Abstract
Scholars and journalists have shown that US state legislators often copy and paste policy text from other sources. This ‘policy plagiarism’ is perceived by critics as symptomatic of process failures and likely to undermine policy success. To proponents, copy and paste legislation stems from an efficient learning process likely to guarantee policy success. The authors test competing hypotheses by measuring success and plagiarism across three areas of US state policy: organ donation legislation, e-cigarette/vaping bans for minors and anti-bullying legislation. They find that higher levels of plagiarism result in significantly less success at reducing youth vaping rates and increasing organ donor registrations. They also find a negative, though not significant, relationship between copying and success for antibullying policy. The evidence favours opponents: legislators risk harming policy success by copying from others. This study of policy plagiarism advances knowledge by moving beyond the simple demonstration of the phenomenon to investigate the potential link between the copying of legislative text and the extent to which the policies studied achieved their goals.
‘when we see copycat legislation…oftentimes it’s clearly not customized to what’s going on in Oklahoma.’ State Representative Emily Virgin
‘If you don’t like the fact that [Oklahoma is] ranked so low in so many things why wouldn’t we want to look around and see what other states are doing…and try to analyze that for application here?’ State Representative Mark Lepak
In 2019, the Arizona Republic, USA Today, and Center for Public Integrity jointly published the results of a two-year investigation into the use of model legislation in the American states titled ‘Copy, Paste, Legislate’. This award-winning report found that over 10,000 bills introduced in state legislatures had been copied from interest group authored model bills during an eight year period (O’Dell and Penzenstadler, 2019). Lawmakers admitted to relying on bills written by others, as one legislator claimed he had ‘signed on to 72 such bills without knowing or questioning their origin’ (O’Dell and Penzenstadler, 2019). The report concluded that ‘fill-in-the-blank bills’ may be ‘supplant[ing] the traditional approach of writing legislation from scratch’ (O’Dell and Penzenstadler, 2019).
Many journalists, legislators, and academics argue that plagiarising policy from other sources is a sign of process failures that undermine the ability of the legislature to achieve policy success. Policy success occurs when the adopted policy is effective at achieving or making substantial progress on stated goals.1 Policy success exists on a continuum that ranges from complete programmatic success – meaning the policy fully achieves its desired outcomes – to outright programmatic failure – meaning the policy fails to improve (and perhaps even worsens) the targeted conditions – with shades of partial success arrayed between (Bovens et al, 2001; McConnell, 2010). The amount of success achieved along this continuum may be limited when policy is plagiarised because the policy may not have been updated, expanded, and customised, which could lead to ineffective implementation (for example, Boushey, 2010; Hertel-Fernandez, 2014; Bologna, 2019; Jansa et al, 2019; McVan, 2019; O’Dell and Penzenstadler, 2019; Hansen and Jansa, 2021). Further, copy and pasting could be a sign that the legislature relied on shortcuts provided by wealthy interest groups rather than crafting legislation with significant input from affected constituent groups (Hertel-Fernandez, 2014). A process marked by such characteristics is indicative of a legislature that may have failed to carefully design and assess policy before making the decision to adopt, leading to less policy success.
Yet, copy and paste lawmaking can also be viewed as an efficient learning process likely to deliver policy success. Defenders of the practice argue that mimicking policies from elsewhere stems from legislators engaging in an efficient search for effective policy. This learning process results in the emulation of policies written by expert lawmakers and interest groups that have been thoroughly tested in other jurisdictions (for example, England, 2019; Ladner, 2019; Sergent et al, 2019).
Our goal is to bring clarity to the debate over whether copy and paste lawmaking hinders or helps achieve policy success. We offer competing hypotheses that extend from the ongoing debate. If proponents are correct, then copying should lead to more policy success, especially for states with fewer legislative resources and for policies that are meant to create uniformity across the states. If critics are correct, then copying should lead to less policy success. We also apply the literature on bureaucratic professionalism to offer an alternative hypothesis: the potential negative impact of plagiarised policy on success can be moderated during implementation by a professional bureaucracy.
We test these hypotheses by measuring success and plagiarism across three policies nearly universally adopted by the states: organ donation legislation, e-cigarette/vaping bans for minors, and anti-bullying legislation. Policies with higher plagiarism scores tend to exhibit less specificity and be less customised than policies with lower plagiarism scores. As a result, these copied policies also exhibit less success; higher levels of plagiarism on state’s vaping bans and organ donor laws resulted in significantly less success at reducing youth vaping rates and increasing donor registrations, respectively. We also find negative, though not significant, relationship between plagiarism and success for anti-bullying laws. The results show that legislators risk undermining policy success if they shortcut the policy process by copying and pasting policy written elsewhere.
Emulation, learning and policy success
One of the primary mechanisms of policy diffusion is emulation, or the copying of policy nearly word-for-word from another source (Mooney, 2020).2 Scholars have found that emulation occurs when lawmakers imitate policies emerging from innovator states (for example, Desmarais et al, 2015), take cues on how to write policy into law from policy experts and advocacy networks (for example, Garrett and Jansa, 2015), or seek an efficient means to policymaking in limited time and resource environments (for example, Hansen and Jansa, 2021). Unlike other diffusion mechanisms such as learning, emulation can occur without legislators engaging with information about the political or programmatic success of the policy being emulated (Gilardi and Wasserfallen, 2019). Instead, legislators emulate to keep up with emerging policy norms, feeling the pull to imitate esteemed innovator states, prestigious policy experts, and trusted advocacy networks (Gilardi and Waserfallen, 2019; Mooney, 2020).
As a result of this process, emulated policy is likely to exhibit little to no change in bill text (Walker, 1969; Mooney, 2020). This contrasts with the learning diffusion mechanism. If lawmakers are engaged in learning, they will encounter information that helps them understand the changes needed to a policy to ensure its success; thus, policies that diffuse via the learning mechanism are likely to be reinvented, leading to significant changes in text (Hays, 1996; Jansa et al, 2019).
