patterns of reinvention : the nature of evolution during policy diffusion

17
PoUcy Studies Journal, Vol. 24. No. 4. 1996 (551-566) Patterns of Reinvention: The Nature of Evolution During Policy Diffusion Scon P. Hays This study considers how states change policy innovations as they diffuse. This process of policy evolution during diffusion, known as reinvention, has been the focus of recent work on innovation diffusion in the states (Clark. 1985; Glick. 1992; Glick & Hays. 1991; Mooney & Lee, 1995). This manuscript focuses on differences in the patterns of reinvention for three different policy innovations with different rates of diffusion: child abuse reporting laws, crime victim compensation laws, and public campaign funding laws. Moreover, how the pattern changes i^er states amend their initial laws is examined. The data show evidence of reinvention during diffusion but provide only limited support for the first hypothesis that the pattern of reinvention is one of increasing comprehensiveness over time. Consistent with expectations, the data reveal little relationship between the date of adoption and the current law's comprehensiveness, including amendment. Reinvention, ihe process of evolution of an innovation during diffusion refers to the changes adopters make to an innovation and is based on tht assumption that innovations evolve as they diffuse (Rice & Rogers, 1980 Rogers, 1978; Rogers, 1995). Recently, this approach to innovation diffusion has been integrated into research on the diffusion of state policy innovations. Mooney and Lee (1995) looked at reinvention during the diffusion of pre-Roe abortion reform, and Glick and Hays (1991) examined the reinvention of living will laws. Clark (1985), while not referring specifically to reinvention, examined the evolution of several policy innovations as they diffused. This existing research on policy change over time generally supports the notion that later adopters adopt laws of greater scope (Glick, 1992; Glick & Hays, 1991; Mooney & Lee, 1995). Earlier research, however, suggested a pattem of policy evolution whereby early and later adopters adopt taws of the greatest scope, while midrange adopters adopt laws of greater variation in scope (Clark, 1985). Therefore, some controversy exists regarding the emergent patterns of reinvention. Is there a general tendency to increase the scope of a law during diffusion, or do different policies not only diffuse differently but exhibit different patterns of reinvention? By taking a comparative approach and examining policies likely to be reinvented among states, this study joins this debate. After considering the extent of policy reinvention, this study examines the pattems of reinvention for each of the policies to consider several research questions regarding reinvention. Does the scope of the laws increase during diffusion, or do later adopters adopt weaker laws of more restricted scope? Or, possibly, are reinvention pattems random, depending on the circumstances and needs of adopters? Finally, is there a second round of reinvention when prior adopters reinvent their laws by amendment? To analyze these questions, this study compares the patterns of reinvention over time during the diffusion of child abuse reporting laws, crime victim compensation laws, and public campaign funding laws. The extent of reinvention is determined by coding the substantive effect of individual provisions 551

Upload: scott-p-hays

Post on 26-Sep-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Patterns of Reinvention : The Nature of Evolution During Policy Diffusion

PoUcy Studies Journal, Vol. 24. No. 4. 1996 (551-566)

Patterns of Reinvention: The Nature ofEvolution During Policy Diffusion

Scon P. Hays

This study considers how states change policy innovations asthey diffuse. This process of policy evolution during diffusion, knownas reinvention, has been the focus of recent work on innovationdiffusion in the states (Clark. 1985; Glick. 1992; Glick & Hays. 1991;Mooney & Lee, 1995). This manuscript focuses on differences in thepatterns of reinvention for three different policy innovations withdifferent rates of diffusion: child abuse reporting laws, crime victimcompensation laws, and public campaign funding laws. Moreover, howthe pattern changes i^er states amend their initial laws is examined.The data show evidence of reinvention during diffusion but provide onlylimited support for the first hypothesis that the pattern of reinvention isone of increasing comprehensiveness over time. Consistent withexpectations, the data reveal little relationship between the date ofadoption and the current law's comprehensiveness, includingamendment.

Reinvention, ihe process of evolution of an innovation during diffusionrefers to the changes adopters make to an innovation and is based on thtassumption that innovations evolve as they diffuse (Rice & Rogers, 1980Rogers, 1978; Rogers, 1995). Recently, this approach to innovation diffusion hasbeen integrated into research on the diffusion of state policy innovations. Mooneyand Lee (1995) looked at reinvention during the diffusion of pre-Roe abortionreform, and Glick and Hays (1991) examined the reinvention of living will laws.Clark (1985), while not referring specifically to reinvention, examined theevolution of several policy innovations as they diffused.

This existing research on policy change over time generally supports thenotion that later adopters adopt laws of greater scope (Glick, 1992; Glick & Hays,1991; Mooney & Lee, 1995). Earlier research, however, suggested a pattem ofpolicy evolution whereby early and later adopters adopt taws of the greatest scope,while midrange adopters adopt laws of greater variation in scope (Clark, 1985).Therefore, some controversy exists regarding the emergent patterns of reinvention.Is there a general tendency to increase the scope of a law during diffusion, or dodifferent policies not only diffuse differently but exhibit different patterns ofreinvention? By taking a comparative approach and examining policies likely tobe reinvented among states, this study joins this debate. After considering theextent of policy reinvention, this study examines the pattems of reinvention foreach of the policies to consider several research questions regarding reinvention.Does the scope of the laws increase during diffusion, or do later adopters adoptweaker laws of more restricted scope? Or, possibly, are reinvention pattemsrandom, depending on the circumstances and needs of adopters? Finally, is there asecond round of reinvention when prior adopters reinvent their laws byamendment?

