One of our regular and more thoughtful commenters, LFC, writes in response to my earlier post:
You say there has been “no secular decline in trust in govt. It waxes and wanes with economic growth.”
Yet your own graph, if I am reading it correctly, does not support this assertion. The year 1958 was a mild recession year, I seem to recall reading somewhere, yet trust in govt, acc. to your graph, was just below 80 percent. It never gets up to that level again, even in the boom years of the 90s. Thus your own graph suggests there has been a decline, albeit not as sharp as Lilla and others suggest, in trust in govt.
I read Lilla’s piece quickly a while ago, and I nodded at his line about the 65 War on Poverty. It is difficult to imagine a comparable response now, and your reference to the the health care question doesn’t really rebut this convincingly.
Lilla overdraws the picture, but to suggest that there has been no change in public attitudes toward government in the past 60 years — well, I’m not quite ready to sign on to that assertion. For one thing, it implies that the period conventionally dubbed the 60s (ie. c.1964-c.1973) had no lasting effects on public attitudes toward govt. Is that true? Can it possibly be the case that the Vietnam War and Watergate had no lasting effect on public attitudes? That from this standpoint they might as well never have happened? I find that a little hard to believe.
I am correct to say that there has been no “secular decline”—in the dictionary definition sense of “secular” as meaning “of or relating to a long term of indefinite duration.”
But LFC is correct about 1958: my earlier graph in this post shows that the economy was weak, but trust was unnaturally high. Indeed, many of the data points in the 1960s are “above” the regression line in that graph, suggesting that trust was higher than expected based on the relationship between trust and the economy over the entire period.
But I don’t think that the 1990s are the exception that LFC suggests. In 1992, 1996, and 2000, the economy grew and trust increased, but trust was not much lower than expected, give the level of growth. This is especially true in 2000.
Does the relationship between trust and the economy imply that the Vietnam War and Watergate were irrelevant? Not necessarily, although it’s somewhat difficult sort out their effects from that of the economy. For example, consider the decline in trust in government between 1964 and 1974, a period that largely spans both events. During this period, with the exception of 1972, the economy grew at progressively lower and lower rates in each successive election year (when the surveys from which the trust data come were conducted). That said, some evidence suggests that Watergate still mattered. Luke Keele’s analysis (pdf) of the period from 1970-2000 suggests that Watergate pulled down trust by about 2 points, over and above the effects of the economy and other variables. The same may be true of the Vietnam War.
The thrust of Lilla’s argument and LFC’s comment is that today is different than yesterday (although, to be sure, LFC thinks Lilla overstates things). That is, the nature of trust in government is qualitatively different now than in the 1950s or early 1960s—and that events like the Vietnam War and Watergate are possibly to blame. I tend to react against this narrative because it typically stops there and makes it seem as if we used to live in some Eden but have fallen from grace. I am more impressed by how much trust simply appears to follow the economy and thus will ebb and flow over time. I am also a little cautious simply because we don’t have any similar measures of trust before 1958; perhaps trust was much lower before then and the 1950s and early 1960s were unusual.
For all these reasons, I remain skeptical of Lilla’s argument, but, with more data, LFC’s point may be confirmed: even with comparable levels of economic growth, trust is stubbornly lower than it was in this earlier period. Indeed, Keele believes that the broader decline in social capital—which he finds is strongly related to trends in trust—may ultimately be the culprit:
Scholars have long debated what caused the slide in trust from the 1950’s to today. Many explanations have been offered but all have been found wanting. Social capital provides the best evidence yet. First, it declined over the same period. Second, the evidence, here, demonstrates that social capital exerts a powerful effect on trust. So while poor government performance undoubtedly contributed to the slide, the loss of social capital must have been decisive. The slide in social capital also explains why trust has never rebounded. Without some resurgence in social capital, especially civic engagement, we cannot expect trust in government to return to its previous levels.
Time will tell. In the meantime, I’ll continue to emphasize that political trust goes both up and down over time, and that these trends are strongly tied to the economy. I do this not because that’s the end of the story but because these are two basic facts that conventional wisdom—a version of which Lilla presents—fails to recognize.