Yet, the scholarly community does not have a clear sense of the consequences of emulation and learning for policy success. The learning process more closely adheres to the ideal policy process outlined by scholars of policy success, beginning with the foundational work of Lasswell (1971). Successful process involves gathering information, defining problems, engaging with stakeholders, predicting implementation difficulties and customising policy, which should yield successful policies (for example, Lasswell, 1971; Bratton and Ray, 2002; Preuhs, 2007; McConnell, 2010; Johannes, 2015). Indeed, information about programmatic success often forms the basis of lawmakers’ decisions during the learning process, leading to policy adoption and reinvention cascades. Emulating legislators, though not without information, are more likely to copy policies based on political or normative signals, resulting in little reinvention – a hallmark of a successful policy process.
Sources of plagiarism in American legislatures
Although its empirical relationship with policy success remains obscure, emulation has garnered increased attention as it come to light that legislators sometimes copy interest group model bills. Interest groups write and promote model legislation in order to influence policy across jurisdictions, which are used by legislators looking for effective and efficient means of policymaking. The emulation of model bills first drew widespread attention during the trial of George Zimmerman for the shooting of Trayvon Martin. Zimmerman’s defence rested on Florida’s Stand Your Ground statute which allows individuals to engage in deadly self-defence. The law was written in consultation with the American Legislative Exchange Council (ALEC), which then used Florida’s law as a model bill to influence similar policies in over 30 states (Garrett and Jansa, 2015; DeMora et al, 2019).
Indeed, ALEC has been identified as a ‘super interest group’ (DeMora et al, 2019) and ‘dangerously effective’ (Hertel-Fernandez, 2019) due to its seeding emulation across the states. In an authoritative study of the group, Hertel-Fernandez (2019) finds that tens of thousands of ALEC-sourced bills were introduced in state legislatures over a 20-year period with some states – such as Arizona – accounting for nearly 1,000 introductions and hundreds of enactments (Hertel-Fernandez, 2019). Mostly conservative legislators introduced ALEC-authored bills enacting policies such as Stand Your Ground, voter identification, Right-to-Work, and challenges to Medicaid expansion across the states (Garrett and Jansa, 2015; Hertel-Fernandez et al, 2016; Hertel-Fernandez, 2019; DeMora et al, 2019). Journalists echoed these findings, observing that model legislation has affected ‘nearly every area of public policy’ (O’Dell and Penzenstadler, 2019).
Copying model bills, though, is just one way in which legislators emulate policy word-for-word across the states. In fact, the phenomenon of emulation was first noted by Walker (1969) when he observed that a 1931 California law had been copied by ten states, typos and all. In a recent study of the sources of state laws that challenge federal law, Callaghan, Karch and Kroeger (2020) find that emulation is more likely to stem from states copying one another than interest group model bills: in challenges to federal gun control rules, about 10% of bills were sourced from model bills while nearly 47% were sourced from other states. The biggest predictor of word-for-word emulation is legislative professionalism; specifically, legislators with few staff resources are more likely to copy policy text from other states (Jansa et al, 2019; Hansen and Jansa, 2021; Linder et al, 2020) and interest group model bills (Hertel-Fernandez, 2014; 2019).3
Potential benefits of copy and paste lawmaking
Even if overstating the prevalence of interest group model bills, the ‘Copy, Paste, Legislate’ report sparked debate as to whether word-for-word emulation is evidence of a broken process doomed to failure or a resourceful process likely to yield success. Defenders of the practice have noted that it helps legislators efficiently use their time and resources. One of the defining characteristics of state legislatures is variation in resources, with some states having well-funded professional staffs for full-time legislators and other states having part-time staffers shared between citizen legislators (for example, Squire, 2017). Scarce resources can constrain policy success if legislators do not have the capacity to research and tailor policy for their state’s needs. One Oregon legislator noted such, saying ‘We have such limited staff that [ALEC] helps us look at things and consider them’ (Hertel-Fernandez, 2019: x). In this way, copy and pasting is an efficient path to policy success for statehouses operating on limited resources.
Lawmakers may also benefit from copying policy because they desire to implement identical solutions. This may be the case if states face similar problems and seek vetted solutions to those problems (for example, Volden, 2006), or if legislators want to provide uniformity for individuals and businesses across lines of jurisdiction (England, 2019; Ladner, 2019; Sergent et al, 2019). In the case of the former, copying keeps lawmakers from having to ‘reinvent the wheel if somebody else has already been successful’, as one lawmaker put it (Sergent et al, 2019). In the case of the latter, copying is ‘a good use of resources to make sure from one state to the other the laws are similar’ (Sergent et al, 2019).
Copying also provides experts a central role in shaping policy across the states. Lawmakers cannot divine what other states are doing and instead rely on cues from expert policy entrepreneurs and lobbyists about which policies work and how to write them into law. Though ALEC gets the most attention, a variety of interest groups, professional associations and think tanks use model bills to spread policy ideas and information (for example, Callaghan et al, 2020). As one legislator noted, ‘we get approached all the time by different uniform code commissions and get model bills from different legislative organizations… and we go to these conferences during the summer in off session times and hear about model legislation’ (Sergent et al, 2019). Though model bills are not the only source of copying, they are written to be generalisable and applicable to different situations (Garrett and Jansa, 2015). Legislators, as political tacticians and policy generalists, welcome prefabricated policy as shortcuts to success, allowing them to shift focus away from drafting and toward other legislative activities. In this way, organisations provide not only a valuable subsidy to legislators, but also a more effective policy than if the legislator crafted it from scratch.
