final essay Professional Writting

When the Smoke Clears: Focusing Events, Issue Definition, Strategic Framing, and the Politics of Gun Control∗

Anthony K. Fleming, University of West Georgia

Paul E. Rutledge, University of West Georgia

Gregory C. Dixon, University of West Georgia

J. Salvador Peralta, University of West Georgia

Objective. This article explores the strategic nature of framing following a focusing event. We argue that focusing events serve as catalysts for bill introductions along three particular causal stories prevalent in gun control policy: restrictive, punitive, and lenient. Methods. We employ a negative binomial regression model to investigate the effect firearm focusing events have on restrictive, lenient, and punitive bills introduced in both the House and the Senate. Results. Focusing events lead to an increase in restrictive, punitive, and lenient bills introduced in the House. In the Senate, however, focusing events lead to an increase in the number of punitive and lenient bills, while having no significant impact on the number of restrictive bills. Conclusion. This represents an increase in attention to gun control policy regardless of the causal story.

Firearm focusing events have become an unfortunate part of the American experience over time. Examples of these events include, mass shootings at schools and colleges, such as Columbine High School in Colorado and Virginia Tech University in Virginia; assassina- tions of prominent leaders, such as President John F. Kennedy and Dr. Martin Luther King Jr.; and assassination attempts such as the shootings against George Wallace and Ronald Reagan. Although relatively rare, these events can become catalysts for policy entrepreneurs to advance their preferred policy solutions in the form of bill proposals. The purpose of this article is to explore the impact of firearm focusing events on attention to gun control policy. We argue that focusing events serve as catalysts for agenda attention through bill introductions fitting three particular causal stories: restrictive, punitive, and lenient. In the aftermath of a focusing event, policy entrepreneurs use these causal stories to exploit the opening of a “policy window” (Kingdon, 1984) as well as to strategically counter opposing issue definitions. These efforts are made in order to redefine the gun control debate at a time of heightened public attention, which could potentially lead to issue redefinition and eventually policy change.

Using a negative binomial regression model, we investigate whether firearm focus- ing events lead to an increase in restrictive, lenient, and punitive bills introduced in both the House and the Senate. We empirically test for the effect of focusing events on the type of frame used in bill introduction by using frames in bills as a mea- sure of strategic framing. We find that focusing events lead to an increase in restric- tive, punitive, and lenient bills in the House of Representatives. This represents an

∗Direct correspondence to Anthony K. Fleming, Department of Political Science, University of West Georgia, 1601 Maple St., Carrollton, GA 30118 〈[email protected]〉. SOCIAL SCIENCE QUARTERLY, Volume 97, Number 5, November 2016 C© 2016 by the Southwestern Social Science Association DOI: 10.1111/ssqu.12269

When the Smoke Clears 1145

increase in attention to gun control policy regardless of the issue definition. In the Senate, however, focusing events lead to significant increases in the number of puni- tive and lenient bills, while having no significant impact on the number of restrictive bill introductions.

These findings are important because we make theoretical and methodological con- tributions to help answer the gridlock puzzle in gun control policy. While our findings suggest that focusing events have a strong impact on attention to gun control, the lack of policy change noted in previous research provides further evidence of the difficulty that is imposed by institutional friction in the policy process for attention leading to change. This is especially true in a salient and divisive area such as gun control.

Focusing Events, Agenda Setting, Framing, and the Gun Control Debate

Focusing events, similar to the mass shootings discussed above, frequently lead to a public outcry for policy change. In the process of agenda setting, the degree of systemic attention to a given problem changes dramatically. Important political questions are often the subject of scant coverage for years, only to then become the subject of featured attention across virtually all media outlets. External shocks to the system tend to impact all actors involved in the policy process simultaneously, causing rapid attention change in each of them rather than in only one or a few (Baumgartner and Jones, 1993). Focusing events, and the associated shifts in attention, tend to have a mobilizing effect on even the previously apathetic. Focusing events can thereby expand the scope of conflict (Schattschneider, 1960), engaging many who were previously relegated to the sidelines of a policy debate. As the scope of conflict changes significantly, majorities can coalesce around a given issue definition, which can lead to sweeping and rapid policy change.

Focusing events are important to policy making for two reasons. First, they can serve to focus attention on a problem. Second, focusing events can affect problem definition in combination with other similar events, making previous issue definitions being advanced by policy entrepreneurs more or less plausible as a result. The issue definitions that result from a focusing event or series of similar events can result in major shifts in governmental attention, as well as sweeping policy change. As Baumgartner and Jones (1993) demonstrate, policy windows open due to shocks in the system, which can allow for significant changes in policy. Rapid, sweeping change is frequently the result of changing policy definitions, the attachment of new symbols, and positive feedback.