To analyze these questions, this study compares the patterns ofreinvention over time during the diffusion of child abuse reporting laws, crimevictim compensation laws, and public campaign funding laws. The extent ofreinvention is determined by coding the substantive effect of individual provisions

551

Page 2: Patterns of Reinvention : The Nature of Evolution During Policy Diffusion

PoUcy Studies Journal. 24:4

comprising each of these laws.' This study is unique from previous research, firstby taking an explicitly comparative approach to the study of reinvention, second,by examining the resultant changes to these laws by amendment since their initialadoption, and third, by detailing the changes in particular provisions in threepreviously unexamined policy areas.

Innovation Diffusion and Reinvention

Most of the eariy research conceming the diffusion of policy innovationsamong the states assumed that policies remained unchanged during diffusion. Thatis, policy adoption was coded dichotomously: Either the innovation was adoptedor it was not. While Walker (1959, p. 565) realized that this assumption perhapswas unrealistic, it was necessary for the coding of the large aggregation of policiesincluded in his seminal work. Furthermore, this dichotomous coding of adoptionmethod was used by subsequent studies of state innovativcness (Gray, 1973;Savage, 1978). Indeed, these early studies were concemed primarily with assessingstate innovativeness based on whether states were among the first iLo adopt severaldiverse innovations, and not with the diffusion of particular innovations, as isreinvention research. More recently, the innovation research of Berry and Berry(1990, 1992) focused on the diffusion of particular innovations to developexplanatory models of the timing of adoption of state innovations. While thisresearch was revealed the importance of motivation, obstacles, and resources, aswell as regional influences on policy adoption, it also relied on dichotomouscoding of the dependent variable.

Importantly, in many, perhaps most, instances, this assumption of anunchanging innovation may hold true since many innovations by their nature donot lend themselves to reinvention. For example, with state lotteries or taxincreases, policy adoption is basically a dichotomous choice, although there maybe variation in the nature of the specific games offered (Berry & Berry 1990,1992). Innovations such as the institution of particular taxe:>, professionallicensing, or setting up certain agencies clearly are less susceptible to policyreinvention. 2

However, for those polices which lend themselves to reinvention,examining this reinvention adds a crucial dimension to the study of stateinnovativeness. To the extent that later adopters do adapt and improve upon anearlier adopted innovation, these states could be considered innovative in their ownright, despite their later adoption of the law. This type of innovative behavior isleft unexamined by the large aggregated studies of state innovation diffusion.

Reinvention suggests that policies change systematically over theirdiffusion period. Previous research conceming reinvention established thisrelationship between the content of a policy and the date of its adoption (Clark,1985; Glick & Hays, 1991; Mooney & Lee, 1995; Rogers, 1995). Whileevidence of change in comprehensiveness over time indicates tlie presence ofreinvention, the expected nature of that pattern is somewhat less clear, andtheoretical guidance in this area is somewhat limited. Rogers (Rice & Rogers,1980; Rogers, 1978, 1995) suggested that other pattems, even nonlinear pattems,are possible since reinvention may result from both the needs and circumstances ofsubsequent adopters as well as a social learning process (Hays, 1996; Rogers1995).

552

Page 3: Patterns of Reinvention : The Nature of Evolution During Policy Diffusion

Hays: Patterns of Reinvention

Reinvention and the subsequent changes in comprehensiveness thereforeraise an intriguing question regarding the diffusion of innovations among thestates. While certain states by virtue of their willingness to venture into new,untried policy areas may be viewed as pathbreakers, the policies they adopt maylook very different from subsequent adoptions. Early adopting states could adopttentative, watered-down, or outright ineffective versions of the policy due to thepolitical compromises necessary to secure its adoption, particularly given the lackof prior experience in the policy area. Consequently, subsequent adopters are inthe relatively more advantageous position of being able to observe patiently theexperiences of the earliest states and to leam from their experiences. Ironically,adopters at the end of the diffusion period, while clearly not innovative in thesense of being early adopters, have the opportunity to adopt the most innovativeversion of a policy by learning from the collective experiences of all earlieradopters.

The findings of state policy research, while generally supporting thenotion of increasing comprehensiveness over time, simultaneously offer evidenceof the possibility of more complex relationships. For example, while Mooneyand Lee (1995) found a generally increasing pattern of abortion law"permissiveness," they speculated that adoption of more restrictive laws by lateradopting states may have been precluded by the intervention of the Roe v. WadeSupreme Court case. Glick and Hays (1991) and Glick (1992) provided generalsupport for a linear trend of increasing "facilitativeness" of living will laws. Butthey also identified certain provisions that decrease in facility as living will lawsdiffuse, creating a "contrary substream" of policy reinvention (Glick & Hays,1991, p. 845). Moreover, Clark (Allen & Clark, 1981; Clark, 1985; Clark &French, 1984) found that the innovative early adopters along with the latestadopting "laggards" adopt laws of the greatest scope, while middle adopters showgreat variation in the scope of their laws. While Hoefler (1994) found aconvergence in living will court cases, he found wide diversity among early livingwill law adoptions and convergence among later adopters. Thus, increasingcomprehensiveness over time is not an assumption of this research, but ahypothesis to be tested with empirical data. This hypothesis is as follows:

/ / / ; The comprehensiveness of legislation increases over time.

