Getting Past “Purposeful”: Exploring Dimensionality in Nonprofit Executive Performance Information Use

Although performance information use (PIU) among public managers is a growing and increasingly relevant research area, the existing evidence base has two significant limitations for those interested in its application to nonprofit executives. First, large survey investigations, the predominant method used to assess PIU behaviors, have rarely sampled outside of government. Second, despite theoretical arguments and empirical support for PIU being a multidimensional behavior, only ‘purposeful’ use (i.e., the deliberate and instrumental use of performance information in decision-making to improve organizational operations) has been examined with any regularity. Thus, in addition to developing theory around PIU for nonprofit executives (rather than just public managers within governments), I test established drivers of purposeful and political PIU using survey data from 260 nonprofit executives throughout the United States. Results show that nonprofit executive PIU is driven by different considerations than public manager PIU. Additionally, results show that leadership support of performance measurement is an important driver of purposeful and political PIU, with organizational goal clarity and networking behavior also, specifically, driving political PIU.

In this article, I develop theory around nonprofit executive PIU and respond to these limitations by re-testing established drivers of purposeful and political PIU among nonprofit executives. Two research questions guide this inquiry. First, what drives nonprofit executives to use performance information purposefully and politically? And, secondly, do different drivers promote specific types of use?

Literature Review
There is an increasing body of empirical research showing that the adoption and use of performance management in the public sector does not always lead to the desired effects (Kroll & Moynihan, 2017;Poister et al., 2013). Explanations for this varied performance of performance management are multifaceted (Moynihan & Kroll, 2015). Of particular interest in this study, however, is the failure of managers to use information to drive decisions. This failure is the primary mechanism of organizational change in performance management doctrine. Indeed, when information is not utilized, managers establish organizational processes for which creating data is a suitable end in-and-of itself (Kroll, 2015). Ultimately, these managers fail to transition from performance measurement into performance management.
Despite often being used interchangeably, performance measurement and performance management are not synonymous. Performance measurement explicitly refers to "the regular collection and reporting of information about the efficiency, quality, and effectiveness of…programs" (Martin & Kettner, 1996, p. 3). Performance management, meanwhile, is widely understood as a system of internal organizational processes (e.g., rewards and sanctions) based on regular, formal tracking of quantitative objectives to achieve results (Hatry, 2007;Melkers & Willoughby, 2005;Speckbacher, 2003). Performance measurement may occur in the absence of meaningful performance management, but effective performance management requires good performance measurement.
Performance management is generally characterized by PIU in decision-making. It is, therefore, reasonable to question-what actually constitutes substantive PIU? Although definitions of PIU abound (e.g., Kroll, 2013;Lee & Clerkin, 2017;Johansen et al., 2018;Moynihan, 2008;Moynihan, Pandey, & Wright, 2012a), studies conceptually converge on PIU as the deliberate and instrumental use of performance information in decision-making in order to improve organizational operations (Johansen et al., 2018). This is what Moynihan (2008) refers to as purposeful use.
Since the desired outcome of performance management is well-evidenced incremental improvement, the empirical focus on purposeful use as the default form of PIU is perhaps unsurprising. Still, the focus on purposeful use among government managers within the current body of research means that the differences between nonprofit and governmental actors are undertheorized and largely unexamined.
Although the topic of performance management, particularly PIU, in the nonprofit sector is less studied than it is in governments, there is rich nonprofit research on evaluation use (EU). PIU and EU are behaviors related to the two primary forms of knowledge production about the performance of public services or public serving organizations: performance management and program evaluation respectively. Although related, EU and PIU have important conceptual distinctions, different epistemic logics, and different professional audiences (Kroll & Moynihan, 2017).
On the one hand, program evaluation leans heavily on applying rigorous social research methods to the assessment of design, implementation, and impact of intervention programs. Performance management, on the other hand, is used to assess program outcomes in order to understand performance and hold managers accountable for results (Heinrich, 2007).
In this article, I draw primarily on performance management traditions. Nonprofit scholarship on performance management has demonstrated that there is substantial variation in nonprofit executives' attitudes toward it. Indeed, nonprofit executives are not universally positive about performance management (Caers et al., 2009). Although some view it purposefully, as a strategic management tool, others consider it to be a political tool for marketing and promotion (Carman & Fredericks, 2008). Nonprofit executives, therefore, increasingly find themselves stretched in two ways.
First, these executives are often stretched purposefully; that is, they are expected to evidence the completion of activities and the achievement of outcomes in order to continually improve and deliver against their articulated mission. Secondly, they are stretched politically; that is, they are expected to evidence the completion of activities and the achievement of outcomes, often specified by funders, to demonstrate contract fulfilment, secure legitimacy, evidence effectiveness, and ensure continued or additional resources (Moynihan, 2008;Tassie, Murray, Cutt, & Bragg, 1996;Tassie, Murray, & Cutt, 1998;Eckerd & Moulton, 2010).
For nonprofit executives, this duality is particularly salient since funder discourse regarding accountability, which initially centered on making nonprofits more transparent concerning their fundraising, spending, and governance, has shifted to the demonstration of "impact" in "addressing complex social problems such as poverty and inequality" (Ebrahim & Rangan, 2014, p. 118;FitzGerald et al., 2020). Thus, performance reports tend to now require these organizations to include social outcomes as more comprehensive measures of organizational achievement.
This requirement has, inevitably, made competition for funding more focused on evidence of organizational impact (Lee & Clerkin, 2017;MacIndoe & Barman, 2013;Mitchell & Berlan, 2017;Witesman & Fernandez, 2012;Thomson, 2010;Ebrahim & Rangan, 2014); and, nonprofit executives now operate in an environment where their rationale for using information may be more directly driven by the preferences of those who provide financial support to their organizations (e.g., contractors and donors). This could mean that purposeful and political PIU will be mutually beneficial for them.
The following sections offer an overview of existing studies on performance management and drivers of purposeful PIU. This is followed by a presentation of the hypotheses on relationships between various stakeholders' support for performance measurement and corresponding purposeful or political PIU by nonprofit executives. I then test these hypotheses using hierarchical regression analysis. I conclude the article with a discussion of the results; and, an overview of the study's limitations as well as directions for future research.