Politicians, interest groups, and policy entrepreneurs are left scrambling to respond to the issue very quickly, attempting to attach solutions to problems. Kingdon (1984:94–95) describes focusing events as a “crisis or disaster that comes along to call attention to the problem . . . a powerful symbol that catches on.” The issue is already on the minds of policy entrepreneurs who are focused on a policy area, and the event serves as a potentially powerful symbol that can open a window for policy change to occur (Kingdon, 1984). Policy entrepreneurs who are focused on a given policy area will advance their preferred policy solutions as causal stories (Stone, 1989), hoping to attach the solution to the problem while the window for policy change is open.

Birkland (1997), in his study of focusing events such as natural disasters, oil spills, and nuclear power plant accidents, defines a focusing event as:

an event that is sudden, relatively rare, can be reasonably defined as harmful or revealing the possibility of potentially greater future harms, inflicts harms or suggests potential harms that are or could be concentrated on a definable geographical area or community

1146 Social Science Quarterly

of interest, and that is known to policy makers and the public virtually simultaneously. (Birkland, 1997:21)

Haider-Markel and Joslyn (2001), focusing on public support for gun control frames, argue that focusing events are a major driving factor in the framing of issues involving guns. As their work makes clear, following mass shootings, policy entrepreneurs spring to action in advocacy of a preferred causal story in recognition of the opening of a policy window (Kingdon, 1984) to shape public opinion. Issue framing, in this case, refers to the causal stories advanced by entrepreneurs to increase salience and promote a particular definition, interpretation, or policy solution, which in turn become the way in which those events are understood by both the public and political elites (Gamson, 1992; Entman, 1993; Haider-Markel and Joslyn, 2001). Political debate, and the likelihood of policy change, is determined by which of the issue frames successfully expands the scope of conflict (Schattschneider, 1960).

In summary, focusing events occur suddenly, news of them spreads very quickly and nationally, affect a large number of people, and are relatively rare. Focusing events are capable of getting agenda attention for an item, but this attention does not guarantee policy success. In the case of gun control policy, Fleming’s (2012) qualitative analysis demonstrates that while firearm focusing events lead to punctuated attention to gun control, policy change is extremely limited and rare. This emphasizes the need for examination of framing in the gun control literature because, unlike many studies of framing that focus on effectiveness, there is considerable resistance to framing in gun control policy (Haider- Markel and Joslyn, 2001). In the analysis that follows, we build on the work of Haider- Markel and Joslyn (2001) on framing the public debate by analyzing the quantitative impact of focusing events on congressional attention to gun control, separating bill introductions into three separate frames posited by Fleming, which will be described in more detail below.

Gun Control Issue Frames

The restrictive issue definition is defined as bills that make it more difficult to obtain a firearm. Examples of this bill type are restrictions on assault rifles, restrictions on handguns, restrictions on ammunition, taxes on ammunition, waiting periods, and any other bill that makes it harder to own or use a firearm. The principal argument here is that taking guns out of circulation, or preventing an increase in guns available, will lead to fewer gun crimes.

The punitive issue definition focuses on the individual rather than the gun, punishing those who use firearms in an illegal manner. The primary examples of this bill type are limits on convicted felons from obtaining firearms and increased penalties for using a firearm in the commission of a crime. This issue definition is based upon providing a deterrent. Stronger penalties imposed on those who commit a crime with a gun will serve as a deterrent for individuals to commit a crime with a gun.

The lenient issue definition is defined as bills that make it easier to obtain or own a firearm. Lenient examples include making illegal firearms legal, eliminating taxes on firearms, allowing firearms to be carried in locations they are banned, and any other legislation that makes it easier to own or obtain a firearm. This issue definition is focused on making it easier to obtain a gun, assuming that law-abiding citizens who carry guns will be able to stop firearm focusing events from happening. Groups advocating this issue definition argue that an individual who wishes to do harm with a gun will be deterred from doing so if the victims are well armed.

When the Smoke Clears 1147

Theory and Hypotheses

This article takes into account every gun bill proposed in Congress between the years of 1947 and 2010. Following Fleming (2012), these bills were coded as being restrictive, punitive, or lenient. The purpose of this article is to demonstrate the impact that gun violence focusing events have had on congressional attention to gun control. We use bill proposals as our measure for agenda attention, which we code separately for each of the issue definitions advanced in the gun control debate (Fleming, 2012). Our analysis is separate for the House and Senate because of the salience of the gun control debate and the differing diversity of opinions among the constituents served by members of each chamber.

