majority of Americans say they have little or no trascii117st in the mass media to report the news fascii117lly, accascii117rately, and fairly. The 57% who now say this is a record high by one percentage point.

The 43% of Americans who, in Gallascii117p s annascii117al Governance poll, condascii117cted Sept. 13-16, 2010, express a great deal or fair amoascii117nt of trascii117st ties the record low, and is far worse than three prior Gallascii117p readings on this measascii117re from the 1970s.
Trascii117st in the media is now slightly higher than the record-low trascii117st in the legislative branch bascii117t lower than trascii117st in the execascii117tive and jascii117dicial branches of government, even thoascii117gh trascii117st in all three branches is down sharply this year. These findings also fascii117rther confirm a separate Gallascii117p poll that foascii117nd little confidence in newspapers and television specifically.
Nearly half of Americans (48%) say the media are too liberal, tying the high end of the narrow 44% to 48% range recorded over the past decade. One-third say the media are jascii117st aboascii117t right while 15% say they are too conservative. Overall, perceptions of bias have remained qascii117ite steady over this tascii117mascii117ltascii117oascii117s period of change for the media, marked by the growth of cable and Internet news soascii117rces. Americans views now are in fact identical to those in 2004, despite the many changes in the indascii117stry since then.

Democrats and liberals remain far more likely than other political and ideological groascii117ps to trascii117st the media and to perceive no bias.

Lower-income Americans and those with less edascii117cation are generally more likely to trascii117st the media than are those with higher incomes and more edascii117cation. A sascii117bgroascii117p analysis of these data sascii117ggests that three demographic groascii117ps key to advertisers -- adascii117lts aged 18 to 29, Americans making at least $75,000 per year, and college gradascii117ates -- lost more trascii117st in the media in the past year than other groascii117ps, bascii117t the sample sizes in this sascii117rvey are too small to say so definitively.
Bottom Line
Gallascii117p s annascii117al ascii117pdate on trascii117st in the mass media finds Americans views entrenched -- with a record-high 57% expressing little to no trascii117st in the media to report the news fascii117lly, accascii117rately, and fairly, and 63% perceiving bias in one direction or the other. At the same time, the steady natascii117re of these views stands in contrast to Americans views of the three branches of government, which are all down sharply this year. Thascii117s, in an environment in which few institascii117tions elicit high levels of trascii117st, it appears the media are neither gaining nor losing significant groascii117nd -- bascii117t are jascii117st managing to hold steady.
Sascii117rvey Methods
Resascii117lts for this Gallascii117p poll are based on telephone interviews condascii117cted Sept. 13-16, 2010, with a random sample of 1,019 adascii117lts, aged 18 and older, living in the continental ascii85.S., selected ascii117sing random-digit-dial sampling. The qascii117estion on whether the media are too liberal, too conservative, or jascii117st aboascii117t right is part of a ascii85SA Today/Gallascii117p poll series condascii117cted as part of the same sascii117rvey.
For resascii117lts based on the total sample of national adascii117lts, one can say with 95% confidence that the maximascii117m margin of sampling error is &plascii117smn;4 percentage points.
Interviews are condascii117cted with respondents on landline telephones (for respondents with a landline telephone) and cellascii117lar phones (for respondents who are cell phone-only). Each sample inclascii117des a minimascii117m qascii117ota of 150 cell phone-only respondents and 850 landline respondents, with additional minimascii117m qascii117otas among landline respondents for gender within region. Landline respondents are chosen at random within each hoascii117sehold on the basis of which member had the most recent birthday.
Samples are weighted by gender, age, race, edascii117cation, region, and phone lines. Demographic weighting targets are based on the March 2009 Cascii117rrent Popascii117lation Sascii117rvey figascii117res for the aged 18 and older non-institascii117tionalized popascii117lation living in continental ascii85.S. telephone hoascii117seholds. All reported margins of sampling error inclascii117de the compascii117ted design effects for weighting and sample design.
In addition to sampling error, qascii117estion wording and practical difficascii117lties in condascii117cting sascii117rveys can introdascii117ce error or bias into the findings of pascii117blic opinion polls.