This hypothesis will be tested by regressing the comprehensiveness scoreof each piece of legislation on the year of its adoption. A positive slope indicatesincreasing comprehensiveness over time, and a negative slope indicates decreasingcomprehensiveness over time. A very low or zero slope indicates that there is nosupport for any linear trend in the law's comprehensiveness. The absence of anylinear trend would suggest that, while reinvention may still be present, it probablyresults not from a temporal learning process but from factors particular to eachadopting ^

Reinvention and Amendment

Research also suggests that amending earlier laws provides opportunitiesfor early adopters to revisit their legislation. In the only studies to examine thisprocess systematically, Glick and Hays (1991) and Glick (1992) studied theprovisions of living will laws after amendment and found little systematicdifference in the facility of living will laws. Therefore, one would expect a

553

Page 4: Patterns of Reinvention : The Nature of Evolution During Policy Diffusion

Policy Studies Journal. 24:4

convergence in policy content as states use communication networks to revisit andrevise their laws. Combined with an increase in comprehensiveness, thisconvergence should be at a higher point since the amendment process enables theearliest adopters to increase the scope of their laws as they leam from their ownexperiences or from those of subsequently adopting states. Hoefler's (1994)research on living will laws has shown diversity among early adopters butconvergence among later adopters of living will laws. Perhaps tliie diversity thatexists among the laws after amendment is a function of the needs andcircumstances of the particular state and is unrelated to when the law was adopted.This lack of a relationship between comprehensiveness and the date of adoption isthe second hypothesis to be tested by this research:

H2: After allowing a period of time for amendments,comprehensiveness is unrelated to the date cfthe law's adoption.

Data and Method

To select the policies included in the study, I used a comparativeapproach, selecting three separate policies relying on several criteria. Thesecriteria allow me to test for pattems of reinvention under different conditions.Obviously, the policies cannot be chosen randomly, since many innovations bytheir nature do not allow for substantial variation among provisions. Thus, theselected policies must consist of a variety of provisions that may or may not beadopted during the diffusion process. To test for different pauems of reinventionunder different conditions, the policies chosen reflect different diffusion rates.Thus, this study examines one very rapidly diffusing policy, one with a mediumrate of diffusion (reflecting the average diffusion rate of about 25 years), and onewith a very protracted diffusion period, only adopted by half the states despite thepassage of more than 25 years since its initial adoption. The policies finallyselected were child abuse reporting laws, which require certain individuals to reportevidence of child abuse; crime victim compensation laws, which allow victims ofcrime to claim compensation against the state for expenses incurred resulting fromtheir victimization; and public campaign funding laws, which provide for statefunding to the campaigns of political candidates or to political parties to distributeto candidates.

The data consist of the provisions of each of these laws as initiallyadopLcd by states. Similar to other studies of legislative content, the scoringtechnique involves a content analysis of the provisions of each law (Fairbanks1980; Glick, 1992; Glick & Hays, 1991; Rice & Rogers, 1980). Eiach provisionin each law was coded based on its substantive effect on the overallcomprehensiveness of the law. That is, provisions were coded 1 if they increasedthe law's comprehensiveness or 0 if they narrowed comprehensiveness. Forexample, victim compensation laws including a provision paying for psychiatricexpenses were coded 1 for that provision since it expands comprehensiveness,while those including a provision that requires victims to demonstrate financialneed in order to qualify for compensation were coded 0 for that provision since itnarrows comprehensiveness by reducing the number of qualified applicants.Summing these codes creates a score measuring the legislation's aggregatecomprehensiveness. The initially enacted legislation in every state adopting eachlaw has been coded including all slate adoptions through 1991. To test H2,1 alsocoded the comprehensiveness of each state's current law reflecting all amendments

554

Page 5: Patterns of Reinvention : The Nature of Evolution During Policy Diffusion

Hays: Patterns of Reinvention

added to all state laws through 1991. If no amendments were added to a state'slaw, obviously the comprehensiveness score is not changed.

Child Abuse Reporting Laws

Of the three policies analyzed, child abuse reporting laws diffused mostrapidly. Initially adopted by 9 slates in 1%3, all 50 states had adopted such a lawby 1967. I based the comprehensiveness measure of child abuse reporting laws onthe laws' degreeof protection of the interests of abused children. Table 1 i^esentseach state's comprehensiveness score by the date the law was adopted. Despite theextremely rapid diffusion period, adopting states clearly reinvented these laws asthey diffused. The comprehensiveness scores range from Califomia's 2 in its 1963law to a 10 for Nevada's 1965 enactment. The mean score for all states was 4.78,with a standard deviation of 1.78.