We theorize that more punitive and restrictive gun bills will be proposed during the year in which the focusing event took place or in the year immediately following the focusing event. The heightened sense of awareness to the issue drives members of Congress into proposing legislation, seeing the event as an opportunity that opens a policy window. The exception to this is the Senate, where the diverse nature of the statewide constituency as compared to the possibly homogenous House districts will lead senators to be less likely to introduce restrictive bills than their House counterparts. Finally, we argue that lenient bills will be proposed at similar rates whether there has been a gun violence event or not. We argue this because Republicans and their allies in the National Rifle Association (NRA) will push to repeal or make it easier to own a firearm with policy proposals at regular intervals that are not guided by focusing events. In fact, it may be strategically beneficial for the NRA and gun proponents to remain relatively quiet, allowing time to pass following a focusing event (Fleming, 2012).

Thus, in our next section, we test the following four hypotheses:

1. The presence of a focusing event will have a significant impact on the number of restrictive bills introduced in the House.

2. The presence of a focusing event will have no significant impact on the number of restrictive bills introduced in the Senate.

3. The presence of a focusing event will have a significant impact on the number of punitive bills introduced in both chambers of Congress.

4. The presence of a focusing event will have no significant impact on the number of lenient bills introduced in either chamber of Congress.

Data and Method

Data on focusing events is taken from Anthony Fleming’s (2012) data set of gun-related focusing events in the United States. This data set includes all firearms-related focusing events of national importance in the United States for the years 1947–2010. The 12 focusing events included in our analysis are listed in Table 1. From these data a series of variables were generated for use in this analysis. A dichotomous variable was created to mark the presence or absence of a focusing event in a given year. This variable is coded a “1” for each year in which a focusing event took place and “0” for each year in which no focusing events took place. A second variable was generated to show the number of years since the prior focusing event took place. As Birkland (1997) notes, the policy learning and euphoria for change decay over time. This variable is a count of the number of years that have passed since the immediately prior focusing event. In the year of the focusing event it is coded as zero. A third variable was generated to account for the potential that the impact of focusing events could be additive. Birkland (1997) also indicates that multiple focusing

1148 Social Science Quarterly

TABLE 1

Firearm Focusing Events

Firearm Focusing Event Date

JFK Assassination November 22, 1963 Texas Bell Tower Shooting August 1, 1966 MLK Assassination April 4, 1968 RFK Assassination June 6, 1968 Wallace Shooting May 15, 1972 Easter Sunday Massacre March 29, 1975 Reagan Shooting March 30, 1981 Stockton Massacre January 17, 1989 Killeen Texas October 16, 1991 Columbine April 20, 1999 D.C. Sniper Shootings October, 2–22, 2002 Virginia Tech April 16, 2007

events of a similar nature within a short amount of time could have an additive effect on policy attention to an issue. This variable counts the number of focusing events that have occurred in the years covered by the data. This is a simple count of the number of focusing events that have taken place in previous years.

Data on the proposed legislation come from the congressional Bills Project data set (Adler and Wilkerson, 2014) and include all bills proposed in both the House of Representatives and the Senate in each year. These data include the author of each bill and the date the bill was proposed, for a total of 1,130 House bills and 259 Senate bills. The bills are divided into three categories of bills following Fleming (2012): restrictive, punitive, and lenient using our coding to reflect the nature of the bill. Some legislation can appear in multiple categories, such as an attempt to roll back gun control legislation while increasing the punishments for crimes committed with a firearm. This legislation is coded in both of the categories reflected in the legislation for the purpose of generating our counts of the number of proposals. For example, a piece of legislation that was both lenient and punitive would be included in the count for lenient and the count for punitive in that year.

To control for the reduced congressional workload in election years, we include a dummy variable for election years. During election years there are significantly fewer bills proposed than in nonelection years. To account for this we create a dichotomous variable coded as “1” for election years and “0” for nonelection years.