Drivers of Performance Information Use
To date, there has only been one systematic review, which included 25 studies, on drivers of purposeful PIU (Kroll, 2015). In this review, drivers were categorized as either "important," "promising," or "insignificant" depending on the strength and consistency of the evidence base linking each one to purposeful PIU. Across the 25 studies, the mean R-squared was 39%, indicating that the included drivers accounted for a substantial amount of observed variance in purposeful PIU. For the purposes of this study, I focus on drivers that Kroll (2015) categorized as either "promising" or "important," since these variables may relate to different patterns of PIU among nonprofit executives (see Table 1). For a summary of hypotheses presented in this section, see Table 2.
Stakeholder Involvement. Nonprofits operate in politicized environments where performance measurement can function as a transaction cost imposed in order to support the monitoring of grants and contracts by funders (Carman, 2011;Carnochan, Samples, Myers, & Austin, 2014;MacIndoe & Barman, 2013). This arrangement allows external stakeholders, including government agencies, foundations, individual donors, and national headquarter organizations, to resemble principals who rely on nonprofit agents to deliver services. Performance measurement, then (like evaluation), acts as a transaction cost imposed on nonprofits so that funders can monitor the impact of grants and contracts (Carman, 2011)-potentially, at the expense of direct service provision (Lynch-Cerullo & Cooney, 2011). Nonprofit executives are likely to use mandated performance measurement not only in accordance with contractual obligations, but also to manage the principal-agent relationship and advocate for continued support (Davis, Schoorman, & Donaldson, 1997;Van Slyke, 2007).
Qualitative evidence suggests that the opposite could also be true, however. Indeed, case studies of 18 Detroit-based nonprofit organizations demonstrated that although funders' reporting requirements boosted performance measurement, they "generally do not lead to greater use of outcomes measures in decision-making overall" (Thomson, 2010, p. 54). Therefore, external stakeholders may still be needed to track whether performance plans, goals, reporting, and measurements are followed, which could be difficult if these stakeholders do not have the in-depth programmatic knowledge necessary to assess managerial information use.
External stakeholder involvement, then, could actually encourage a more passive form of PIU, where performing data use is principally a way to appease grantors, funders, or boards without changing operations. Thus, I propose the following competing hypotheses: Hypothesis 1: External stakeholder support of performance measurement is positively associated with political PIU.
Hypothesis2: External stakeholder support of performance measurement is not associated with political PIU.
Internal stakeholders (e.g., board members, clients, and staff), on the other hand, do possess the programmatic knowledge necessary to encourage executives to remain up-to-date and make databased decisions Bourdeaux & Chikoto, 2008;Ho, 2006;. Not surprisingly, then, PIU studies have shown that when managers are aware that these internal stakeholders care about performance, they work to stay current on their program, department, and/or organization's data, performance trends, and explanations of outliers Bourdeaux & Chikoto, 2008;Ho, 2006;Moynihan & Hawes, 2012a;. Thus, I propose: Hypothesis3: Internal stakeholder support of performance measurement is positively associated with purposeful PIU. For nonprofits, the pressure to engage in performance measurement often comes from external sources as a condition of receiving funding (e.g., contractual obligations to government-run human service agencies and demands for financial and program accountability by private foundations) (Ebrahim & Rangan, 2014;FitzGerald et al., 2019). It is less frequent that performance management initiatives come from internal sources (Carnochan et al., 2014) suggesting that performance measurement is less intrinsically valuable to nonprofit executives. Thus, I propose: Hypothesis4: External stakeholder involvement is a stronger predictor of political PIU than internal stakeholder involvement is of purposeful PIU.