Potential drawbacks of copy and paste lawmaking
Others have expressed concern over copy and paste methods, noting its potential to have a negative impact on policy success. First, copying signifies a lack of policy customisation. Early adopting jurisdictions tend to create narrow, experimental policies (for example, Hays, 1996). As these experiments diffuse, later adopters reinvent policy by learning from previous adopters’ experiences. Through this learning and reinvention process, policies become more comprehensive and, therefore, robust to different scenarios and contexts (for example, Glick and Hays, 1991; Mooney, 2020). If lawmakers opt to copy and paste instead, they may be importing narrow policies rather than improving upon what has been learned through experimentation. Scholars have implied that this could have negative repercussions for policy success, though these claims have not been tested. For example, Boushey argues that quickly copying from others can mean policies are ‘unsuitable for their immediate problems’ (2010: 171) while Jansa, Hansen and Gray note that it could reflect ‘ignorance of the exact implications of a bill’s content for one’s own state’ (2019: 741).
Critics also note that lawmakers often fail to fully vet copied policies before they are introduced, leading to critical implementation problems. For example, Indiana had to revise its controversial Religious Freedom Restoration Act (RFRA) when it became clear that provisions it had borrowed in pieces from others states created a structure of legal discrimination against LGBT citizens (Cook et al, 2015). Lawmakers recognise the need for careful vetting, even if it does not always happen. As Missouri State Representative Justin Hill said ‘To take up a bill and just blindly say, ‘OK, I’m going to sponsor this’ – that’s not good practice’ (McVan, 2019). Hill’s colleague State Representative Derek Grier also noted that model bills should be made ‘more specific to Missouri’ before being passed into law (McVan, 2019). But these quotes came on the heels of Missouri lawmakers passing an ALEC-sourced bill that referenced the wrong federal statute (Lieb, 2014). Ultimately, drafting errors can lead to problems with and protracted battles over implementation as lawmakers are forced to deal with the consequences of ‘the politics of haste’: unworkable legislation (Lewallen, 2016).
Critics also condemn copy and paste lawmaking as an indicator that the underlying process is unrepresentative of key constituent groups. Scholars have noted that citizens may not be getting ‘as much specialized representation’ from ‘solutions developed elsewhere’ (Hansen and Jansa, 2021). Copying model bills produced by wealthy interest groups could mean that legislators are responding to ‘the preferences of organized business and upper-income citizens, rather than the preferences of lower-income citizens’ contributing ‘to a representational imbalance’ (Hertel-Fernandez, 2014). Some opponents have even claimed that the emulation of model bills allows ‘deep pocketed outsiders’ to have more influence in public affairs (Bologna, 2019). The ‘Copy, Paste, Legislate’ report succinctly expresses this concern ‘You elected [legislators] to write new laws. They are letting corporations do it instead’ (O’Dell and Penzenstadler, 2019).
Representative deficits are alarming because policy success is often achieved via input from broad constituencies and their representatives. When legislators represent the unique perspectives of large constituent groups, they can produce more innovative and successful policies that are broadly accepted by the public (Bratton and Ray, 2002; Preuhs, 2007). Lawmakers can better anticipate possible implementation problems and stave off failure in representative policy processes (Bovens et al, 2001; McConnell, 2010). As Johannes explains, ‘a truly responsive and representative government is likely to be judged effective, since by definition it mirrors the views and characteristics of its people and has considered and acted on their wishes and welfare’ (2015: 9–10). Copy and paste methods that are overly responsive to narrow interests could harm actual and perceived effectiveness of policy.
Competing hypotheses for effect of copying on policy success
H1a) Policy plagiarism leads to policy success that is equal to or greater than other states.
H1b) Policy plagiarism leads to policy success that is equal to or greater than other states, especially for states with less professional legislatures.
H1c) Policy plagiarism leads to policy success for policies that are meant to create uniformity across the states.
H2) Policy plagiarism leads to less policy success than other states.
H3) Policy plagiarism leads to less policy success than other states, though this relationship is moderated by the professionalism of the state’s bureaucracy.
Measuring success across three different policies
To test these predictions, we collect data on the degree of policy success achieved by states across three different policies: e-cigarette/vaping bans for minors, anti-bullying legislation and organ donor legislation. These policies were chosen for several reasons. First, the policies were nearly universally adopted by states over a similar timeframe (1999–2017). Specifically, all 50 states adopted vaping bans for minors and anti-bullying legislation, while 46 states adopted organ donor legislation. The diffusion patterns provide temporal variation within each policy – since nearly every state adopted but at different timepoints – as well as a similar sampling frame to facilitate pooling data across policies. This allows us to examine the impact of whether the policy was adopted, the timing of the adoption, and the amount of policy language copied on the success of the policy. Further, these policies’ universal adoption means that they are less likely than other policies to be adopted in pursuit of ideological agendas, or as a result of legislators attempting to emulate the political successes of their peers. Indeed, the correlations between a state’s government ideology4 and the decision to adopt (r = -0.08), the rank order (out of 50 states) in which the state adopted (r = 0.01), and the amount of text copied from previous adopters (r = 0.01) are extremely small, indicating little relationship between ideology and the diffusion dynamics of these three policies. Second, each of these policies has been previously studied in the context of copy and paste lawmaking (for example, Hansen and Jansa, 2021), making the texts of each state’s law readily available for operationalising our key independent variable policy plagiarism. Third, each policy has a clearly associated goal with publicly available data that tracks the achievement of that goal. This is critical for measuring our key dependent variable policy success. Finally, the inclusion of organ donor laws in particular allows us to test the claim that plagiarism ensures success on policies designed to create uniformity (H1c). Below we define each policy and describe its associated measure of policy success.
E-cigarette/vaping bans for minors
E-cigarette/vaping bans seek to decrease youth consumption of vapour products by prohibiting the sale and consumption of vaping products to people under the age of 18. New Jersey was the first state to adopt a youth vaping ban in 2009 with all states following suit by 2017.5 We use Center for Disease Control (CDC) Youth Risk Behavior Surveillance System (YRBSS) survey data to calculate the change in the percentage of high schoolers who report using an electronic vapour product at least once in the past 30 days.6 This survey question was asked in 2015, 2017 and 2019 and the CDC reports these survey results at the state-level for each of these years.7 Using 2015 as our baseline, we are able to calculate the state-level, biannual per cent change in high school vaping rates for 2017 and 2019.
