Coping with competition: The impact of charter schooling on public school outreach in Arizona

Efforts to "reinvent government" and enhance the performance of public sector organizations have widely advanced the notion that public sector clients need to be treated more like customers (Barzelay with Armajani, 1992; Osborne & Gaebler, 1992; Osborne & Plastrik, 1997). This premise has been particularly evident in choice-based efforts to reform education, as policymakers have devoted increasing attention to the idea that markets will improve school effectiveness and responsiveness (Hassel, 1999). This impulse has marked the advent of "charter schools," relatively unregulated state-sponsored schools that compete with traditional public schools for students.

A critical question in the charter school debate has been how to promote school accountability and ensure that market incentives function as desired (Bierlein, 1997; Hill, Lawrence, & Lake, 1998; Manno, 1999; Manno, Finn, Bierlein, & Vanourek, 1998; Nathan, 1996; Schneider, Teske, Marschall, & Roch, 1998; Schneider, Teske, Roch, & Marschall, 1997; A. S. Wells & Associates, 1998). One important consideration has been the recognition that markets work most effectively when consumers have cheap, reliable, and plentiful information with which to make decisions, while the quality and availability of information on schooling has been a subject of considerable concern (Lowery, Lyons, & DeHoog, 1995; Teske, Schneider, Mintrom, & Best, 1993, 1995). School choice advocates suggest that, while existing information may be inadequate, the need to attract students will compel market-dependent schools to work more diligently to communicate information about their services and performance to the public (Chubb & Moe, 1990). Consistent with this hypothesis, survey data suggest that parents of students in charter schools are more satisfied with school communications than are parents in traditional public schools (Finn, Manno, & Vanourek, 2000).

Subjected to less scrutiny has been the assumption that competition will compel traditional public schools to work harder at communicating information on their performance and services to parents and the community. This question is particularly significant in light of ongoing disciplinary consideration as to whether and how parents and families may respond to educational choices and information on school quality (Schneider et al., 1997, 1998; Schneider, Teske, & Marschall, 2000; Smith & Meier, 1995a, 1995b; Wrinkle, Stewart, & Polinard, 1999). How schools supply information will help to explain how and why families choose the schools they do, or why they may not choose at all.

Here, we focus on this supply-side component of the information process. Parents may have greater incentives to collect information in a choice environment, but this will be unlikely to have significant effects unless the information is actually made available at a reasonable opportunity cost. Previous work suggests that private sector organizations under increased competitive pressure make greater efforts to inform customers and potential customers (Osterhoff, Locander, & Bounds, 1991; Peters, 1987, pp. 144-157). Hale (1996) and Barzelay with Armajani (1992) make the same case for previously monopolistic public sector organizations entering a competitive environment.

Do schools operating in choice environments seek to inform parents about what they are doing? School choice advocates suggest that markets will compel schools to operate more like firms, including taking steps to advertise their services (Nathan, 1996). If schools do not behave in this fashion, the ability of families to make informed judgments, and thereby hold schools accountable, will be impaired. Parents will not be without resources, as they will continue to collect information from informal networks, but the information is likely to be transmitted and used in an uneven fashion (Schneider et al., 1998). Schools possess much valuable and nuanced information that can inform familial decisions if schools choose to communicate it.

Whether public schools seek to communicate their programs and performance will help determine the effect of charter schooling on systemic accountability. In addition, the district school response to the emergence of charter competition can also tell us a great deal about how constrained and traditionally public organizations may respond to the introduction of market pressures.

We examine how the degree of local charter school competition affects school efforts to inform parents about school activities and programs. Our reference here to the implications of outreach for market-induced accountability, rather than the role of regulatory or administrative bodies (see Hassel & Vergari, 1999; Manno, 1999; Meier, Polinard, & Wrinkle, 2000), is not meant to underestimate the importance of regulatory or administrative accountability in education. Rather, it reflects the need to consider more fully how market mechanisms may or may not empower consumers.

Market advocates have long suggested that under a system of school choice parents will collect useful data and use that information to select the appropriate school for their child (Friedman, 1962). Those schools that gain students are expected to flourish, while those that lose students will close. However, this chain of events depends on parents having good information with which to make choices. While we cannot determine here whether increased district outreach necessarily translates into the provision of good or useful information, this work will help us begin to address this question.

The question investigated here is but one element of market-driven accountability. We cannot and do not seek in this article to broadly assess the value of choice-based reforms, their effect on information, or whether they produce "good" or accurate information on school quality. Rather, we explore how market pressures affect organizational "advertising" and communication, in the expectation that such efforts will help us begin to more fully understand market effects on information, consumer behavior, and the possible implications for educational quality.