We include data related to the partisan makeup of the House and Senate as well as the partisan loyalty of the president as controls for the partisan makeup of government. These data are taken from the websites of the respective houses of Congress (〈www.House.gov; www.Senate.gov〉). For the House and Senate, we include a variable that reports the percentage of each house that is made up of Democrats. Separate variables are calculated for the House and for the Senate. The partisanship of the president is measured with a dichotomous variable, with “1” being a Democratic president and a “0” being a Republican president. Partisanship is a potential explanation for the number of bill proposals. If the chamber of Congress is controlled by the Democratic Party, the likelihood that a restrictive bill would pass is greater than in a chamber controlled by the Republican Party. Similarly, if proposers consider the likelihood of presidential support, the partisanship of the president may have an impact on the number proposed.

A dichotomous dummy variable is included to account for the “Cincinnati Revolt” within the NRA. This meeting of the NRA in 1977 led to a purge of more moderate

When the Smoke Clears 1149

leadership and a strong turn toward opposition to all forms of gun control (Davidson, 1998). This changed the leadership and goals of the largest organization opposed to gun control and could potentially impact the fate of relevant legislation in Congress. This variable is coded a “1” for all years after 1977 and a “0” for 1977 and all years before that.

Data on violent crime in the United States were derived from the FBI Unified Crime Reporting Project (UCR). These data were compiled from the UCR Online Data Tool and are based on violent crime data for the United States as a whole (U.S. Department of Justice, 2013a). Two variables were taken from these data: the overall violent crime rate and the murder rate. The violent crime rate is expressed as the number of violent crimes per 100,000 persons in the population. Violent crime in this measure includes murder (and nonnegligent manslaughter), rape, armed robbery, and aggravated assault (U.S. Department of Justice, 2013b). The violent crime rate is expected to have a positive and significant relationship with congressional attention to gun control because when the violent crime rate increases, citizens are likely to place increased pressure on Congress for gun control legislation. The murder rate is similarly expressed as the rate of murders per 100,000 persons in the population.

It is possible that the murder rate has a separate influence from the overall rate of violent crime, so we control for its independent effects. In addition, it is likely that the effects of violent crime and murder are conditional on each other. Murder may matter less when violent crime is high and more when it is low. To test this effect, an interaction term was generated as the product of the violent crime rate and the murder rate.

The result of this is an annual data set that includes the above data for all 63 years (1947–2010).1 To test our hypotheses we generated a series of models that test the impact of focusing events on the number of proposed bills of various types. All models reported here are negative binomial regression models. The negative binomial model is the appropriate model for count data that are characterized by overdispersion (Long, 1997; Greene, 2011). A series of tests were performed using a variety of models including the Poisson model and the generalized negative binomial model. These tests indicated that a standard negative binomial model was the most appropriate.2 The results of these models are presented in a series of tables that are divided by the type of legislation. Table 2 shows the results for models using restrictive legislation. Table 3 shows the results for the models using punitive legislation. Finally, Table 4 shows the results for the models using lenient legislation. Each table groups the models by the relevant house of Congress. Two models are included for each type of legislation in each house of Congress to allow for the inclusion of two alternate specifications of the focusing event effects. Model 1 refers to models that do not include a cumulative impact of focusing events (the count of prior events). Model 2 refers to models that count the total number of focusing events.

The inclusion of the violent crime/murder rate interaction term as a control leads to some differences in the tables. The interaction term for violent crime and murder rates was not significant in a number of the models. In Table 2, the interaction term is omitted entirely as it was not significant in any of the models shown.3 In Tables 3 and 4, the models in which the interaction term was not significant include “n/a” in the interaction term row. As the interaction term is included among the controls rather than the variables of interest, we do not discuss these differences in detail.

1Missing data in the crime rate reduce the effective number of observations to 51 years. 2All models and tests were estimated in STATA, with replication data and do-files available upon request. 3The results as well as the replication data and code for the various models with insignificant interaction

terms are available upon request.

1150 Social Science Quarterly

TABLE 2

Impact of Focusing Events on Proposed Restrictive Legislation

House of Representatives Senate

Model 1 Model 2 Model 1 Model 2

Focusing event 1.013∗∗∗ 1.030∗∗∗ 0.293 0.293 (0.28) (0.27) (0.29) (0.29)

Years since last focusing event

−0.076∗∗ −0.125∗∗∗ −0.030 −0.029 (0.04) (0.04) (0.04) (0.05)

Percent Democratic −5.565∗∗ −6.911∗∗∗ −1.300 −1.292 (2.47) (2.47) (2.71) (2.74)

President’s party 0.722∗∗∗ 0.633∗∗ 0.318 0.319 (0.25) (0.25) (0.27) (0.27)

