The following is a guest post by GW Ph.D. student “Jake Haselswerdt”:http://www.gwu.edu/~psc/people/stu_haselswerdtj.htm.
*****
Last week, in reaction to “this Rasmussen poll”:http://www.rasmussenreports.com/public_content/politics/general_state_surveys/wisconsin/wisconsin_poll_support_for_budget_cutting_not_for_weakening_collective_bargaining_rights of Wisconsin voters, Josh Marshall “wrote”:http://www.talkingpointsmemo.com/archives/2011/03/rasmussen_walker_flailing.php:
bq. [W]hen you look at this poll the pro-union / anti-union division turns heavily on age too. The younger votes have a far more progressive views on public employees, unions, collective bargaining and so on…[F]or those of us with somewhat longer political memories this is actually pretty different from the way things used to be in the ’80s and ’90s. Back then it was older voters who tended to have more Democratic views on bread and butter economic issues. And it was younger voters who had more libertarian inclinations.
bq. There’s a fairly straightforward explanation for this. Back then older voters had formative political experiences from the 30s through the 50s. Today they’re based in the 60s through the 80s. Straightforward doesn’t mean it’s accurate. But I think that’s at least part of the explanation.
bq. There’s an argument from conservatives that this doesn’t represent a trend. It’s just that as you get married and have kids and have to handle a family budget, you develop more conservative economic values. And I think there’s definitely something to that argument. But again, I don’t think history really bears it out.
Marshall discusses a question that scholars of political behavior have studied for years: what is the nature of the relationship between age and political attitudes (in this case, attitudes toward labor unions)? Is it best understood as a generational effect (with age cohorts differing from one another consistently over time due to different formative experiences) or a life cycle effect (in which political attitudes shift predictably as a cohort ages)?
Luckily, the American National Election Studies have been collecting data on Americans’ attitudes toward labor unions since 1964, which helps us answer this question. Specifically, ANES respondents are asked to rate labor unions on a 100-point “feeling thermometer,” with higher scores representing warmer feelings toward unions. The figure below shows the aggregate trend in the mean thermometer score over time (note that data are missing for 1970, 1978, 1982 and 2006).
At least in the aggregate, Americans’ attitudes about labor unions appear surprisingly stable over time. There are spikes and dips to be sure, but the mean thermometer score never falls below 55.3 (1986) and never rises above 67.2 (1972). Although union membership may have declined precipitously in recent decades, Americans’ favorable feelings toward unions have not; in fact, the mean thermometer scores in 2004 and 2008 (around 63) are actually higher than the scores in the 1960s (between 57 and 59). Of course, this graph tells us nothing about life cycle or generational effects. The following graph overlays the trend lines for the four most recent age cohorts coded by the ANES. Note that the y-axis on this graph is truncated so the differences between cohorts are more visible.
First, there appear to be generational differences here, though they are not always especially pronounced. During and after the late 1990s, for example, the 1975-1990 cohort (roughly the “under 40” voters that Rasmussen found to be most supportive of the Wisconsin Democrats) is the most enthusiastic about unions, followed by the 1959-1974 and 1927-1942 cohorts. (The large spike for the latter in 2008 may be a statistical anomaly, since there were only 70 respondents in this cohort in the 2008 survey.) The 1943-1958 cohort is the least enthusiastic, at least until 2008. The modestly more conservative attitude toward unions among the 1943-1958 cohort relative to the 1927-1942 cohort supports Marshall’s memory regarding the economic conservatism of younger people during the 1980s and 1990s (though this cohort was hardly “young” by the 1990s).
Second, the cohorts trend together much of the time; Americans of different ages usually react to the same events in the same way. Generational differences may account for differences in underlying baseline opinion toward unions (the separation between the trend lines), but current events clearly play a powerful role in shaping reported opinion at any given time. This casts doubt on the notion of a life cycle effect, which is based on the idea that events specific to the age cohort (e.g. moving into the workforce or retiring) lead to predictable changes in political attitudes. The graph shows little evidence of such a pattern. For example, if we focus on just the 1927-1942 and 1943-1958 cohorts (the blue and yellow lines), it is difficult to see any pattern in the former’s labor union sentiments that is echoed a decade or two later by the latter. Instead, the lines move together simultaneously with few exceptions. Even if we smooth out the fluctuations induced by current events using a lowess smoother, as in the following graph, there is little evidence of a predictable life cycle pattern.
The evidence seems to square with Marshall’s intuition: generational effects are more important than life cycle effects in explaining the attitudes of different age cohorts toward labor unions. Still, even the generational effects are modest in most years.