The effect of charter school competition on public school communication is examined here in the context of Arizona. Arizona has the nation's most competitive local education markets, due to an expansive charter school law passed in 1994. The 1994 law permits two state-level boards to issue a total of 50 charters annually and allows a single charter holder to open multiple campuses. By January 2001, this expansive regime had resulted in the opening of 402 Arizona charter campuses. These schools enrolled 55,000 students. This figure amounted to 6.3% of K-12 enrollment, while the national figure was slightly less than 1 %.

Moreover, because state subsidies accounting for a mean of 57% of total K- 12 education expenditures follow student enrollments, Arizona districts have strong incentives to worry about losing students to charter schools. In contrast, school choice schemes in many other states include "hold harmless" provisions that protect district schools from budgetary losses owing to enrollment losses from competition (Rofes, 1998; Schneider et al., 2000).

Arguably, the unprecedented extent of the state's charter school sector could make it difficult to extrapolate from Arizona's experience. Yet studying how Arizona district schools react to charter schools is important for two reasons. First, Arizona can serve as a "canary in the coal mine." Because Arizona moved faster and further than other states, we learn from its experiences to see how the impact of charter competition may eventually play out in less ambitious charter regimes. Second, as is noted above, the existence of an educational market makes it possible to test theories regarding how public sector organizations respond to market pressures (Barzelay with Armajani, 1992; Osborne & Gaebler, 1992). Third, a number of states are considering expanding their school choice programs. Indeed, by 2000, several cities (including Washington, DC, Philadelphia, and Kansas City) were enrolling 10% or more of their students in charter schools. Arizona's ambitious charter experiment may offer lessons as to the likely longterm impact of these aggressive efforts. In short, Arizona may be an "outlier" today, but also may offer a useful preview of what is to come.

Previous Research

Organizations in established marketplaces may respond to competition with a variety of tools (Lawrence & Lorsch, 1967; Thompson, 1967). However, relatively constrained organizations such as public school systems (Bryk, Sebring, Kerbow, Rollow, & Easton, 1998; Elmore, 1997; Wilson, 1989) have a more limited range of responses. For instance, they will find it extremely difficult, to say the least, to redefine their core mission, alter their targeted clientele, reconfigure the service mix they provide, or target different geographic populations. This is due to the conflicting notions of purpose that schools must negotiate, the civil service rules under which they operate, and the pressures produced by political contestation regarding the mission, practice, and distribution of schooling (Chubb & Moe, 1990; Hess, 1999). Because it is difficult for public schools to readily change either personnel or educational practice (Elmore, Peterson, & McCarthy, 1996), efforts to reassure customers about current service provision present an alluring response to competitive pressure.

Not only are schools constrained by their nature, they also suffer under the constraints endemic to large organizations more generally. These constraints have often been found to take three forms. First, organizational leaders are unable to fully assess their situation; instead they rely on proxies and easy cues (March, 1988; March & Olsen, 1987; Simon, 1979, 1995, 1997). Second, leaders tend to rely upon routines that they already know how to do when confronted with a new or threatening situation (Allison, 1971; Peterson, 1976). Third, when organizational leaders find it difficult to provide the public with the good it demands, they often turn to symbolic responses and gestures (Hess, 1999; Meyer & Rowan, 1991; Tyack & Cuban, 1995).

Previous research on educational markets has not systematically examined whether competition affects school outreach and communication in the United States. Several studies of schooling in Great Britain (Gorard, 1997; Walford, 1994; Woods, Bagley, & Clatter, 1998) found evidence that schools step up public relations and marketing in choice-based environments. Studies of urban school system response to public voucher programs in Milwaukee and Cleveland have found some evidence that districts may work harder to market themselves or to reach out to families (Hess, 2000; McGuinn & Hess, 2000).

More generally, there are mixed signals in the literature about whether competition induces significant responses from public schools. Several researchers have found evidence that public school systems seek to improve their performance and appeal when confronted with competition (Borland & Howsen, 1992; Dee, 1998; Hoxby, 1998, 2000), while others have found otherwise (Smith & Meier 1995a, 1995b; Wrinkle, Stewart, & Polinard, 1999) or have presented more ambiguous results (Hess & Leal, in press; Schneider, Teske, Clark, & Buckley, 2000). Others have found that only a minority of districts respond in constructive ways (Armor & Peiser, 1998; Rofes, 1998; Schneider et al., 2000), and still others have found no evidence of a significant constructive response to competition from charter schooling (Hassel, 1999; Smith & Meier, 1995b; A. S. Wells & Associates, 1998).