Election year—yes/no? −0.717∗∗∗ −0.732∗∗∗ −0.633∗∗∗ −0.633∗∗∗ (0.23) (0.22) (0.25) (0.25)

Violent crime rate 0.004 0.006∗∗ 0.007∗∗ 0.007∗∗ (0.00) (0.00) (0.00) (0.00)

Murder rate −0.263 −0.479∗∗ −0.633∗∗∗ −0.630∗∗∗ (0.17) (0.20) (0.19) (0.24)

Cincinnati Revolt −1.108 −0.662 −1.109 −1.114 (0.68) (0.69) (0.85) (0.88)

Total number of focusing events

n/a −0.271∗∗ n/a 0.003 (0.13) (0.15)

Constant 5.922∗∗∗ 8.109∗∗∗ 3.243∗ 3.220 (1.30) (1.62) (1.67) (2.01)

N 51 51 51 51 χ 2 35 40 31 31

NOTE: Results are negative binomial regression coefficients with standard errors included in parentheses. ∗p > 0.1, ∗∗p > 0.05, ∗∗∗p > 0.01.

Results

As expected, the models below demonstrate the strong impact of focusing events on restrictive and punitive bill introductions in the House of Representatives. Contrary to expectations, however, focusing events also have a positive and significant impact on lenient bill introductions in the House. The impact of focusing events in the Senate is more mixed. While focusing events are associated with increased punitive and lenient bill proposals in the Senate, they are not significantly associated with restrictive legislation.

The results for restrictive bills are presented in Table 2. The table includes two models for each house of Congress that include control variables to account for the partisan makeup of the House, the party of the president, the leadership of the NRA, the presence of an election, and the crime and murder rates. Model 1 includes the presence of a focusing event and the time since the last focusing event. Model 2 extends this by adding the total number of focusing events to see if there is an accumulating impact as more events take place.

Across both models, the focusing event variable is statistically significant and positive for the House. The presence of a focusing event significantly increases the number of proposed restrictive bills in the given year. The number of years that have passed since the last focusing event reduces the number of proposed restrictive bills. The findings in both House models for the restrictive issue definition are consistent with expectations set forth in hypothesis 1. In the Senate, the focusing event variables are not statistically significant, consistent with hypothesis 2. Models 1 and 2 show the same result: neither the focusing event nor the time

When the Smoke Clears 1151

TABLE 3

Impact of Focusing Events on Proposed Punitive Legislation

House of Representatives Senate

Model 1 Model 2 Model 1 Model 2

Focusing event 0.824∗∗∗ 0.822∗∗∗ 0.831∗∗ 0.674∗∗ (0.28) (0.28) (0.34) (0.31)

Years since last focusing event

−0.140∗∗∗ −0.144∗∗∗ −0.121∗ −0.027 (0.05) (0.05) (0.07) (0.07)

Percent Democratic 0.910 0.998 −1.167 −5.492∗ (2.45) (2.46) (3.78) (2.88)

President’s party 0.822∗∗∗ 0.817∗∗∗ 0.271 0.341 (0.24) (0.24) (0.32) (0.31)

Election year—yes/no? −1.159∗∗∗ −1.162∗∗∗ −1.259∗∗∗ −1.168∗∗∗ (0.22) (0.22) (0.29) (0.28)

Violent crime rate 0.008∗ 0.009 0.015∗ −0.004 (0.00) (0.01) (0.01) (0.00)

Murder rate 1.262∗∗∗ 1.264∗∗∗ 0.991∗∗ 0.584∗∗∗ (0.26) (0.26) (0.49) (0.20)

Crime/murder interaction term

−0.001∗∗∗ −0.002∗∗ −0.002∗ n/a (0.00) (0.00) (0.00)

Cincinnati Revolt −0.150 −0.120 0.118 −0.714 (0.47) (0.47) (0.74) (0.64)

Total number of n/a −0.049 n/a 0.488∗∗∗ focusing events (0.15) (0.16)

Constant −5.645∗∗ −5.818∗∗ −6.537 −0.583 (2.62) (2.66) (5.27) (2.01)

N 51 51 51 51 χ 2 69 69 35 41

NOTE: Results are negative binomial regression coefficients with standard errors included in parentheses. ∗p > 0.1, ∗∗p > 0.05, ∗∗∗p > 0.01.

since the last event is significant. In Model 2, we do see that the total number of focusing events is significant and negative, indicating that the greater the number of past events, the fewer restrictive bills will be proposed in a given year.