Table 1Comprehensiveness Score by Date of Adoption: Child AbuseReporting Laws, 1963-1967

YearState

1963CAMNORTNCOFLOHIDWY

1964MDMANYAZNJRILAMIKYSD

ScoreL'

233344456

2224445568

A"

131312131315111012

9131211111212141111

YearState

1965TXCTMEMOGANCOKSCVTILINNHWAWIARDEKSMTNMND

ScoreL"

23334444455555666666

A**

121111147

1010149

15121199

16128

n8

12

YearState Score

L"1965 (continued)

NBALAKIAUTNV

1966VAMS

1967PAHI

78888

10

58

68

A"

121313131314

1212

1612

Notes . Initial Laws: Mean = 4.78 (sUndard deviation = 1,78), Amended Laws: Mean = 1L72(standard deviation = 2,03),' Comprehensiveness score of originally enacted law.^ Comprehensiveness score as of all amendment activity through 1991.

555

Page 6: Patterns of Reinvention : The Nature of Evolution During Policy Diffusion

PoUcy Studies Journal. 24;4

Patterns of Reinvention of Child Abuse Reporting LawsFigure 1 presents the scatterplot resulting from regressing the

comprehensiveness score on the date of adoption. The slope coefficient of .694 isboth strong and statistically significant. This means that the comprehensivenessscores of child abuse reporting laws increased by seven-tenths of a point each yearof the law's diffusion. This analysis provides support for HI, in iJiat even in thisextremely rapid diffusion period these laws became increasingly comprehensive asthey diffused.

Figure 1Comprehensiveness Score by Date of Adoption: Child AbuseReporting Laws, 1963-1967

oU

62 63 64 65 66

Date of Adoption

67 68

Notes- r =.380 (p^.(X)3): slope = .694 (sUfidard error = .243). After amendment: r =-.063.330); slope - -.133 (standard error = 301).' denotes AR. DE. IA. KS, MT. ND, NM^ denotes AK, IL, IN, NH, WV. WI" denotes GA. NC, OK. SC. VT, WA

Reinvented ProvisionsThe earliest child abuse reporting laws adopted in 1963 mandated only

that physicians and other medical personnel report abuse. The comprehensivenessof the law expanded over subsequent adoptions, however. New Mexico, Utah,Nevada, and Nebraska required any person (not just physicians or other medicalpersonnel) with knowledge of abuse to report it. Illinois added a provision,subsequently adopted by Hawaii and Pennsylvania, estabUshing a statewide centralregistry of abuse reports, but due to its rapid adoption and the length of timepassing since 1967, most of the story of reinvention during the diffusion of childabuse reporting laws must be told through amendment.

556

Page 7: Patterns of Reinvention : The Nature of Evolution During Policy Diffusion

Hays: Patterns of Reinvention

The Effects of AmendmentConsistent with H2, the slope of the comprehensiveness of the amended

legislation (with alt states scored for the comprehensiveness of their laws in 1991)with its adoption date drops to a nonsignificant -.133. This indicates that, afterstates are given the opportunity to amend their initial laws, there is no support fora relationship between a law's comprehensiveness and the date of its initialadoption. Moreover, not only is there a sizable increase in meancomprehensiveness, to 11.72, as of 1991, since the initial round of adoptions,every individual state increased the comprehensiveness of its law.

Since the original round of adoptions, states have added provisionsexpanding the list of those required to report abuse, establishing a centra! statewidechild abuse registry, guaranteeing the reporter's confidentiality, and authorizing X-rays or photographs without parental permission. Other, more recent,amendments require reporting of fetal alcohol syndrome as well as mental oremotional abuse of children.

Overall, the pattem of reinvention of child abuse reporting laws showsthat states expanded Uie comprehensiveness of these laws even as they diffusedrapidly. Consisu^nt with HI, states increased the comprehensiveness of this policynot only during its initial diffusion, but they continued to increasecomprehensiveness by amendment. Consistent with H2, after states have theopportunity to amend, any relationship between their initial date of adoption andthe comprehensiveness of their law disappears.

Crime Victim Compensation

Califomia fu st adopted a crime victim compensation law in 1965, and 43states had enacted such laws by 1988. Thus this policy is roughly consistent withWalker's (1969) mean of 25 years for policy diffusion. I based thecomprehensiveness measure of victim compensation laws on the extent to whichlaws compensate all victims of any criminal act quickly, thoroughly, and easily.Table 2 presents each state's comprehensiveness score by the date the law wasadopted. The scores demonstrate substantial reinvention of these laws over theirlengthy diffusion period. They range from 6 for Missouri's law adopted in 1981,to 17 for Utah's law adopted in 1986, with a mean value of 11 and a standarddeviation of 2.62.

Patterns of Reinvention of Crime Victim Compensation LawsThe scatterplot in Figure 2 shows that regressing this score on the year of

adoption results in a slope coefficient of .110, indicating on average a 1.1 pointincrease in comprehensiveness over a period of 10 years. While this certainly ismore attenuated than is the slope for child abuse repCHting laws, it is still positiveand statistically significant at the .10 level. The table and the scatterplot alsoshow wide variation in state comprehensiveness scores. Although considerablyweaker, the direction of the coefficient lends further support to HI regardingincreasing comprehensiveness of policy innovations over time.