We are interested in whether the degree of competition affects school outreach efforts. To examine this question, at the most basic level, we ask whether the emergence of a threatening local charter school presence prompted schools to step up their outreach efforts over a 3-year period. We expect that, in those districts where charter schools did not emerge, schools would be less likely to enhance their outreach efforts.

School Level Responses to Competition: The Role cf Organization Structure

Absent a highly competitive environment, the business literature suggests that organizations can often be effectively managed via centralized decisionmaking (Burns & Stalker, 1961; Courtright, Fairhurst, & Rogers, 1989; Thompson, 1967; Wright, Pringle, & Kroll, 1992). When change is relatively slow and fairly predictable, most decisions follow a routine pattern, and uniform procedures can often be established in advance. Like other large, public organizations with multiple constituencies and ambiguous missions (Wilson, 1989), school systems normally rely heavily upon centrally directed standard operating procedures.

However, organizations in rapidly changing environments-such as schools facing significant market competition for the first time-will usually benefit from decentralizing their decision-making and allowing front-line workers to make and implement context-sensitive decisions. Firms that remain centralized in such environments are at risk of making decisions slowly and without the most useful information.1 Yet schools find change difficult (Fullan, 1991), and research suggests that schools are more likely to follow preexisting practice than to radically alter their routines when pressed to change (Bryk et al., 1998; Tyack & Cuban, 1995). Researchers have found that empowerment may foster significant changes in teachers' attitudes and behaviors, by enhancing their stake in the school and their sense of control (Marks & Louis 1997; Rowan 1990). Here, we hypothesize that school communities that already operate in a decentralized manner when competition emerges are better equipped to respond to the changed environment. Using the extent of teacher empowerment at an individual school in 1994-95 (just prior to Arizona charter school competition) as our proxy for decentralized decisionmaking at the school level, we hypothesize that when school decisionmaking is decentralized, and thus is more flexible, charter competition will more likely increase school wide outreach efforts.

Methods

Using a 145-item questionnaire mailed in March 1998, we surveyed teachers regarding school behaviors during the 1994-95 and 1997-98 school years. Individual teacher responses were used to measure classroom outreach, while aggregated responses were used to assess school-level outreach. The dependent variables were the change in reported outreach activity from 1994-95, before charter competition was introduced in Arizona, to 1997-98. By asking teachers to report on events in 1994-95 and in 1997-98, we measured effects that may have occurred as competitive responses, since 1994-95 was the last school year prior to the start of charter schooling in Arizona.2 This "recall" approach poses obvious problems, but it was the only viable strategy for collecting the necessary data. While the data are subject to faulty respondent recollection, there is little reason to expect that any imprecision or misrepresentation would be correlated with the extent of local charter school competition. Absent such a correlation between these errors and the key explanatory variables, any inefficiency or misreporting due to the data-collection technique will not bias the results (King, Keohane, & Verba, 1994). Moreover, previous research has shown the recollections of public sector bureaucrats regarding organizational changes tend to be highly accurate (Maranto, 1991).

We studied elementary schools (grades K-8) in the sample districts, because competition for "typical" students is most intense in this market segment. At the high school level, the prohibitive cost of facilities prompts charter schools to focus on marginal and "at-risk" populations (Maranto & Gresham, 1999).

Of Arizona's 204 school districts with elementary schools, 45 had charter schools within their borders in 1997-98. To maximize variation on the key explanatory variable, data were collected in high-penetration and low-penetration districts. The high-penetration districts included 24 of the 25 school districts where charter schools accounted for 30% or more of public school enrollment.3 The low-penetration group included 19 of the 159 charter-free districts. The districts were selected so as to match up as closely as possible on district enrollment, poverty, and racial composition with the 24 high-penetration districts. To improve the group match, we added two large districts with very low charter penetration to the low-penetration group. This process yielded high- and low-- penetration samples that were very similar in all observable dimensions (see Table 1). Save for charter school market share, t tests found no differences significant at p < .10.