Anti-bullying legislation
Anti-bullying laws provide guidance and resources to school districts for development of bullying prevention and intervention plans for public schools in hopes of reducing instances of bullying in school. Georgia first adopted this policy in 1999 with all 50 states adopting some form of anti-bullying policy by 2015. We use CDC YRBSS data to track success on anti-bullying laws. The CDC asks highschoolers if they were bullied on school property in the previous 12 months and the CDC reports state-level results.8 This data is available every two years from 2009 to 2019. We use the first cycle (2009) as our baseline and calculate the biannual per cent change in school bullying rates for each state from 2011 to 2019.9
Organ donor legislation
Organ donor laws create pathways for state residents to become organ donors and provide legal rights to organ donors. The goal of the legislation is to increase the availability of life-saving tissue and organs for patients in need. National Conference of Commissioners on Uniform State Laws (NCCUSL) authored a model bill for introduction by legislatures in order to create uniformity in organ donor regulations across states. Following the commission’s recommendations, many states began adopting in 2007. By 2013, 46 states had adopted the policy; the remaining four states – New York, Delaware, Pennsylvania and Florida – had not adopted by 2020. We measure policy success as change in the percentage of the state population over the age of 18 that are registered organ donors. This data is tracked by Donate Life America, a nonprofit organisation that promotes organ donation across the US. Using 2012 as a baseline, we measure per cent change in donor registration every two years from 2013 to 2017 for each state.
After collecting policy success data for each policy, we re-coded the variable so that for each policy higher values indicated greater success and lower values indicated less success. This meant recoding the per cent change in bullied students and per cent change in youth e-cigarette users, for which negative values would represent greater success, to per cent change in nonbullied students and per cent change in youth nonvapers, for which positive values would represent greater success. Thus, positive coefficient estimates in the forthcoming analyses can be interpreted as the predictor leading to greater success, even though success may ultimately be the reduction bullying or vaping conditions.
Measuring policy plagiarism
For each state’s policy, we measure policy plagiarism as the amount of legislative language (or text) that was borrowed from previous adopters. Specifically, we take the highest Smith-Waterman alignment score between the text of a state’s policy and the text of the policies from all previous adopters.10 The Smith-Waterman algorithm is used to match sequences of words and has been previously used to measure similarity between congressional and state bill texts (Wilkerson et al, 2015; Linder et al, 2020; Hansen and Jansa, 2021). Specifically, the algorithm calculates an alignment score by rewarding matched words and penalising mismatched words and gaps.11 If two texts have more matched sequences and fewer gaps, they will have a higher alignment score, indicating more plagiarism. There is substantial variation in alignment scores overall and across all three policies, as shown in the online supplemental material.12
Textual similarity scores like the Smith-Waterman alignment pick up on substantive differences in policies; bills that are textually similar enact policies in similar domains, move the policy status quo in a similar direction, and contain similar characteristics such as the level of specificity and comprehensiveness (Linder et al, 2020). Most importantly, policies that are highly similar to previous adopters have been shown to be less (re)inventive, meaning the text contains fewer new, customised provisions (Hansen and Jansa, 2021).
To illustrate, take Vermont’s and Arizona’s respective vaping bans. An excerpt from each is shown in Figure 1. Vermont adopted a vaping ban on minors in 2014 that borrowed significantly from Minnesota’s law, extending the state’s existing penalties for individuals and businesses that sell tobacco products to minors to include electronic products. The alignment score for the Vermont vaping policy is 49.0. In contrast, Arizona adopted a vaping ban on minors in 2013 that used original language to define ‘vapor products’ and include them in a comprehensive list of banned products for sale and consumption, establishing a $100 fine for minors who possessed e-cigarettes, and described conditions for offences with greater specificity. The alignment score for Arizona’s policy compared to previous adopters is just 27.6.
Figure 2 also helps illustrate, showing that antibullying laws with higher alignment scores (that is, more plagiarised) are much less likely to include new provisions. After the initial adoption of an antibullying law in Georgia in 1999, which defined bullying and cyberbullying, required school districts to develop antibullying policies, and provided training to detect and quell bullying, subsequent state adopters added provisions such as explicit protections against bullying students based on colour, appearance, academic performance, gender, sexual orientation, religion, health and socioeconomic status, created state boards to oversee the implementation of antibullying programmes in schools, and specified legal remedies should a student be subject to bullying. In this way, the alignment scores capture reinvention and specificity in a summary measure even as the policies move the status quo in the same direction.
Alignment scores by new provisions for antibullying policies
Citation: Policy & Politics 2022; 10.1332/030557321X16445954252430
Alignment scores by new provisions for antibullying policies
Citation: Policy & Politics 2022; 10.1332/030557321X16445954252430
Alignment scores by new provisions for antibullying policies
Citation: Policy & Politics 2022; 10.1332/030557321X16445954252430
Additional variables and modelling strategy
We measure a number of additional variables. To test the relationship between copying and success being conditional on legislative professionalism (H1b), we interact our alignment scores with legislative professionalism scores from Squire (2017) which is a summary measure of how professional each state legislature is compared to the US Congress in terms of salary, staff resources and session length. We also interact alignment scores with bureaucratic professionalism to test H3. Following Boushey and McGrath (2017), we measure bureaucratic professionalism by taking the average annual nominal salary across 50 different executive branch officials and creating a ratio of executive pay to legislative pay.13 For each of the professionalism measures, we use a state’s score at the time of policy adoption; for non-adopters we use the score from the first year in the data for each policy. We do this to capture the level of professionalisation when the policy was passed and began to be implemented.
