Public Opinion

This month

Job Approval

In comparison with last month, the government approval rating oscillated 2% points upward, while those disapproving downward at the same rate. For 10% of the respondents, the government is doing a “great” or “good” job, but for 69% of them, it’s rather doing “bad” or “terrible”. For others 21% of the respondents, the government performance is “regular”.

Key points driving negative views on Dilma’s administration:

  • A stubborn chilling economy.
  • A deteriorating political situation.
  • A negative-stained image of the major parties in the governing coalition caused by long-standing corruption scandals.

What is public opinion?

For few, public opinion is an aggregate of the individual views, attitudes, and beliefs about a particular topic. Some scholars treat the aggregate as a synthesis of the view of all or a certain segment of society. For the American sociologist Charles Horton Cooley (1918) public opinion is a process of interaction and mutual influence rather than a state o broad agreement. For the american political scientist V. O. Key (1961), public opinion is opinions held by private persons which government find…

What is approval ratings about?

Job Approval

Do you approve or disapprove of the way [first & last name] is handling his job as president? Approval Ratings were first introduced in the 1930s by George Gallup. He wanted to quantify how much the populace approved of the President of the United States. Although approval ratings can tell much about the humor of the electorate, it’s by far a precise measure of the government performance or the leader. In addition, because of the swingable nature of politics and polling methodologies, these ratings are generally noisy, even taken as unscientific by few. The Brazilian Presidency Project

Zaller begins by arguing that expressions of opinion are not statements about a fundamentally stable state of mind but rather functions of the considerations at hand when a question is asked. Considerations are “any reason[s] that might induce an individual to decide a political issue one way or another” and reflect both cognition and affect (Zaller, 1992, pp. 40-44).

What causes house effects?

These systematic house effects can be large — in fact, they can be larger than the nominal “margin of error” that is reported along with every poll. What causes house effects? I look at some of the possible explanations here, which is a more detailed comparison of systematic lean in both Presidential approval and head-to-head election polls for the 2004 election. In particular, note that the Zigby, Harris, and AP polls present their “job approval” questions differently than the other houses: they offer respondents a slightly wider choice of answers than a simple “approve” or “disapprove.” There are also differences in the sample and the average amount of “undecideds” by polling house: Fox polls survey registered voters (most other polls survey “adults”) and they tend to have slightly higher numbers of undecideds.

Gross vs. net approval

The plots above show net approval (that is, approval minus disapproval) but when the newspapers report the President’s approval ratings, they usually report only gross (that is, just the approval rating without subtracting the disapproval). What is the difference? A large part of the difference among polls is in the proportion of “undecided” or “don’t know” or “decline-to-state”. For example, the Fox poll tends toward high proportions of undecideds while the Zigby poll tends toward low proportions. When the proportion of undecideds varies greatly between two polls, the gross approval doesn’t tell the whole story. For example, in the first few weeks of his administration, President Bush had an approval rating of 46% on the Fox poll – but 40% of respondents were undecided and only 14% reported disapproval. Contrast this the Fox poll shortly before Hurricane Katrina, when his approval rating was 45%, but he had only 5% undecided and a disapproval rating of 50%. Gross approval was quite close in these two cases (46% vs. 45%) but no one would say that they represented the public’s perception of the President equally well. In the earlier poll, the President’s net approval was 46% - 14% = +32%; in the later poll, his net approval was 45% - 50% = -5%.

To see the difference between gross and net approval over the last 12 months place the mouse pointer over the image below; then move the mouse pointer away to return to gross approval. In each panel the black line shows what the trend would have been without that firm’s polls. This gives a rough idea of how much each firm’s polls influences the overall trend, both in gross and net approval.

Statistical summary

As you can see in the chart in the top, the lowest approval rating ever recorded is 23, while the highest is 87, and the mean is 56.31. The Office of the President was pretty popular during this time frame, though it did have it’s ugly moments (Watergate). there were still times when the approval rating dipped way down.

Overall approval rating over time

From the chart below, we can see that Brazilian public tends to be a fickle bunch. While no incumbent president lost a re-election bid since the redemocratization, no one had high levels of popularity either. President Lula was the sole case the incumbent made his successor and that his government enjoyed high levels of approval.

Approval ratings by party

What does this mean?

Approval Ratings are fun for the public, but they don’t have much bearing on the direction of elections or the Presidency. For example:

Harry Truman has the lowest rating ever recorded; even lower than Richard Nixon, who was forced to resign! Truman also has the highest rating ever recorded, though that was the very first survey taken, so the methodologies may have been imperfect Dwight Eisenhower’s approval rating never dropped below 50, but that popularity was still not enough for his Vice President (Richard Nixon) to win the 1960 presidential election.

Infrequently asked questions

Where do these polling figures come from? Here and here.

What are these trend lines? They’re estimates from a model which treats latent party support as something which evolves smoothly over time, and is made manifest through particular potentially biased polling snapshots.

How are the effects of the different polling companies identified? By assuming that on average, polling companies are unbiased.