557

Page 8: Patterns of Reinvention : The Nature of Evolution During Policy Diffusion

PoUcy Studies Journal, 24:4

Table 2Comprehensiveness by Date of Adoption: Crime VicilimCompensation Laws, 1965-1990

YearState

1965CA

1966NY

1967MAHI

1968MD

1969NV

1971NJ

1972AKLARI

1973ILWA

1974DEMN

ScoreL»

9

8

712

7

9

10

101515

1012

1313

A"

15

17

1015

10

13

12

141416

1513

1515

YearState

1975

1976

1977

1978

1979

WIND

KYMIVATNOHPA

INFL

CTKSNE

TX

ScoreL'

913

888

111414

7U

101215

13

A**

1114

139

12101515

1310

111216

16

YearState

1981

1982

1983

1984

1985

1986

1987

1988

MONMCOOKWV

IASC

NC

AL

WY

UT

AR

GA

S(L'

68

101313

813

13

12

10

17

13

12

:or(A

99

141818

1015

13

12

11

18

13

12

Initial Laws: mean = 11.0 (sUndard deviation = 2.62). Amended laws: mean = 13.11(standard deviation = 2.41).' Comprehensiveness score (rf originally enactfd law.

ComfM^ensiveness score as of all tmendment activity through 1991.

558

Page 9: Patterns of Reinvention : The Nature of Evolution During Policy Diffusion

Hays: Patterns of Reinvention

Figure 2Comprehensiveness by Date of Adoption:Compensation Laws, 1965-1990

Crime Victim

60 65 70 75 80

Date of Adoption

85

Notes , r = .l-il (p = .062); slope = .110 (standard error = .070).After amcndmeni: r= -.110 (;> - .240); slope = -.047 (standard error = .066).

Reinvented ProvisionsThe earliest victim compensation laws compensated only medical

expenses arising from physical harm, and a few states compensated for lostearnings. Hawaii's 1967 law was the first to include payment for mental anguish,but this provision was not copied by any other state. Louisiana and RhodeIsland's 1972 law first expanded victim compensation to include payments forpsychiatric treatment and funeral expenses. Subsequently, 22 other states includedthe psychiatric treatment provision and 35 included payment for funeral expenses.Louisiana's 1972 law also was first to add a provision requiring police to informcrime victims of their eligibility for the program. State officials soon realized thatthe anticipated flood of victims claiming compensation did not materialize(Carrow, 1980; McGillis & Smith. 1983). and 33 more states followedLouisiana's lead.

A good example of reinvention in practice is provided by a provisionprohibiting relatives from receiving compensation. While adding to the law, sucha provision decreased comprehensiveness since it disqualifies most victims ofdomestic abuse from receiving compensation (in calculating comprehensiveness, itwas coded 0 if present in a state's law and 1 if absent). The iwovision prohibitingrelatives from receiving compensation was included first in New York's 1966 lawand subsequently was included in the legislation of 35 other states. However, overthe years, this provision often was challenged in court as discriminating unfairlyagainst a specific class of victims—those of domestic abuse. Nine states thatoriginally had included it, beginning with Alaska in 1983, subsequently repealedthis restrictive provision.

559

Page 10: Patterns of Reinvention : The Nature of Evolution During Policy Diffusion

PoUcy Studies Journal. 24:4

In contrast, two other somewhat controversial provisions were added tothese laws successfully, thereby reducing their comprehensiveness. One. alsoadded first by New York in 1966, required a person claiming compensation todemonstrate financial need for compensation. This restrictive provision, copiedsubsequently by only 14 other slates, subsequently was repealed by 4 of those(including New York). The other provision, adopted first by MiU7land in 1968,required victims to cooperate with police as a condition of receiving acompensation award. It was adopted subsequently by 36 other states, to placatepolice officers' demands and to promote law enforcement (Meiners, 1978).

The Effects of AmendmentAfter amendment, the mean comprehensiveness of victim compensation

laws rises slightly to 13.11. Moreover, of 34 states amending their laws, all but2 increased the comprehensiveness of their legislation after adding amendments.The overall effect of amendment has been to reduce further the relationship of thecomprehensiveness score to the date of adoption to a nonsignificant value of -.047(p = .240). This lends support to H2, which hypothesizes such lack of arelationship.

Overall, the pattern of reinvention for crime victim compensation lawsreveals increasing comprehensiveness over time, although at a much slower pacethan child abuse reporting laws. Moreover, later adopting states have reduced thescope of victim compensation laws by adding two provisions that placerestrictions on receiving compensation. Thus, while most of the changes increasethe comprehensiveness of the laws, these two provisions reveal a "contrarysubstream" of reinvention to reduce comprehensiveness similar to that found byGlick and Hays (1991) for living will laws. Unlike child abuse reporting lawsthen, the diffusion of crime victim compensation laws indicates that reinvention,or the addition of provisions to a law, also may result in a reduction incomprehensiveness, at least for particular provisions. The pattern of reinventionin the amendment process also is as predicted, with any relationship betweencomprehensiveness and the original adopting date disappearing after states have hadthe opportunity to amend.

Public Campaign Funding Laws

Public campaign funding laws provide a test of policy reinvention undervery different conditions. These laws were first adopted by 4 states in 1973, and19 had followed suit by 1991. Unlike child abuse reporting laws or victimcompensation laws, public campaign funding laws have been adopted by fewerstates over a more protracted diffusion period. I based the comprehensivenessmeasure of public campaign funding laws on the capacity of ihe law to raise anddispense funds for the greatest variety of electoral contests (Jones, 1980; Moore &Labelle, 1985; Noragon, 1981). Table 3 shows the comprehensiveness scores ofeach state by the date it adopted the policy, and again illustrates reinvention duringthe diffusion of these laws. The scores range from 2 for New Jersey's 1974 law to9 for the law adopted in that same year in Minnesota. The mean state score was5.44, with a standard deviation of 2.33.