Negative binomial regression coefficients can be somewhat difficult to interpret directly. To express the impact of focusing events in more practical terms, Table 5 illustrates the change in the predicted counts of proposed bills in years with and without focusing events. Looking at the House of Representatives, in years with focusing events we would expect 12 restrictive bills to be proposed. In years with no focusing events this number is six bills. The presence of a focusing event roughly doubles the number of restrictive bills introduced in the House of Representatives in a given year.

Partisan effects show a series of interesting results. The partisan makeup of the House shows a statistically significant impact, but not the expected one. This relationship is significant and negative across both models. The more Democrats in the House, the fewer restrictive bills are proposed. In the Senate, the results are also consistent across both models, but in this case the results are not statistically significant, indicating that there is no significant effect of the changes in the percentage of Democratic senators. The partisan effect of the president is consistent across the models and is in the expected direction for the House. For the Senate, the president’s party is not statistically significant. The presence

1152 Social Science Quarterly

TABLE 4

Impact of Focusing Events on Proposed Lenient Legislation

House of Representatives Senate

Model 1 Model 2 Model 1 Model 2

Focusing event 0.670∗∗∗ 0.633∗∗∗ 0.576∗ 0.577∗ (0.23) (0.21) (0.32) (0.32)

Years since last focusing event

−0.090∗∗ −0.040 −0.143∗∗∗ −0.101∗ (0.04) (0.04) (0.05) (0.06)

Percent Democratic −1.925 −2.294 −0.157 0.444 (2.16) (2.02) (2.88) (3.00)

President’s party −0.031 0.046 −0.034 0.059 (0.22) (0.21) (0.31) (0.32)

Election year—yes/no? −1.177∗∗∗ −1.154∗∗∗ −0.872∗∗∗ −0.846∗∗∗ (0.19) (0.18) (0.27) (0.27)

Violent crime rate 0.010∗∗ −0.001 −0.001 −0.004 (0.00) (0.01) (0.00) (0.00)

Murder rate 1.038∗∗∗ 1.020∗∗∗ −0.132 0.139 (0.25) (0.23) (0.16) (0.21)

Crime/murder interaction term

−0.002∗∗∗ −0.001∗ n/a n/a (0.00) (0.00)

Cincinnati Revolt 0.513 0.175 0.571 0.244 (0.46) (0.43) (0.77) (0.76)

Total number of focusing events

n/a 0.383∗∗∗ n/a 0.264∗∗

(0.11) (0.13) Constant −2.973 −2.217 2.125 0.092

(2.46) (2.31) (1.80) (2.08) N 51 51 51 51 χ 2 47 61 23 27

NOTE: Results are negative binomial regression coefficients with standard errors included in parentheses. ∗p > 0.1, ∗∗p > 0.05, ∗∗∗p > 0.01.

TABLE 5

Mean Predicted Number of Bills Introduced

House of Representatives Senate

No Focusing Focusing No Focusing Focusing Event Event Event Event

Restrictive legislation 6 12 2 3 Punitive legislation 9 17 1 2 Lenient legislation 4 6 1 2

NOTE: Predicted number of each type of bill under the specified conditions (rounded to the nearest whole number).

of a Democratic president increases the number of restrictive bill proposals in the House, but not in the Senate.

The controls showed interesting results for the restrictive models. The election year dummy variable was significant and negative as expected in all models. In election years, members propose fewer bills than they do in off years. This relationship held for both houses of Congress (as well as across all three bill types). The overall violent crime and murder rates were not statistically significant in Model 1. In Model 2, the violent crime

When the Smoke Clears 1153

rate did have a significant and positive relationship to the number of bills proposed, but the murder rate had a significant and negative impact on the number of restrictive bills proposed. Similarly, the “Cincinnati Revolt” of 1977 in which the NRA leadership became more dogmatic did not have a significant impact on the number of restrictive bills that were proposed.

Looking at punitive legislation we see that the relationships follow a different pattern to the one we observe in restrictive bills. Focusing events increase the number of punitive bill proposals in the House and the Senate, as demonstrated in Table 3. The results for the focusing event dummy are consistently positive and significant in both houses of Congress and in both models for each house, consistent with the expectations posited in hypothesis 3. Table 5 shows the substantive impact of focusing events on punitive legislation: the predicted number of punitive bills in the House in a year without a focusing event is nine. For a year with a focusing event, the predicted number of punitive bills is 17, slightly less than twice the number of proposed bills. In the Senate a similar magnitude can be observed as a result of a focusing event, with the number of bills proposed rising from one to two. Focusing events lead to a clear increase in the number of punitive bills proposed in both houses. In the House, the number of past focusing events was not significant, but in the Senate, the total number of past events was significant and positive.

