An exploration of spillover effects: evidence from threat-induced education reform

ABSTRACT

This study explores spillover effects between national security and education issues. I examine whether the rise of new foreign threats is correlated with education issues rising onto the congressional agenda and with the passage of education laws. To answer this question several data sources on military conflicts and congressional activity from 1947–2015 were combined. I estimate the relationship between changes in deployed troops, the presence of education issues on the congressional agenda, and the passage of education laws. I find that total troop changes were a significant predictor of the passage of education legislation. This relationship is partially mediated by education issues rising onto the congressional agenda. The arrival of a foreign threat appears to create a shock that raises national security issues and subsequently education onto the congressional agenda. These findings help to explain delays to policy maintenance and the focus on short term goals in education reform.

KEYWORDS:

• Policy attention
• policy agenda
• education policy
• spillover
• agenda-setting

Disclosure statement

No potential conflict of interest was reported by the author.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Notes

1. In Stimson’s polling data the mean education public support is 67 and it never fell below 61 (Stimson 2016a). Polling data on education reform (available from 1956 to 2010) bear out Davies’ hypothesis (2007) about the popularity of increasing spending.

2. From this point education issues refers exclusively to primary and secondary education issues.

3. Adler and Wilkinson (2017) collected data from 1947 to 2014. I appended data from 2015, when the Every Student Succeeds Act passed.

4. The subject of congressional bills was coded using the Policy Agenda Codebook (Policy Agendas Project 2015e). I restricted primary and secondary education laws to the following subcodes: 600-Education General, 602- Elementary and Secondary Education, 603- Education of Underprivileged Students, 607- Educational Excellence.

5. Two education laws were passed in five years: 1965, 1969, 1977, 1990, and 1993.

6. Binder (2003) used unsigned editorials written by the entire board.

7. Tables 2 and 4 show that the association between hearing days lagged and passaged of an education law is a precise zero after conditioning on covariates. The results are quite similar when including other variables measuring characteristics of hearings (i.e. total hearings, number of sessions, length of hearings in days, both lagged and not lagged).

8. I restricted the gridlock data to include only issues with macro code 6 for education.

9. UCDP/PRIO defines an armed conflict as ‘a contested incompatibility that concerns government and/or territory where the use of armed force between two parties, of which at least one is the government of a state, results in at least 25 battle-related deaths in a calendar year’ (UCDP/PRIO 2016).

10. In the DoD report it appears that the Korean War began in 1951. The DoD created the report prior to the start of the Korean War in June 1950. The troop change value in 1950 was changed to reflect this fact. All conflict start dates were validated with multiple sources (Torreon 2017; UCDP/PRIO 2016).

11. Congress formally revoked authorization for force in Vietnam, which President Nixon subsequently escalated.

12. The national security issues measure was imputed in 2015 using the predicted outcome from regressing the national security issues measure on a linear year trend. The mediation results have similar sign, size, and significance when the observation for 2015 is case-wise deleted and if the variable is excluded from the model.

13. The correlation between the two measures is low (r = −0.001). Ideally this variable would measure nonviolent events like the Soviet launch of Sputnik or the Cuban missile crisis. However, this variable takes on a below average value in the years in which these events took place.

14. State education revenues are measured in tens of billions of dollars. State education revenues were not available for five years (1947, 1949, 1951, 1953, 1955). In these years state education revenues were imputed to be the average of the adjacent years. The mediation results are qualitatively similar both when these years are excluded and when the covariate is removed from the model.

15. Henig’s data are available through 2010. Following his coding strategy (Henig 2013) I identified education governors in subsequent years (2011–2015).

16. All SAT scores were converted to the recentered SAT scale (The College Board 2018).

17. I took the average of the polarization variable for the Senate and House, which is the ‘Difference in Party Mean,’ for the first dimension DW-Nominate score and rescaled the variables to have a theoretical range from −100 to 100.

18. Linear probability models and logistic regressions ought to produce equivalent estimates if the outcome event is neither rare nor common (i.e. probability of outcome is approximately between 0.2 and 0.8) (Hellevik 2009). Results from logistic regressions have similar sign, size, and significance (see Table B1).

19. One potential concern with time series data is the presence of an autoregressive process where residuals are correlated across time. If the passage of an education law was correlated with education issues rising onto the congressional agenda in year t and year t-1, the estimates would be biased. To test for the presence of auto-correlation I estimated Durbin-Watson test statistics for the main models. Each were arbitrarily close to 2, which suggest there was no correlation between period t and period t-1.

Additional information

Notes on contributors

Joshua Bleiberg

Joshua Bleiberg studies standards-based reform, education policy, and governance at the Leadership, Policy, and Organizations at Peabody College, Vanderbilt University.