560

Page 11: Patterns of Reinvention : The Nature of Evolution During Policy Diffusion

Hays: Patterns of Reinvention

Table 3Comprehensiveness by Date of Adoption: Public CampaignFunding Laws, 1973-1990

Year Year

1973

1974

1975

1976

State

MERl

urIANJMTMDMN

MAIDNC

MIKY

ScoreL' A"

State Score

1977

1979

1982

1983

1986

1987

INORWI

HIOK

VA

ALCA

FL

OH

357

77

4

45

4

3

347

7(rqxialed)

4

45

4

3

Notes . Initial laws: mean = 5.44 (sundard deviation = 2.33). Amended laws: mean = 5.30 (standarddeviation - 2.27)." Comprehensiveness score of originally enacted law."Comprehensiveness score as of all amendment activity through 1991.

Patterns of Reinvention of Public Campaign Funding LawsThe scatterplot of comprehensiveness scores over time in Figure 3

suggests two things. First, later adopters generally have adopted lesscomprehensive laws than have some early adopting states, but second, amongearly adopters the variation in the laws was much greater than among lateradopters. Regressing comprehensiveness over time results in a weak, negativeslope of -.150 (p = .093). This indicates a 1.5-poini average decline in score over a10-year period. This finding therefore contradicts HI, which predicted increasingcomprehensiveness over time. Moreover, no state adopting since the late 1970shas scored higher than 5 in comprehensiveness.

Thus, the comprehensiveness of public campaign funding laws declinesover time, although the scatterplot indicates that the negative relationship likelyresults from later adopters not adopting the type of comprehensive laws that wereadopted early in the diffusion period by states such as that adopted by RhodeIsland, Oklahoma, Minnesota, and North Carolina. It also is telling that the mostcomprehensive laws of Rhode Island and Oklahoma subsequently were repealed.Perhaps later slates realized the ambitiousness of such grandiose schemes andpurposefully adopted scaled-down plans. Interestingly, this interpretation isevidence of the social learning process characterizing policy reinvention.

561

Page 12: Patterns of Reinvention : The Nature of Evolution During Policy Diffusion

Policy Studies Journal. 24:4

Figure 3Comprehensiveness by Date of Adoption: Public CampaignFunding Laws, 1973-1990

i :

c

V.C

fl 1

cCLEoU

10

fio

6

4

2

0

RJ*

•NC

IA * Wi* *

MDUT \ r r Si:\-K¥- 9R —ME ID

MI [N

NJ

0.K

HI

CA

OH'

72 74 76 78 80 82 84

Date of Adoption

86

'' = .286(p = .093):slope= .150(standarderrof = .109). After amendment: r =-.270 (p.107); slope = -.143 (standard eiror = .111).

Reinvented ProvisionsAU of the laws adopted to date provide financing for gubernatorial

elections. Most of ihe laws fund primary elections and provide for revenuegeneration through a tax checkoff similar to the federal presidential campaignfunding law. Some slates also imposed spending limits and provided that statefunds would be used as a match to existing campaign funds. Scrutiny of particularprovisions reveals no clear patterns, and few provisions stand out as being addedonly by the laws of later adopting states. All of the 1973 adopters permitted thosemaking contribution.s to the fund through income tax checkoffs to specify thepolitical party they wished to receive the funds. But Maryland, Montana, and NewJersey did not include such a provision in their 1974 laws, nor did eightsubsequently adopting states. Minnesota was First to provide financing through ageneral appropriation in addition lo a taxpayer checkoff.

While the scope of these laws has declined overall, one example ofincreasing scope is that while some early adopters did not provide financing forelections other than the general election, all laws adopted since 1977 have providedprimary election financing. None of the laws adopted since 1979 have fundedlegislative elections, and all have conditioned disbursement of funds on matchingfunds from the candidate rather than providing absolute amounts based on availablefunds.

The Effects of AmendmentThe changes resulting from the few amendments to these laws do not

result in much change to the earlier relationship. On average, the meancomprehensiveness score declined to 5.3. Iowa was the only state to haveincreased its comprehensiveness score by amending its law. Maryland reduced thecomprehensiveness of ils legislation from 6 to 2, and Oklahoma repealed its

562

Page 13: Patterns of Reinvention : The Nature of Evolution During Policy Diffusion

Hays: Patterns of Reinvention

campaign funding law. The negative relationship between the original adoptiondate and current comprehensiveness score decreases slightly to a nownonsignificant -.143. Yet this nonsignificant relationship is consistent with H2,which predicts no relationship between time and comprehensiveness afteramendment

In contrast to HI. the diffusion of public campaign funding taws providesevidence that policy reinvention can result in a pattern of decliningcomprehensiveness over time. Possibly, the evolution of these laws suggests asomewhat different diffusion model whereby reinvention of controversiallegislation is a function of political or ideological factors specific to a state. Thereduction of policy comprehensiveness likely results from political compromisesmade necessary by the controversial nature of this policy. Moreover, amendmentdoes not alter this relationship fundamentally as it did for the other policies,although consistent with the findings for the other laws, there is no relationshipbetween time and comfwehensiveness after amendment of these laws.