The impact of our control variables was different for punitive bills than for restrictive ones. Most noticeably, the violent crime and murder rate variables are significant in the punitive models, as is the interaction term. This shows that there are conditional effects of the two rates on the number of punitive bills proposed. The negative coefficient for the interaction term indicates that the impact of each rate decreases as the value of the other rate rises.4 The Cincinnati Revolt variable is not significant in any of the four models. The partisan variables are mixed across the models. In the House, the percent of the House that is Democratic is not significant in either model. The presidential partisanship variable is significant and positive as with restrictive bills. In the Senate, a greater percentage of Democratic senators leads to a lower number of proposed punitive bills in Model 2. In Model 1, the partisanship was not significant. The partisan allegiance of the president is not significant in the Senate.

Focusing events also have a significant impact on the number of lenient bill proposals, contrary to expectations posited in hypothesis 4 above. Table 4 summarizes the results of the models for lenient bills, and it is clear that lenient bills follow a similar pattern to that observed in their punitive and restrictive counterparts. In the House, the presence of a focusing event is associated with a higher number of lenient bill proposals. Interestingly, the years since a focusing event has taken place is also associated with a reduced number of bill proposals in most models. The total number of focusing events is associated with an increased number of lenient bill proposals, indicating that more firearm focusing events over time have played a role in the lenient issue definition gaining more traction in Congress.

In the House, a year with a focusing event has six bill proposals, while a year without one has four proposals. This is a 50 percent increase in proposed lenient bills in a year with a focusing event. In the Senate, the familiar doubling of the number of bills from one to two is observed.

The election year variable was associated with a strong reduction in the number of bills proposed in a given year. The remaining control variables were inconsistent in their effects and significance. The partisan variables were not significant either for the president’s

4As these are included as controls, we do not include detailed information on the conditional relationship. We merely note the results.

1154 Social Science Quarterly

party or for Democratic control in either house. Violent crime and murder rates were significant and positive only in Model 1 in the House. The murder rate was significant in Model 2 for the House. The violent crime/murder interaction was significant in the House, but not in the Senate. The Cincinnati Revolt was not significant in any of the models.

On the whole, the results show strong support for the role of focusing events in the rate of bill proposals for firearms-related legislation. The occurrence of a focusing event doubles the number of predicted restrictive bill proposals in a given year in the House of Representatives. In the House, the number of years that have passed since the last focusing event leads to a decrease in both restrictive and punitive bills proposed. Con- trary to expectations, the results for lenient bill proposals in the House follow a similar pattern to the results for restrictive and punitive issue definitions, indicating the gun pro- ponents are surprisingly active following focusing events in advancing the lenient issue definition.

In the Senate, focusing events increase the number of proposed punitive and lenient bills. The time since the last focusing event reduces the number of lenient bills proposed in the Senate. The results clearly demonstrate that focusing events have a significant impact on bill introductions of each type of frame, with the exception of restrictive legislation in the Senate. Again, this conforms to expectations (hypothesis 2) because of the far more diverse constituency served by senators, and the salience of the gun control debate, which could potentially jeopardize reelection for senators much more than counterparts in the House. Further, attention to gun control tends to decrease as time passes and focusing events fade from the public memory.

Discussion

In this article, we have investigated the impact that firearm focusing events have on bill introductions in each chamber of Congress across three causal frames—restrictive, punitive, and lenient—that are prevalent in the gun control debate. Our findings demonstrate that Congress is responsive to focusing events in the agenda-setting stage. Consistent with expectations, both the restrictive and punitive issue definitions are advanced in House bill introductions as a response to focusing events. Consistent with expectations, the Senate does not respond to focusing events with restrictive bill proposals. As we suggest above, this could be a result of the more diverse constituency coupled with the salient nature of the gun control debate. Senators put themselves particularly at potential electoral risk in a statewide constituency by advocating the restriction of access to guns.

Additionally, contrary to expectations posited in hypothesis 3, both the House and the Senate tend to see an increase in lenient bill introductions following focusing events. The results also demonstrate that the cumulative impact of firearm focusing events over time has been to increase the traction associated with the lenient frame, with more bills introduced that fit the lenient frame. As tragedies continue to occur, the lenient frame seems to become an increasingly plausible legislative proposal. This signifies the work of policy entrepreneurs such as the NRA advancing their proposed issue definitions and the associated policy solutions when policy windows are opened as a result of a focusing event. Our evidence is also generally supportive of Birkland’s (1997, 2006) argument that focusing events are a critical factor in determining the level of attention devoted to a policy area.