We also include a dichotomous measure for policy adoption (1 for all state-years in which the policy was adopted and thereafter, 0 otherwise) since states that adopt a version of the policy are more likely to experience success than states that do not. We also control for problem severity, which is a lagged, raw version of the policy success variable. Higher values indicate higher youth vaping rates, higher bullying rates in school, and higher share of adults unregistered as organ donors. States facing more severe underlying conditions have more room for improvement and could see larger than average changes from year to year regardless of other factors. We control for time since adoption, measured in years since the policy was adopted by a state, to capture whether success is a function of how long a policy has been in effect. Finally, we control for a state’s logged population and per capita income in thousands of dollars. These slack resources should be associated with more success as it taps each state’s capacity to draw on human and economic capital to ensure improved outcomes. Descriptive statistics for the model variables are available in the online supplemental material.
We model the dependent variable, policy success, as a function of policy adoption, policy plagiarism, time of adoption, legislative professionalism, bureaucratic professionalism and problem severity using a pooled dataset that combines data from all three policies. This is possible because the dependent variable policy success is measured on the same scale (that is, per cent change) for all three policies. The pooled dataset, due greater degrees of freedom, provides greater statistical power for precisely estimating the effects of predictor variables on policy success. But the pooled dataset includes repeated observations for state-years (for example, Alabama’s success in 2015 at reducing vaping and Alabama’s success in 2015 at registering organ donors) which need to be accounted for in order to avoid biased coefficient estimates. To account for repeated observations of state-years, we estimate the model using a maximum likelihood multilevel model with state and year random effects (RE). In addition to controlling for repeated state-year observations, RE help account for unit and time heterogeneity across policies without masking the effect of ‘sluggish’ or nearly-time-invariant predictors such as legislative professionalism and policy plagiarism. This model (Table 1, Model 1) provides a direct test to H1a and H2 while subsequent pooled models that interact legislative professionalism with policy plagiarism test H1b (Table 1, Model 2) and interact bureaucratic professionalism and policy plagiarism test H3 (Table 1, Model 3). To test H1c we provide models of the individual policies (Table 2, Models 1–3) so we can compare the effect of policy plagiarism on policy success for organ donor legislation (a policy meant to create uniformity) to the other policies. These individual models also allow us to assess the degree to which the results in the pooled model are driven by the dynamics of one or more individual policies. Individual models are estimated using maximum likelihood with state RE.
Models of policy success
| Model 1 DV = policy success | Model 2 DV = policy success | Model 3 DV = policy success | |
|---|---|---|---|
| Policy adoption | 24.837* (8.588) | 24.493* (8.860) | 25.015* (8.584) |
| Policy plagiarism (alignment score) | –0.270* (0.074) | –0.282* (0.105) | –0.280* (0.075) |
| Bureaucratic professionalism | –0.010 (0.011) | –0.010 (0.011) | –0.024 (0.020) |
| Legislative professionalism | 6.339 (15.895) | 3.371 (24.610) | 6.403 (15.883) |
| Problem severity | 1.021* (0.106) | 1.022* (0.106) | 1.122* (0.106) |
| Time since adoption | 0.778* (0.359) | 0.774* (0.360) | 0.813* (0.361) |
| Population (logged) | –2.274 (1.716) | –2.276 (1.716) | –2.324 (1.715) |
| Per cap income ($1000s) | –0.219 (0.182) | –0.215 (0.184) | –0.217 (0.182) |
| Legislative prof x Alignment score | 0.066 (0.416) | ||
| Bureau prof x Alignment score | 0.000 (0.000) | ||
| Constant | –4.287 (27.759) | –3.523 (28.177) | 3.561 (27.751) |
|
|
|
|
|
|
|
|
Standard errors in parentheses
p<0.05
Models of policy success for individual policies
|
|
| |
|---|---|---|---|
| Policy adoption | 1.213 (6.260) | 41.313* (18.539) | 117.429* (38.432) |
| Policy plagiarism (alignment score) | –0.039 (0.051) | –0.426* (0.198) | –1.376* (0.501) |
| Bureaucratic professionalism | –0.015+ (0.008) | –0.012 (0.019) | –0.023 (0.029) |
| Legislative professionalism | –17.622+ (10.136) | –14.691 (28.464) | –0.986 (48.667) |
| Problem severity | 1.951* (0.281) | 1.308* (0.161) | 6.473* (0.809) |
| Time since adoption | 0.228 (0.199) | 0.439 (0.785) | –8.180* (3.056) |
| Population (logged) | 2.660* (1.130) | –6.406* (3.167) | 5.318 (5.265) |
| Per cap income ($1000s) | 0.262* (0.111) | 0.385 (0.326) | 0.097 (0.720) |
| Constant | –89.655* (20.324) | 20.847 (50.526) | –254.303* (102.519) |
| State RE | Yes | Yes | Yes |
|
|
|
|
Standard errors in parentheses
+ p<0.1, * p<0.05
Analysis and results
The results in Table 1 show that the most considerable influence on policy success is policy adoption (p<0.05 in 3 of 3 models) and problem severity (p<0.05 in 3 of 3 models); success is directly influenced by whether a state adopted a policy solution and how severe its problem was. A consistently positive and significant relationship emerges between time since adoption and policy success (p<0.05 in 3 of 3 models); as time elapses since a policy has been in effect, the state sees greater progress toward achieving its goals. Together, the results for policy adoption, problem severity and time since adoption provide face validity to our models.
With regards to our principle hypotheses, Table 1 shows that more copying is associated with less success. The relationship is consistently negative and significant. Table 1, Model 1 shows the baseline effect, which is a 1 point increase in alignment scores (on a 100 point scale) is associated with a 0.27 point decrease in policy success (on a 100 point scale). From one standard deviation below to above the mean on alignment scores, this seemingly small effect size equates to a 12.63% decrease in success, all else equal. This is a meaningful effect size that means worse outcomes for constituents. But it is important to note the effect is much smaller than policy adoption (having a policy is associated with about 24% more policy success than not having one) and problem severity (an increase from one standard deviation below to above the mean severity of conditions at t is associated with 31.5% more policy success at t+2).
