Understand Polls
About forty million people in California from many different walks of life live in an incredible diversity of communities. How can we successfully understand so many opinions? Many academic inquiries about public opinion use surveys to ask people about their positions. How valid are people's responses? They may not want to be truthful, or perhaps their views change. Are we capturing the intensity of people's beliefs? Firmly held beliefs held by a minority may be far more influential than widespread beliefs that motivate little or no action.
These are some of the challenges of the study of public opinion. After World War II, political scientists addressed these challenges by developing public opinion surveys that are administered in scientific ways so that we may understand their validity. Three characteristics should be evaluated. First, have the pollsters randomly selected the respondents from the defined population(s) we wish to study? Random, or probability, sampling means that each respondent has an equal chance of being selected from a population for questioning. Data from reputable polling organizations are best to follow because their sampling techniques will be more likely valid and published for our consideration as part of the survey. Be wary of ones that do not clearly explain how respondents have been chosen. They are likely to be for entertainment only.
Second, a scientific survey will clearly display its sampling error. This statistical measure of the probability difference between the sample and the population is based purely on a mathematical calculation. A good survey has a margin of +/- 3% or less, meaning that if there is a difference in responses of six points or less, we cannot say that the two groups have different opinions. It is also essential to note how the survey randomly selected respondents. For example, a few years ago, if a survey was randomly selected only from landlines and not cell phones, the sample would likely skew toward an older population.
Third, we must study the wording of the survey questions. Are the questions effectively addressing the issue that we want to study? Are they going to be understandable? Are the response choices fair or biased? If we are comparing surveys over time, have words changed in meaning?
Here is an example of a scientific poll from the Institute of Government Studies at UC Berkeley regarding whether California should retain the death penalty. Five thousand thirty-six registered voters, including subgroups, were selected randomly to respond to an internet poll, with a random sample of +/- 2.5%. Respondents were asked:
"The legislature is considering placing a constitutional amendment on the 2022 statewide election ballot asking voters to formally abolish the death penalty as a punishment for certain crimes in California. If you were voting today, how would you vote on the constitutional amendment to abolish the death penalty in California?"
Tables 4.3.1, 4.3.2., 4.3.3, and 4.3.4 show the responses by subgroup. Registered voters who identified themselves as Democrats or as liberal were far more likely to favor the repeal of the death penalty. Voters living in coastal counties were somewhat more in support of this proposed amendment than voters living in inland counties. Women and younger people were also somewhat more likely to favor the repeal.
Table \(\PageIndex{1}\): Opinion about the Death Penalty by Party Registration
Registered Voters |
Repeal the Death Penalty |
Keep the Death Penalty |
Undecided |
All Voters |
44 |
35 |
21 |
Democrats |
63 |
19 |
18 |
Republicans |
12 |
68 |
20 |
No Party Preference |
43 |
31 |
26 |
Minor Parties |
29 |
41 |
30 |
Table \(\PageIndex{2}\): Opinion about the Death Penalty by Political Ideology
Political Ideology |
Repeal the Death Penalty |
Keep the Death Penalty |
Undecided |
Strongly conservative |
12 |
69 |
19 |
Somewhat conservative |
19 |
62 |
19 |
Moderate |
33 |
37 |
30 |
Somewhat liberal |
64 |
17 |
19 |
Strongly liberal |
83 |
7 |
10 |
Table \(\PageIndex{3}\):Opinion about the Death Penalty by Region
Region |
Repeal the Death Penalty |
Keep the Death Penalty |
Undecided |
Los Angeles County |
49 |
28 |
23 |
San Diego County |
41 |
42 |
17 |
Orange County |
36 |
44 |
20 |
Inland Empire |
34 |
43 |
23 |
Central Coast |
41 |
31 |
28 |
Central Valley |
35 |
44 |
21 |
San Francisco Bay Area |
54 |
27 |
19 |
North Coast/Sierras |
44 |
34 |
22 |
Table \(\PageIndex{4}\):Opinion about the Death Penalty by Gender
Gender |
Repeal the Death Penalty |
Keep the Death Penalty |
Undecided |
Male |
41 |
40 |
19 |
Female |
46 |
31 |
23 |
Now, consider the validity of this poll. Respondents were randomly selected using state records of registered voters. Since the poll asks about a proposed referendum, selecting registered voters rather than all adults makes sense. However, all California adults are not the target population. We know that registered voters are typically a bit more conservative. If registration increases between the time of this survey and election day, its validity will be weakened.
Second, consider the question that is asked and the possible response. The strength of asking about the death penalty is that nearly everyone will likely understand the issue because of its relative clarity and prevalence in the media. The question wording may be confusing because the poll asks if the respondent favors repealing the death penalty rather than just asking if they favor or oppose it. It is also likely that if the survey question were more complex to consider more choices such as "life imprisonment without possibility of parole" as an alternative, then support for the death penalty would decrease.
The challenges of studying Californian public opinion go beyond just survey design. An additional challenge is the relative lack of statewide surveys. General surveys about opinion are often conducted nationally by national polling organizations such as Gallup and Pew. Many other academic institutions and polling organizations provide a constant stream of public opinion data. In California, very few polling organizations concentrate only on state and local public opinion. The most prominent are the Public Policy Institute of California and UC Berkeley's Institute of Government Studies. Newspapers and television stations will also conduct surveys, which are usually associated with predicting electoral outcomes. Election-year surveys have value but are not often designed to investigate underlying beliefs. Moreover, the party-driven mass politics, advertisement blitzes, and the immediacy and drama of the campaign limit the scope and depth of these polls.
In summary, political scientists use surveys to learn about public opinion, but they must be evaluated critically for their validity. Moreover, systematic and regular surveys of California public opinion over the long term are limited. Instead, surveys often present a series of snapshots about public beliefs, attitudes, and opinions that give valuable, although incomplete, understandings.