For scholars of gun control and framing, our findings are a critically important addition to the literature because we provide quantitative analyses of the three competing issue definitions that have governed the politics of gun control for the last half century plus. As

When the Smoke Clears 1155

Haider-Markel and Joslyn (2001) note, gun control is a critically important policy area for the examination of framing. In a literature where many previous studies have focused on issues with lesser conflict over frames, this article has demonstrated the dominance, tension among, and resistance to three competing frames in the gun debate over more than a half century. This article again emphasizes the importance of policy entrepreneurs’ advancement of issue frames when policy windows are open, specifically following the stimulus of focusing events (Kingdon, 1984; Birkland, 1997).

Mass shootings and assassinations clearly lead to an increase in attention to gun control policy, as well as the public demand for change. Focusing events attach powerful symbols, such as dead children, to a drive for policy change. Despite the powerful nature of this symbol, increased attention to restrictive, punitive, and lenient legislation has led to limited policy change on the national level. Meanwhile, state governments have become increasingly active in adopting laws that fit each of these frames following focusing events. More work is to be done to understand the impact that focusing events have on attention to and adoption of gun control legislation in the states. This is particularly important given the NRA’s aggressive response to the school shooting in Newtown, Connecticut, signaling an increased push for lenient legislation.

This article has demonstrated that framing effects are issue dependent, which calls for future research on an expanded number of policy areas with competitive issue frames and differing public salience. Further, both the gun control literature and the framing literature could benefit from an expanded focus on governmental institutions and nongovernmental actors in the advancement of frames, specifically following focusing events. For example, how does the president respond in language and proposals to advance frames and causal stories following focusing events? Specifically, in the case of gun control, does the president tend to advance the same frames as we find among members of Congress? Similar to Haider-Markel and Joslyn (2001), we contend that the contribution of this article is in furthering a literature regarding the importance of focusing events and the framing effects that follow, but is by no means the last word on the subject.

REFERENCES

Adler, Scott E., and John Wilkerson. 2014. Congressional Bills Project: (1947–2010), NSF 00880066 and 00880061.

Baumgartner, Frank R., and Bryan D. Jones. 1993. Agendas and Instability in American Politics. Chicago: University of Chicago Press.

Birkland, Thomas. 1997. After Disaster: Agenda-Setting, Public Policy, and Focusing Events. Washington: George- town University Press.

———. 2006. Lessons of Disaster: Policy Change After Catastrophic Events. Washington: Georgetown University Press.

Davidson, Osha Gray. 1998. Under Fire: The NRA and the Battle for Gun Control. Iowa City: University of Iowa Press.

Entman, R. 1993. “Framing: Toward Clarification of a Fractured Paradigm.” Journal of Communication 43(4):51–58.

Fleming, Anthony. 2012. Gun Policy in the United States and Canada: The Impact of Mass Murders and Assassinations on Gun Control. New York: Continuum Publishing.

Gamson, William A. 1992. Talking Politics. New York: Cambridge University Press.

Greene, W. H. 2011. Econometric Analysis, 7th ed. Upper Saddle River, NJ: Pearson Publishers.

1156 Social Science Quarterly

Haider-Markel, Donald P., and Mark R. Joslyn. 2001. “Gun Policy, Opinion, Tragedy, and Blame Attribution: The Conditional Influence of Issue Frames.” Journal of Politics 63(2):520–43.

Kingdon, John W. 1984. Agendas, Alternatives, and Public Policies. Boston: Little, Brown.

Long, S. J. 1997. Regression Models for Categorical and Limited Dependent Variables, Thousand Oaks, CA: Sage Publishers.

Schattschneider, E. E. 1960. The Semi-Sovereign People. Hinsdale, IL: Dryden Press.

Stone, Deborah. 1989. “Causal Stories and the Formation of Policy Agendas.” Political Science Quarterly 104(2):281–300.

U.S. Department of Justice. 2013a. Uniform Crime Reporting Data Tool. Available at 〈http://www. ucrdatatool.gov/〉. Accessed December 30, 2013. ———. 2013b. Uniform Crime Reporting Statistics—UCR Offense Definitions. Available at 〈http://www. ucrdatatool.gov/offenses.cfm〉. Accessed December 30, 2013.

Copyright of Social Science Quarterly (Wiley-Blackwell) is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.