Discussion

To revisit the questions asked early in this paper, first, each of thesepolicies makes clear that policies evolve as they diffuse. While states may followthe leaders in the initial adoption of innovative policies, they do not follow theleaders in reinvention; state policymakers feel free to reinvent these policies bybuilding on or adapting policies previously adopted by more innovative states.Further, the results suggest that the first hypothesis concerning increasingcomprehensiveness during reinvention may have been somewhat naive. In fact,these results suggest that the pattems of reinvention may be related to the rate oftheir diffusion. Child abuse reporting laws, which were the most rapid to diffuse,exhibited the clearest support for the hypothesis of increasing comprehensiveness.While the direction of reinvention for the more slowly diffusing victimcompensation laws also supported HI, this effect was quite weak. Furthermore,the pattern of reinvention of public campaign funding laws, which have been theslowest of these three policies to diffuse, illustrates that reinvention also mayresult in less comprehensive laws over time. Possibly, this fmding suggests thatthe various patterns of reinvention may result from the level of controversyassociated with the policy. Savage (1985) suggests that the rate of diffusion isrelated to the controversial nature of a pwlicy, with noncontroversial policies beingthe most rapid to diffuse. Thus, opposition to controversial laws may increaseduring diffusion, necessitating watering down the law to secure passage, whilesubsequently adopting states continue to expand the scope of noncontroversiallaws, as opposition is minimal. The less controversial the law, the stronger therelationship should be between time and increasing comprehensiveness.

The findings also show evidence for a second round of reinventionthrough amendment whereby earlier adopters reinvent their laws. Moreover, thetrend in amendment activity often follows the general trend in reinvention. Forthe laws exhibiting increasing comprehensiveness, amendment increasedcomprehensiveness further, for public campaign funding that exhibits decreasingcomprehensiveness, amendment decreased comprehensiveness. Moreover, aconvergence in policy seems to occur, since in every case the relationships thatexisted between time and comprehensiveness during the law's initial diffusiondisappear after amendment.

563

Page 14: Patterns of Reinvention : The Nature of Evolution During Policy Diffusion

Policy Studies Journal. 24:4

Possibly, different patterns of reinvention result from characteristicsunique to the adopting state. These would include such conditions as the need forthe policy, public support for the policy, along with the political or ideologicalcleavages that may divide state political institutions. For example, a state withrelatively few instances of child abuse, a very low crime rate, or a healthy,competitive political system would have little need for adopting comprehensivelaws in any of these areas, if they would adopt the law at all. V/here public orinterest group opposition was intense, compromise may necessitate reducing anyof these laws' comprehensiveness to secure passage in any state consideringadoption. While ideological cleavages are relatively less important for broad,consensual policies such as child abuse reporting laws or even victimcompensation laws, it is highly unlikely that a conservative, Republican-dominated state would adopt a comprehensive public campaign funding law.

Implications

Understanding the pattems of reinvention is important for the study ofpolicy diffusion and state policy generally for several reasons. First, reinventionsuggests rethinking previous notions about innovativeness. To thie extent that apolicy permits a degree of flexibility among subsequent adoptions, reinventionillustrates how states adapt and change an innovation as it diffuses. While achronological approach to diffusion implies that only the early-adopting states areinnovative, reinvention illustrates how later adopters may be innovators as well.Previous research on aggregate innovativeness discounts the development of new,innovative provisions as the laws diffuse.

Second, by showing that pattems of reinvention do not always or do notstrongly evidence increasing comprehensiveness, this research demonstrates thatreinvention may have different effects on the policy. Simply because the pattemof reinvention for public campaign funding laws was negative in no way impliesthat there was less reinvention. Altematively, this may indicate leaming amongstates that the grandiose plans of early adopters were all but impossible toimplement Thus, the scaled-back programs of later adopters perhaps illustrate amore realistic approach to public campaign funding.

This raises the additional issue of reinvention and policy implementation.Policy implementation also offers the opportunity for additional policyreinvention. As the agencies responsible for implementing these innovations do soto varying degrees, they also are engaging in reinvention. In fact ihis may be themost crucial aspect of reinvention, as it affects the success or failure of theprogram and may impact subsequent legislative reinvention through amendment,or in some instances, repeal. Future research should examine Uie relationshipbetween implementation and subsequent reinvention.

Third, reinvention provides a different way of thinking about stateswithin the federal system. While states might in fact follow the leaders foradopting the basic idea for a policy, reinvention suggests that later-adopting statesdo not blindly follow these "path-breaking" states or even the federal govemmentPolicies evolve as they diffuse, and all states may have a role to play in policyreinvention. This most likely results from interstate communication andcooperation during diffusion. In the contemporary atmosphere of stateprofessionalism, these communication networks are likely to continue inimpOTtance to the diffusion process.