{ 10 comments }
If you’re looking for secular decline, why not just plot trust against years?
See:
http://yfrog.com/1rtrustkj
Of course, the two aren’t mutually exclusive. One can have a general decline in trust, along with cyclical variations. In fact, I think that a long-term declining trend in trust is quite obvious in your data, aside from the cyclical variations. I’d like to see an analysis going both ways — long term trend (isn’t 50 years considered “long term”?) in trust holding economic conditions constant, and quantify the influence of economic conditions in trust of government despite the (hypothetical) long term decline.
I wonder if the GSS has better data on this. Wish I had the time.
Vincent: If you click through the links above, you’ll see that I do plot trust against years. It shows the lack of a secular decline to which I refer, and which your graph also illustrates.
James: All true. I haven’t seen the exact analysis that you suggest. Wish I had the time too. Incidentally, the GSS data isn’t as good. It starts only in 1973 and then doesn’t ask any questions about trust in “government,” only confidence in each of the three branches of gov’t.
Can you post the data file you produced the scatterplot from? Or alternatively the results of a regression with DV trust and IVs year and growth?
Jeff: Happy to run that regression. Just a reminder, since the details are in earlier posts: the growth measure is Hibbs’. If I regress trust on that measure and year, both are significant:
growth: b=5.2, se=1.8
year: b=-0.5, se=.2
However, a good bit of the effect of year is due to especially high levels of trust in 1958. If you delete 1958 from the dataset, the effect of the economy is larger and the effect of year declines by almost half and becomes marginally significant, if one wants to go by conventional p-values (b=-0.30, se=.17).
So from this back-of-the-envelope analysis, I wouldn’t rule out the possibility that trust has declined over time, other things equal. There is suggestive evidence. But there is much stronger evidence that trust ebbs and flows with the economy.
In other words, I’d say that the strength of each variable (economy and year) is in inverse proportion to the ink spilled discussing them. Pundits love to discuss what they perceive as the trend (i.e., the alleged decline) but not so much the dynamics and their relationship to the economy.
I don’t agree with your casual disposal of that inconvenient year 1958. After all, 8 years later “trust in government” was even higher, and 1966 was before the major escalation of the war, the Tet Offensive, the revelation that the government was cooking the war numbers, the anti-war demonstrations, on and on. See Vietnam War > Events
I’d posit that if you had pre-1958 numbers, trust in government would be at comparable levels. You may be too young to remember the 50′s and 60′s, but I’m not. It was a much different time, my friend. As I remember it, and I was a young adult at the time, 1967 was a year with much the same impact on the national psyche as 2001. There was before, and then there was after.
So I wouldn’t be so quick to discard the little data point that ruins your hypothesis, though I understand and agree with your major point.
Here is some more robust data from Pew. http://people-press.org/trust/
You will see that data points between 1958 and 1966 show the same high level of trust in government, and even excluding everything before 1970, you still get a declining long-term trend.
J.S.:
Thank you for your reply to my comment; the earlier graph that you reproduce in this post – and that I hadn’t recalled – is helpful, as is the rest of the post.
To the extent that we may still differ a little on emphasis (and the data, as I gather, allow a bit of room for that), my perspective here is not that of a scholar — since my field is not American politics — but rather that of a politically aware citizen who lived through the period in question (I was only a year old in ’58, but I remember, e.g., the ’72 and ’76 presidential campaigns well, and the ’68 campaign to some extent also).
Re what you say at the end of the post: One of the scholar’s functions is, of course, to point out weaknesses in the conventional wisdom, so you’ll get no quarrel from me on that.
Yes, it appears I jumped the gun on this one. The graphs in your earlier post were clearly more than adequate. Thanks for pointing them out to me, sorry for the negligence, and thanks for your great work on the monkey cage!
James: I am not “discarding” a data point, merely pointing out that it has significant leverage on the inferences we can draw from these limited data. I linked to the Pew data in my first post on Lilla:
http://www.themonkeycage.org/2010/06/explaining_the_origins_of_the.html
They show the same trend in the 1950s and 1960s because in these early years they are based on the exact same data that I draw on (from the National Election Studies).
Comments on this entry are closed.