In all, 98 schools were randomly selected in the 45 sample districts.4 For each school, 18 teachers who had taught at the school for 3 years or longer were sampled. If fewer than 18 teachers met the criterion, then all were sampled. The mail survey was conducted in early March 1998; respondents were paid $5 for participating. The return rate for the district teachers was 79.1%, n = 1,065. Of the 1,065 respondents, 75 (7.0%) indicated that they were not working for their current school in 1994-95; these teachers were dropped from the analysis (since they had no direct knowledge of outreach in 1994-95). After eliminating those schools with less than 5 responding veteran teachers, we were left with 87 schools containing 959 teachers (with a mean of 11.02 teachers per school).

Dependent Variables: School Outreach School outreach data were gathered for 1994-95 and 1997-98 on whether the school: [11 made extensive efforts to inform parents about school programs and options, and [21 used flyers and other media to explain school services and programs to parents. Teachers used a 6-point scale to agree or disagree with [1] and [21. The values on the scale were 1 = strongly disagree; 2 = moderately disagree; 3 = slightly disagree; 4 = slightly agree; 5 = moderately agree; and 6 = strongly agree. Individual teacher responses were aggregated to the school level for 1994-95 and 1997-98. For all analyses, the dependent variable used was the change from 1994-95 to 1997-98. For example, if a school mean rating on an item in 1994-95 was 4.50, but changed to 5.15 in 1997-98, then the dependent variable was +.65. Positive numbers indicate a movement on the 1-6 scale toward the "strongly agree" position that outreach activities were being heavily used, and vice versa. Descriptive statistics for these variables are shown in Table 2. Note that the first question asked teachers about outreach, broadly construed.5 A broad measure was deemed most accurate because there are numerous activities schools may use to promote outreach, e.g., school nights with speakers, individual phone calls from teachers, school newsletters, more parent-teacher conferences, etc. Respondents were also surveyed more specifically regarding the use of flyers and other media, however, in order to obtain a discrete and more consistent measure for comparison.

School-level behavior is notoriously difficult to study. Teachers are the observers best placed to observe school-level changes in administrative policies (Lipsky, 1980). Accordingly, surveying teachers to study school-level changes is a common methodological approach (see, for example, Chubb & Moe, 1988, 1990). Indeed, changes in school-level implementation of administration policies are likely to precede changes in such output measures as test scores (Bryk et al., 1998).

Independent Variables

We used two variables to measure the extent of competition: [1] charter market share by district, e.g., the total charter elementary (K-8) enrollment per district divided by the total number of public elementary students (charter plus district) in the district,6 and [21 percentage of district maintenance and operations (M & 0) subsidies received from the state (which follow parental enrollment decisions-denoted as subsidy). School districts with a low subsidy level may be less concerned about a charter entry threat, and vice versa for districts with a high level. Of the 45 sample districts, 8 received less than one-third of their M & 0 budget from the state, while 13 received over two-thirds. This variable was usually stable for 1995-98, varying by less than 5% for 64.4% of the districts, and by less than 10% for 90.0%.

To assess synergistic effects from both existent charter competition and the vulnerability produced by a heavy reliance on state funding, we included an interactive term, e.g., market share times subsidy. Entry plus a high subsidy may intensify competition, spurring more district changes. However, organizational constraints may limit the ability of districts to respond beyond a certain point. If so, the relationship will be negative as districts reach the limit of their ability to respond. Because we hypothesized that response to competition may be reduced by school size, as larger schools may be more difficult to control, we also included a measure of school size. Finally, we controlled for the extent of 1994-95 outreach activity since-if organizations indicate they do more of what they already do-experience with outreach may produce an increase in school outreach, all else equal. The hypothesized relationship means that the associated coefficient is expected to be positive. Table 2 presents descriptive statistics for these variables.

Regression Segmentation

Three regressions are reported for each outreach tool: the aggregate regression (87 schools); the schools with relatively less-empowered staffs (the 34 schools that fell below the 40`" percentile on empowerment); and the schools with relatively more-empowered staffs (the 34 schools that were above the 60`" percentile on empowerment).7 Our empowerment proxy is reported teacher control over school curriculum in 1994-95 (prior to the introduction of charter schooling). This question read: "Using the scale provided, how much influence do teachers in your school have over establishing curriculum?" The variable was reported on a 6-- point scale, with the anchor at 1 being "No influence" and the anchor at 6 being "complete influence." Teacher control over curriculum is a useful empowerment proxy because of the central role of curriculum in shaping school organization and mission (Chubb & Moe, 1990; Tyack & Cuban, 1995). As discussed previously, schools with empowered staffs are more decentralized and expected to react more quickly to competition. The lower 40% of schools all scored less than 2.92 on curriculum control (with a mean of 2.48); the top 40% scored above 3.36 (with a mean of 3.86). We eliminated the 20% of the schools that were clustered around the median empowerment level of 3.00, because the relative centralization for these schools was not obvious. When doing this segmentation, we used 1994-95 values-not changes in value-so as to avoid segmenting based on the dependent variable.