564

Page 15: Patterns of Reinvention : The Nature of Evolution During Policy Diffusion

Hays: Patterns of Reinvention

Fourth, although the outcome of current policy initiatives in Washingtonis uncertain, states in the future clearly will play a greater role in developing theirown solutions to pressing domestic problems such as poverty, crime, and healthcare. Reinvention bodes well for the states' ability to respond to this challenge. Inthe true spirit of federalism, states do become laboratories of democracy, allowingtheir inherent diversity to structure individual policy responses based on theirneeds, while simultaneously looking to other states and revising, adapting, andreinventing what has gone before.

Scott P. Hays is an assistant professor at Southern IllinoisUniversity. His research interests include state politics, policy innovation, andenvironmental politics. He has published articles in the Journal of Politics^PoUtical Research Quarterly, Policy Studies Journal, Publius: The Journal ofFederalism, and Spectrum: The Journal of State Government.

Notes

For purposes of rcplicati(»i or furiher research, all empirical data arc available from theauthor as a Lotus spreadsheet Hie. This data file contains ctily the toul comprehensiveness scores foreach state. Copies of the hand-scored coding sheets for each provision for each law and the datadictionaries for coding these provisions are also available from the author.

TTianks to an anonymous reviewer for these comments.^ Importantly, relying on these measures of association is not meant to impute causality for

time alone, but merely to iUustrate the temporal trend in changes in comprehensiveness. logically,time may be a surrogate for a variety of factors that actually influences the pattems of reinvention thatthemselves may apply uniquely to different ptdicies.

References

Allen, R., & Qark, J. (1981). State policy adoptions and innovatiiMi: Lobbying and education.Slale and Local Government Review, 13. 18—25.

Berry, F. S., & Berry. W. D. (1990). State lottery adoptions as pt^icy innovations: An event historyanalysis. American Political Science Review. 84, 395-415.

Berry, F. S., & Berry W. D. (1992). Tax innovation in the states: Capitalizing on politicalopportunity. A/rurican Journal ofPolilical Science. 36, 715-742.

Carrow, D. M. (1980). Crime victim compensation. Washington, DC: National Institute of Justice.Qark, J. (1985). Policy diffusion and program scope: Research directions./'ujb/iitr, /5 , 61-70.Clark, J., & French, L (1984). Innovaticm and program content in sUte tax policies. State andLocal

Government Review, 16, 11-16.Fairbanks, J. D. (1980). Politics, economics and the public morality: Why some states are more

moral than others. In T. R. Dye & V. Gray (Eds.), Determinants of public policy (pp.125-132). Ixxington, MA: Lexmgton Books.

Glick, H. R. (1992), The right to die: Policy innovation and its consequences. New York, NY:Columbia University Press.

Glick, H. R., & Hays, S. P. (1991). Innovation and reinventicm in sUte policymaking: Theory andthe evolution of Uving will laws. Journal of Politics, 53, 835-850.

Gray, V. (1973). Innovation in the states: A diffusion study. American Political Science Review,67,1174-1185.

Hays. S. P. (1996). Influences on reinvention during the diffusion of state policy innovations.Political Research Quarterly. 49. 6\ 3-632.

Hoefler, J. M. (1994). Diffusion and diversity: Federalism and the right to die in the fifty sutes.Publius,24. 153-170.

Jones, R. S. (1980). Sute public financing and the sUte panjes. In M. J. Malbin (Ed.), Parlies,interest groups and campaign finance laws (pp. 243-267). Washington, DC: AmericanEnterprise Institute.

565

Page 16: Patterns of Reinvention : The Nature of Evolution During Policy Diffusion

PoUcy Studies Journal, 24:4

McGillis, D., &. Smith, P. (1983). Compensating victims of crime: An analysis of Americanprograms. Washington, DC: National Institute of Justice, Office of DevelopmentalTesting and Dissemination.

Meiners, R. E. (1978). Victim compensation. Lexington, MA: Lexington Books.Mooney, C. Z., & Lee, M. (1995). Legislating moraUty in the American states: The case trf pre- Roe

abortion regulation reform. American Journal of PolUical Science, 39, 599-627.Moore. T. R., & Labelle, R. D., ID. (1985). Public financing of elections: New proposals to meet

new obsucles. Florida Stale University Law Review, 13, 863-900.Noragon, J. L. (1981). Political finance and pc^tical refomi: 'Hie experience with state income tax

checkoffs. American Political Science Review, 75, 667-687.Rice, R. E., & Rogers, E. M. (1980). Reinvention in the innovation process. Knowledge, 1, 499-

514.Rogers, E. M. (1978). Re-invention during the innovation process. In M. Radnor, 1. Feller, & E. M.

Rogcn(Ed&.), The diffusion of innovations: Anassessment (pp. 167-1K2). Evanston, IL:Northwestem University, Center for the Interdisciplinary Study of Science and Technology.

Rogers, E.M. (1995). D^usion of innovations (4^ ed.). New York, NY: Free IVess.Savage, R. L. (1978). Policy innovativeness as a trait of American states. Journal of Politics, 40,

212-224.Savage. R. L. (1985). When a policy's time has come: Cases of rapid policy diffusion in 1983-

1984. Publius, 15, 111-124.Walker, J. L. (1969). llie diffusion of innovations among the American sUtes. American Political

ScUnce RevUw. 63, 880-899.

566

Page 17: Patterns of Reinvention : The Nature of Evolution During Policy Diffusion