Table 1, Model 2 shows that the relationship between copying and success is not significantly different at any level of legislative professionalism. Similarly, Table 1, Model 3 shows that there is no evidence to suggest that more professional bureaucracies ensure the success of plagiarised policy. If this were the case, we would expect a negative and significant parameter estimate for the interactive effect of bureaucratic professionalism and alignment scores, as strong agencies would be weakening the effect of copying on success. Instead, the estimated effect is extremely small and statistically indistinguishable from zero.
We re-estimated Model 1 for each individual policy and present the results in Table 2. Table 2 shows alignment scores are negative and significant for both organ donor laws and e-cigarette/vaping bans. Though not significant, the estimated relationship between plagiarism and success is still negative for antibullying laws as well. For antibullying laws, problem severity and slack resources in the form of larger populations and more affluent citizens predicts success on antibullying policy. Indeed, problem severity remains a positive, significant and consistent predictor of policy success across all three policy areas, while policy adoption is positive in all three models, though significant in only the organ donor and vaping models.
The effect detected in the pooled model is driven by negative effects of copying on success for all three policies, though mostly by the significant effects for organ donor and vaping policies. The size of the effect of copying on success is quite large for vaping bans; an increase of 1 point in a state’s alignment score corresponds to a 1.38 percentage point decrease in policy success. This means that a state that had an alignment score of 28.8 (or one standard deviation below the mean level of copying on vaping ban legislation) could expect about 25.9% more success in slowing or reducing youth vaping than a state that had an alignment score of 47.6 (or one standard deviation above the mean level of copying on vaping ban legislation).
To illustrate this effect, consider once again Vermont’s and Arizona’s respective vaping bans. Vermont adopted a vaping ban that borrowed from Minnesota and contained few new provisions, resulting in an alignment score of 49.0. In contrast, Arizona adopted a vaping ban that used original language instituting customised provisions for an alignment score of 27.6. Given the estimated effect size, we should expect Arizona to have substantially more success than Vermont in reducing youth vaping and that is what we observe: between 2015 and 2017 Arizona saw a 41.5% reduction in youth vaping (from 27.5% of high schoolers reporting using e-cigarettes in 2015 to 16.1% in 2017) while Vermont saw a 21.5% reduction in youth vaping (from 15.3% in 2015 to 12% in 2017). Even though Vermont has a less severe problem with youth vaping than Arizona, and the policies move the status quo in the same direction, we control for that and find that, all else being equal, there is greater success for states that plagiarised less when writing their policies into law.
Breaking out the model by individual policy also allows us to test whether copying leads to more success for policies striving to create uniformity than other policies (H1c). We cannot conclude that there is much support for this argument. There is a strong, negative relationship between copying and success for organ donor laws, which is the opposite of what proponents argued as the basis for H1c. Copying does not appear to help legislatures create successful, uniform policies. Indeed, the results lean decidedly in favour of critics’ hypothesis. Policy plagiarism is associated with less policy success (H2). There is also no evidence that less professional legislatures experience greater success by copying more (H1b) nor evidence for the alternative hypothesis that highly professional bureaucracies can aid policy success regardless of the level of policy plagiarism (H3). The failure to add specific, custom provisions and instead import policy whole-cloth is associated with lower success at achieving higher organ donor registration rates and lower youth vaping rates. Lawmakers that eschew customising these policies to fit the conditions in their state risk trading effectiveness for efficiency.
The only support for the argument that copying should lead to at least as much success as other states (H1a) relies on the antibullying laws model; statistically, there is no discernible difference between imitator and innovator states in their success at reducing school bullying. But this support relies on the lack of statistical significance and ignores the consistently negative direction of estimates across individual and pooled models. The spirit of proponents’ argument is that copying is a shortcut to success, but the consistently negative estimates, and significantly negative estimates for e-cigarette/vaping bans and organ donor laws, violates their theory of an efficient learning process.14
Implications and conclusion
Scholars, journalists and policymakers have divergent views on copy and paste lawmaking: some believe it is symptomatic of a broken policy process, while others believe it is emblematic of an efficient learning process. These perspectives highlight a gap in the policy diffusion and policy success literatures: the emulation and learning mechanisms – despite specifying different decision-making processes that lead to different policy texts – are assumed to have a similar effect on policy success. We find, though, that word-for-word emulation is not a shortcut to success. Rather, we find a consistently negative relationship between copying and success across three different policies and significant reductions in success for two of the three policies in particular.
Problem definition, constituent input and customisation are hallmarks of sound policy process, and it appears that plagiarising policy shortcuts this process. While the impact may not be felt on all policies, others may be significantly hampered. That said, it is important to keep the impact of copy and paste methods in context. Copying does not influence policy success as much as policy adoption and problem severity. The effect of having a policy versus not having a policy, and the severity of the conditions in the state, have a larger impact on success than the amount of text copied from other sources. But it is not enough to simply adopt a solution to a lingering problem as, in some cases, lifting policy language without careful reinvention can erode the policy’s impact.
Although we may expect that professional bureaucrats can aid legislatures by implementing policy successfully no matter the level of plagiarism, this appears to not be the case, based on evidence presented here. Further qualitative work could also illuminate how bureaucracies – no matter how well-resourced – react and respond to hastily copied legislation. It could be the case that important work is happening in agencies to fill in the gaps in emulated policies and ensure success that we cannot capture with bureaucratic professionalism. In addition to further exploration of the role of bureaucracy, there are dozens of additional policy areas to test. If, under repeated testing across policies, there emerges a consistently negative and sometimes significant impact of copying on success, opponents’ concerns will be more definitively confirmed. We also do not evaluate whether copying leads to more mistakes or errors in drafting policy. Future research could look at whether incidences of error are higher in emulated rather than reinvented policies. While our work helps to clarify the debate over copy and paste lawmaking by identifying and testing its arguments for the first time, we certainly will not have the last word.