One issue that must be considered is whether the analytic results might suffer from the "regression to the mean" phenomena, e.g., organizational traits significantly above or below the norm will tend to converge over time. However, this would only bias our results if it were correlated with our independent variables. Otherwise it would wash out across the range of sample districts. There is no reason to suspect this effect to be correlated with the other key independent variables.8

Findings

Consistent with expectations, competition as measured by charter school percentage is associated with school-level outreach variables. Examining the effects of competition on school-level outreach, the aggregate analysis presented in Table 3 shows that the competition variables generally have significant positive effects (p < .05). However, the segmented regressions tell a more complex story. Increased charter market share actually reduced efforts to inform parents about school performance at those schools with more centralized governance, while increasing efforts in the schools where faculty play a relatively larger role in decisionmaking (both were significant at p < .05). The reduction in outreach by centralized schools is a puzzling finding that seems counterintuitive. The most likely explanation may be that schools respond to crisis by doing what they already know how to do.9 This could mean that centralized schools batten down the hatches in the face of competition, as principals seek to more carefully regulate teacher and school behavior (including communication with the parents or the community). Such a reaction could produce a short-term reduction in outreach activity.10

Especially in the case of schools distributing informational flyers, both charter market share and the state subsidy percentage lead to statistically significant increases in use (p < .01). However, when the regression is segmented, the competition measures only prove to have this effect in the relatively empowered schools. They have no significant effect in the less-empowered schools. In short, we generally find statistically significant responses to charter market share and to state subsidy percentage only in schools where the faculty are involved in decisionmaking. There is a caveat to this finding, however. The interaction term (subsidy*market share) is negative and statistically significant for both flyers and parental communication in decentralized schools (p < .10). Districts subject to both subsidy and market share pressure evince less of a reaction than we would otherwise expect. Schools with decentralized decisionmaking respond to either latent or realized competition up to a point, but then the response appears to taper off.

To assess the nature of this ceiling, we insert representative values for subsidy and market share into the regressions and estimate the effects. We insert the value at the top of the lowest quintile of districts and the value at the bottom of the top quintile for both market share and state subsidy percentage. This allows us to calculate the results of a 3-quintile change in one or both competition variables. This 3-quintile change produces market shares ranging from 0% to 7.74% and state subsidy percentages ranging from 36.6% to 75.2%.

In schools where faculty are relatively more empowered, increasing market share alone from 0% to 7.74% moves teacher opinion closer to the "strongly agree" position that schools are making extensive efforts to inform parents by .34 (a 6.8% increase on a 1-6 scale); increasing subsidy alone produces a 0.73 (14.6%) increase; and increasing both a 0.57 (11.4%) increase. In schools with more empowered faculties, increasing charter share moves teacher opinion closer to the "strongly agree" position that schools are using flyers and other media by 0.30 (6.0%); increasing subsidy percentage increases their use by 0.49 (9.8%); and increasing both increases their use by 0.31 (6.2%). In other words, a 3-quintile increase in either or both measures of competition moves teacher opinion closer to the "strongly agree" position that decentralized schools are boosting outreach efforts by somewhere between 6% and 14% on the 6-point scale.

By comparison, the case ot less-empowered schools, schools with an increased charter market share were reported to be making less extensive efforts to contact parents. Efforts were reportedly reduced by -0.62 (-12.4%), while increased state subsidy percentage boosted efforts by 0.07 (1.4%). An increase in both produced a significant increase in outreach 0.72 (14.4%). Seemingly, while less-- empowered schools normally do not increase outreach in response to competition, a dramatic double-barreled threat caused them to increase outreach as much as their decentralized associates. (As noted earlier, this result does not apply to the more specific use of flyers and other media-competition had no effect on those in the centralized schools.) In short, when market share alone increased, schools with empowered staffs were more likely to react; when both subsidy and market share increased, both kinds of schools reacted in similar fashion.

Overall, increasing both subsidy and market share appeared to have slight (for flyers) to moderate (for informing parents) positive impacts on outreach. In no case was school size statistically significant, though this may be due to the lack of variation in school size in the sample (see the descriptive statistics for this variable in Table 2). As for the prior outreach level variable, it is negative and statistically significant (p < .05) for the less-empowered schools and vaguely positive for more-empowered schools. Contrary to our expectations, it appears that prior experience in outreach is not correlated with increasing outreach.

