A long-debated question in political science is how coherent public opinion is among the masses. In particular, a seminal finding in this line of study holds that people are very unstable in their policy opinion over time–expressed opinions at one point often change when probed years later. Panel data from the Voter Study Group contains several of the same opinion questions asked in 2011 and 2016 of the same 8,000 people, and thus offers some more recent evidence for answering the question of over time opinion constraint. I take a look at nine of these VSG questions in the graphs below, spanning a wide range of issue areas. All of these have different response scales (some range from 1-5, others just provide just two options and thus range from 0-1). In Figure 1, I calculate the point change on each survey question scale from 2011 to 2016, and then graph the survey respondent distribution for this opinion change indicator (a value of zero indicates no opinion change, while each value greater than that means greater opinion shift).
Interestingly, considerable over time opinion stability emerges. On survey questions that include only two responses, people maintain the same opinion five years later at very high rates (between 82 and 90 percent, with a high of 90 for opinion on the legality of gay/lesbian marriage). On questions with three survey scale points, the percentage exhibiting no opinion change ranges from 73 to 79 percent. More change appears for the four-question racial resentment battery–which makes sense given evidence of racial attitude change over the last decade or so–but even here almost all of the movement is between 0 and 1 point changes on a 1-4 point scale. The greatest opinion change forms for the question of whether it should be made harder or easier for foreigners to immigrate to the U.S. (a majority move at least one point). But again, much of that shift is confined to one-point movement on the scale (35 percent) amid a plurality who see no change (45 percent).
Another way of assessing over time opinion stability has used correlation coefficients (such as in Converse 1964). Table 1 shows the correlation for each item between measurements in 2011 and 2016. Most of these reveal very strong correlations, and thus further reinforces a picture of stable over time public opinion from 2011 to 2016.
Of course, it’s worth keeping in mind some caveats for this evidence of high over time opinion constraint. First, as with all panels, the VSG is likely made up of an especially politically sophisticated sample (staying on the panel and completing a second wave five years later) and thus this might not represent the broader American population too well. Second, the issues that appear in both waves seem to cover more entrenched attitudes–in terms of saliency in the public discourse (abortion, global warming, healthcare) and less policy-based and more group-centered beliefs (anti-black racial resentment and immigrants). In this sense, answers to these questions might not be representative of all public opinion attitudes.