The broadest implication of our work is that legislators’ pursuit of efficiency in the policymaking process has tradeoffs. Using copy and paste methods to gain efficiency may come at the cost of effectiveness. Indeed, policymakers risk the overall level of success of the policy by not taking time to vet, negotiate and customise it for their jurisdiction. Efficiency and effectiveness are key tenets of good government (Johannes, 2015) and trading one for the other is not a clear improvement in the quality of state governance. Our study also has implications for another tenet of good government: representation. Opponents worried that plagiarising policies would undermine their success, but also that the tendency to plagiarise signalled better access and representation for well-resourced interest groups and policy entrepreneurs than for stakeholder groups within the state. In this way, waning success is a byproduct of the unrepresentative and truncated copy and paste policymaking process.
Notes
This definition is akin to programmatic success in McConnell’s (2010) theory of policy success.
Emulation, copy and paste lawmaking, policy plagiarism and other similar terms and phrases will be used interchangeably to refer to the same phenomenon: copying policy nearly word-for-word from another source.
While Republicans are more likely to introduce ALEC bills than Democrats, this could be specific to emulation of ALEC than it is of use of copy and paste techniques overall (for example, Hertel-Fernandez, 2014; 2019).
We use the NOMINATE-based government ideology scores calculated by Berry et al (2010), measured at the time a state adopted a policy. States that did not adopt were assigned the average ideology score over the policy’s observed time period.
Massachusetts did so by judicial action rather than legislative action, and is thus not included in the analyses.
The CDC asks ‘Have you used an electronic vapor product (including e-cigarettes, e-cigars, e-pipes, vape pipes, vaping pens, e-hookahs, and hookah pens) at least once in the past 30 days?’
The data are available for most states in these years. If the CDC reported data for a state in 2015 and 2019 but not 2017, we use linear interpolation to fill in the missing value. We note which observations have been interpolated and test whether our results are sensitive to interpolation.
The CDC also asks high schoolers if they were electronically bullied through texting, Instagram, Facebook, or other social media, during the 12 months before the survey. We use this as an alternative measure of success and present the results in the appendix.
Like policy success for vaping laws, we interpolate where possible and test whether our results are sensitive to linear interpolation.
Due to some extreme alignment scores on bills that were very similar to previous adopters and lengthy, we take the natural log of the alignment score and divide it by the natural log of the highest possible alignment score for that policy. We then multiply this quotient by 100 to create scores scaled from 0 to 100.
The alignment score parameters for this application are +2 for matches, -1 for mismatches, -1 for gaps.
Nonadopters and first adopters were assigned alignments scores of 0 since, by definition, they did not copy text from another source. The results that follow are not sensitive to these low-score outliers.
Data for 1990–2010 was provided by Boushey and McGrath (2017) and updated to 2016 by the authors using Table 4.11 from the Book of the States (Council of State Governments, 2017).
The results are robust to the removal of the interpolated and extreme outlier observations on the policy success dependent variable. See online supplemental material.
Funding
This work was not supported by any outside funding.
Acknowledgments
The author(s) would like to thank Robert McGrath for his generosity in sharing data used in this work.
Supplemental
Data used in this work is available on the UNC Dataverse at https://doi.org/10.15139/S3/4NUGS7
Conflict of interest
The authors declare that there is no conflict of interest.
References
Berry, W.D., Fording, R.C., Ringquist, E.J., Hanson, R.L. and Klarner, C.E. (2010) Measuring citizen and government ideology in the U.S. states: a Re-appraisal, State Politics & Policy Quarterly, 10(2): 117–35. doi: 10.1177/153244001001000201
Bologna, G. (2019) Mississippi leads nation in filing legislation that other people wrote. Here’s why that matters, Ledger, https://www.clarionledger.com/story/ news/politics/2019/04/04/model-bill-investigation-mississippi-copycat-legislation-leads-nation/3204987002/.
Boushey, G. (2010) Policy Diffusion Dynamics in America, New York: Cambridge University Press.
Boushey, G. and McGrath, R.J. (2017) Experts, amateurs, and bureaucratic influence in the American states, Journal of Public Administration Research and Theory, 27(1): 85–103. doi: 10.1093/jopart/muw038
Bovens, M., ’t Hart, P. and Peters, B.G. (2001) Analyzing governance success and failure in six European states, in M. Bovens, P. ’t Hart and B.G. Peters (eds) Success and Failure in Public Governance: A Comparative Analysis, Cheltenham: Edward Elgar, pp 593–640.
Bratton, K.A. and Ray, L.P. (2002) Descriptive representation, policy outcomes, and municipal day-care coverage in Norway, American Journal of Political Science, 46(2): 428–37. doi: 10.2307/3088386
Callaghan, T., Karch, A. and Kroeger, M. ( 2020) Model state legislation and intergovernmental tensions over the affordable care act, common core, and the second amendment, Publius: The Journal of Federalism, 50(3): 518–39. doi: 10.1093/publius/pjaa012
Cook, T., LoBianco, T. and Eason, B. (2015) Governor mike pence signs RFRA fix, Indianapolis Star, https://www.indystar.com/story/news/politics/2015/04/01/indiana-rfra-deal-sets-limited-protections-for-lgbt/70766920/.
Council of State Governments (2017) Book of the States, vol 49. Lexington, KY: Council of State Governments. https://issuu.com/csg.publications/docs/bos2017
DeMora, S.L., Collingwood, L. and Ninci, A. (2019) The role of super interest groups in public policy diffusion, Policy & Politics, 47(4): 513–41.