An important qualification must be added to our results: we can only assess the impact of competition in the short run. As noted earlier, much of the increase in charter school enrollment in Arizona occurred in the 1997-98 period. Hence many of the schools facing actual charter competition in our sample had no more than a year to respond to this competition. Whether this response will grow in the long run, stay the same, or even decline, is unknown at this time.

Conclusions

In general, we find school-level effects from competition at schools with relatively decentralized governance for both informing parents and the more specific use of flyers and other media. A 3-quintile increase in competition prompted about a 10% increase in likelihood that teachers at more schools where curricular control was more decentralized would characterize their schools as communicating more frequently with parents. On the other hand, competition did not induce an increased use of flyers at schools where curricular control was more administratively centralized, and only prompted an increase in broader outreach efforts when competitive threat was rather high. These findings suggest that flexible organizations are somewhat more likely to respond to competition. This accords with the theoretical literature on schools and organizational change, which suggests that responses to competition will be mediated by the prior organizational structure, e.g., by the degree to which faculty are already involved in school-level decisionmaking. Contrary to what some conventional thinking might lead us to expect, however, we do not find that prior organizational experience with outreach leads to greater outreach.

These observed effects of competition were generally small. However, there is evidence that competition has modestly increased the time and energy that some schools put into communicating with parents. Other things being equal, this would seem to be a useful change in its own right and one that may enhance accountability. However, it is important to recognize the relatively modest size of these initial effects. Moreover, the effects are not evenly distributed, and the fact that decentralized schools are more likely to respond than are more centralized schools could potentially accentuate informational inequities.

This research represents a first cut at the question of how competition affects outreach. Further efforts are necessary if we are to understand the size and the nature of these changes and their implications for educational accountability. First, our methods capture teacher perceptions of the extent of outreach activities, not the precise amount of outreach attempted. Moreover, our instrument can only measure the amount of outreach reported; we do not know what the content of these communications was like, or whether it actually conveyed meaningful information on school performance or services.

Second, we are able only to study the short-term impact of competition. Longer-term effects are beyond the scope of our analysis here. This means that we cannot determine whether the effects we observed are early indicators of more substantial changes to come, the entire realized effect, or a symbolic response masking more fundamental inertia. However, even over the short time period studied, a few Arizona school districts had 20% or more of their potential enrollment to charter competition. A study of four districts found short-term losses of that magnitude associated with more significant changes in district behavior than those documented here (Hess, Maranto, & Milliman, in press). Interviews with district officials suggest that they, rightly or wrongly, often appear to use charter market share as a proxy for district school effectiveness. Consequently, while the findings presented here are certainly preliminary, even the short time period studied does provide the opportunity to begin to assess the revealed impact of competition on school outreach efforts.

Third, the evident impacts seem to occur even though competitive pressures were moderated in many Arizona districts due to high enrollment growth. The presence of the observed effects, despite the rapid growth of the Arizona school population in the 1990s (which would lessen the pressure to compete for students), suggests that our estimates of revealed effects may actually be biased downward.

Finally, while outreach efforts may serve to enhance parental and community knowledge of local schooling, they might also serve as a hollow public relations device. Moreover, efforts at outreach could potentially siphon resources from teaching and learning. Sorting these questions out requires further research and analysis. We do not examine the content of the outreach efforts that schools are making, so it is impossible to determine what it is that schools are communicating or how the content of such efforts varies across schools.11 Research into the content of outreach efforts promises to be a rewarding and useful endeavor for future research, but one that was beyond our means in this initial effort.

Our findings suggest that markets may encourage traditional public goods providers to increase outreach efforts, but it is not clear that they will necessarily do so. The nature and scope of such efforts appear to depend greatly on organizational structure, with the evidence suggesting that decentralized schools may be more likely to increase outreach in answer to competition. While our results are tentative, such a finding would be consistent with previous scholarship in the field of public administration. Although it might seem obvious that public organizations would respond to competitive threats by making customers aware of their services and their performance, such a response depends in part upon organizational resources. Markets may indeed spur schools to communicate more with parents, but the short-run extent of this response is not uniform, and the associated long-run impacts remain unknown. These questions require scholarly scrutiny. We hope that this study may help suggest a model for more systematic and less subjective evaluations of the promise that choice-based competition has for changing the behavior of public schools and other public agencies.