Short on Confidence: Changes in Attitudes toward American Institutions and Occupations

By Richard Seltzer and Rhea Roper-Nedd

It is well known that Americans’ confidence in government and other institutions has dropped substantially since the 1950s. Many researchers have theorized about the general decline in confidence and why confidence has fluctuated. Some have examined why several institutions have fared better or worse than others. Few, however, have examined confidence in different institutions over time. The most comprehensive study was done by Seymour Martin Lipset and William Schneider in 1987; their data are now twenty years old.

In this analysis, we look at attitudes toward American institutions, holistically, from many different perspectives: historically, relative to other institutions, in the context of actual events and happenings, and by using different measures. We extend Lipset and Schneider’s analysis by making the data current, adding some new series, and by using an additional analytical technique: the coded table.

A coded table analyzes a two-way table and employs many of the same concepts as a boxplot. It is an exploratory data analysis technique, developed by John Tukey and refined by Paul Velleman and David Hoaglin, that helps the user identify patterns in the data.

Those values falling within the middle 50 percent (the hspread) of the data are represented in the coded table as dots. Relatively large or small values are represented by plus or minus signs (between the hspread and one and a half times the size of the hspread). Extreme values are coded “#” or “=” for plus and minus values, and truly extreme data are coded “P” for plus or “M” for minus.

For example, Table 1 analyzes data on confidence in American institutions from the National Opinion Research Center’s General Social Survey (NORC-GSS). A cursory review of this coded table shows the military with a majority of pluses, indicating relatively high confidence in this institution. The many dots in the middle years, however, also show that confidence slipped to the typical rating of all the institutions between 1978 and 1991. Similarly, the consistent minuses for organized labor indicate its overall low ratings, while the press went from average to relatively negative ratings by the early 1990s. While no “extreme” values appear in the NORC coded table, in the Harris confidence data displayed in Table 2 (see below), we see extreme confidence in medicine prior to 1974 and in the military in 1966, as well as during the Persian Gulf wars.

According to the NORC-GSS, the average rating for people running medicine (47 percent) was almost ten points higher than the next highest rated institutions—the scientific community and the military. Five institutions had ratings below 20 percent: organized labor, Congress, TV, the press, and the executive branch.

Ratings for two of the institutions showed substantially more variability than the others over time. Confidence for the people running banks and financial institutions fell after the savings and loans crisis of the early 1980s. Banks and financial institutions averaged 32 percent in the eight surveys between 1975 and 1984. In later years, the average was 22 percent. These levels fell below 15 percent for 1991 and 1993 following Wall Street scandals involving Michael Miliken, Ivan Boesky, and others.

Confidence in people running the military was relatively steady from 1973 to 1990, averaging 33 percent. It jumped to 60 percent in 1991 with the first Persian Gulf War, fell to the high 30s until 2002, then rose again to 54 percent with the beginning of the second Gulf War.