Desmarais, B.A., Harden, J.J. and Boehmke, F.J. (2015) Persistent policy pathways: inferring diffusion networks in the American states, American Political Science Review, 109(2): 392–406. doi: 10.1017/S0003055415000040
England, T. (2019) A civics education for USA today, Oklahoma Council of Public Affairs, https://www.ocpathink.org/trent-england-show/a-civics-education-for-usa-today?print.
Garrett, K.N. and Jansa, J.M. (2015) Interest group influence in policy diffusion networks, State Politics & Policy Quarterly, 15(3): 387–417. doi: 10.1177/1532440015592776
Gilardi, F. and Wasserfallen, F. (2019) The politics of policy diffusion, European Journal of Political Research, 58(4): 1245–56. doi: 10.1111/1475-6765.12326
Glick, H.R. and Hays, S.P. (1991) Innovation and reinvention in state policymaking: theory and the evolution of living will laws, Journal of Politics, 53(3): 835–50. doi: 10.2307/2131581
Hansen, E.R. and Jansa, J.M. (2021) Complexity, resources, and text borrowing in state legislatures, Journal of Public Policy, 41(4): 752–75. doi: 10.1017/S0143814X20000252
Hays, S.P. (1996) Patterns of reinvention: the nature of evolution during policy diffusion, Policy Studies Journal, 24(4): 551–66. doi: 10.1111/j.1541-0072.1996.tb01646.x
Hertel-Fernandez, A. (2014) Who passes business’s ‘Model Bills’? Policy capacity and corporate influence in U.S. state politics, Perspectives on Politics, 12(3): 582–602. doi: 10.1017/S1537592714001601
Hertel-Fernandez, A. (2019) State Capture: How Conservative Activists, Big Businesses, and Wealthy Donors Reshaped the American States – and the Nation, Oxford: Oxford University Press.
Hertel-Fernandez, A., Skocpol, T. and Lynch, D. (2016) Business associations, conservative networks, and the ongoing republican war over medicaid expansion, Journal of Health Politics, Policy, and Law, 41(2): 239–86. doi: 10.1215/03616878-3476141
Jansa, J.M., Hansen, E.R. and Gray, V.H. (2019) Copy and paste lawmaking: legislative professionalism and policy reinvention in the states, American Politics Research, 47(4): 739–67. doi: 10.1177/1532673X18776628
Johannes, J.R. (2015) Thinking about Political Reform: How to Fix, or Not Fix, American Government and Politics, Oxford: Oxford University Press.
Karch, A. and Cravens, M. (2014) Rapid diffusion and policy reform: the adoption and modification of three strikes laws, State Politics and Policy Quarterly, 14(4): 461–91. doi: 10.1177/1532440014561867
Ladner, M. (2019) Copy, paste, misstate: the republic’s misunderstanding on model legislation, Chamber Business News, https://chamberbusinessnews.com/2019/05/01/column-copy-paste-misstate-the-arizona-republics-misunderstanding-on-model-legislation/.
Lasswell, H.D. (1971) A Preview of Policy Sciences, New York: Elsevier.
Lewallen, J. (2016) Legislative error and the ‘politics of haste’, PS: Political Science and Politics, 49(2): 239–43. doi: 10.1017/S1049096516000135
Lieb, D.A. (2014) Gov. Jay Nixon vetoes background checks for health navigators, Kansas City Star, https://www.kansascity.com/news/state/missouri/article688941.html.
Linder, F., Desmarais, B.A., Burgess, M. and Giraudy, E. (2020) Text as policy: measuring policy similarity through bill text reuse, Policy Studies Journal, 48(2): 546–74. doi: 10.1111/psj.12257
McConnell, A. (2010) Policy success, policy failure and gray areas In-between, Journal of Public Policy, 30(3): 345–62. doi: 10.1017/S0143814X10000152
McVan, M. (2019) Writing the law: from justice reform to the Bible, model legislation travels from state to state, Missourian, https://www.columbiamissourian.com/news/state_news/writing-the-law-from-justice-reform-to-the-bible-model-legislation-travels-from-state-to/article_0f19a888-ee09-11e9-b1b3-03336eead1d2.html.
Mooney, C.Z. (2020) The Study of US State Policy Diffusion: What Hath Walker Wrought?, Cambridge: Cambridge University Press.
O’Dell, R., and Penzenstadler, N. (2019) Copy, paste, legislate: you elected them to write new laws. they’re letting corporations do it instead, USA Today, https://www.usatoday.com/in-depth/news/investigations/2019/04/03/abortion-gun-laws-stand-your-ground-model-bills-conservatives-liberal-corporate-influence-lobbyists/3162173002/.
Preuhs, R.R. (2007) Descriptive representation as a mechanism to mitigate policy backlash: Latino incorporation and welfare policy in the American states, Political Research Quarterly, 60(2): 277–92. doi: 10.1177/1065912907301981
Sergent, J., Wynn, M. and Madhani, A. (2019) They copied bills from corporations. These lawmakers say that’s OK, USA Today, https://www.usatoday.com/in-depth/ news/investigations/2019/07/17/lawmakers-ok-with-copying-model-bills/3424549002/.
Squire, P. (2017) A squire index update, State Politics and Policy Quarterly, 17(4): 361–71. doi: 10.1177/1532440017713314
Volden, C. (2006) States as policy laboratories: emulating success in the children’s health insurance program, American Journal of Political Science, 50(2): 294–312. doi: 10.1111/j.1540-5907.2006.00185.x
Walker, J.L. (1969) The diffusion of innovations among the American states, American Political Science Review, 63(3): 880–99. doi: 10.1017/S0003055400258644
Wilkerson, J., Smith, D. and Stramp, N. (2015) Tracing the flow of policy ideas in legislatures: a text reuse approach, American Journal of Political Science, 59(4): 943–56. doi: 10.1111/ajps.12175