Networking Behavior. Service providing nonprofits increasingly find themselves participating in community-based interventions that have been implemented through local partnerships (Butterfoss, 2007). A number of academics, however, have criticized these networked efforts for failing to provide bureaucratic-like accountability (e.g., Kroll, 2015;. Thus, assessing whether nonprofits operating in these networks are more likely to use performance information could have implications for network effectiveness. In this context, using performance information may represent a strategy employed by nonprofits to "manage and mitigate the effects of inter-organizational relationships and the environment" (Carman, 2011, p. 354).
Although organizations might be in partnership around common goals, these inter-organizational relationships "can become political struggles in which 'different parties [seek] to influence each other to their own advantage'" (Carman, 2011, p. 354 quoting L. Donaldson, 1995. This suggests that network participants are likely increasingly aware of their own organizational performance goals as well as the network's. As such, they may be better positioned to use data to promote or defend their programs and lobby for resources . Given this possibility, nonprofit executives may use performance data as a way to manage network relationships. If so, executives who report higher reliance on, and engagement in, collaboration may also report using performance information more regularly. Thus, I propose: Hypothesis 5: Networking behavior is positively associated with political PIU. After imputation to account for minimal but non-random missing data in 31 responses (Garson, 2015), original data included 260 responses for a response rate of approximately 17%. These original data along with five imputations were used in this analysis.
In terms of geographic spread, according to the U.S. Census Bureau regional definitions, 74 organizations in the sample were located in the South (approximately 29%). Fiftynine organizations were located in the West (approximately 23%). Fifty-two organizations were located in New England (approximately 20%); and, 46 organizations were located in the Midwest (approximately 18%).
The average total revenue of organizations in the sample was $2,527,215; and, the average total assets were valued at $5,282,706. I conducted one-sample T-tests on total revenue and assets. Results from these tests indicate that the senior executives in this study are more representative of larger service providing youth development nonprofits.
The survey utilized previously validated scales to measure public service motivation (Wright, Christensen, & Pandey, 2013), perceived social impact , developmental culture (Kroll, 2013;Zammuto & Krakower, 1991), and both forms of PIU . All items included in the analysis are provided in the Appendix (see Table A1).
The scale measuring purposeful PIU aligns with common use behaviors. Previous research has shown that these behaviors tend to load onto a single factor (de Lancer Julnes & Holzer, 2001;Kroll, 2015); and, this study is no exception. Regarding political use, Moynihan, Pandey, and Wright (2012a) were the first to demonstrate its distinctness from purposeful use. I use their scale in full in this analysis.
Using principal component factor analysis, I was able to confirm convergent and divergent validity between purposeful and political use. That is, all items (a total of eight) loaded as expected and provided evidence of two distinct latent factors (see Table 3).
Tables 4 and 5 provide descriptive statistics and correlations. Correlation is low between independent variables suggesting that multicollinearity is not a concern. This is further confirmed by variance inflation factors ranging from 1.12 to 2.07, which do not approach five (i.e., the typical threshold for problematic collinearity) (Garson, 2014). There is some overlap in dependent variables, suggesting that executives who use data tend to do so both purposefully and politically. However, I do not use these simultaneously in the model estimations.

Controls
I include leadership support, goal clarity, support capacity, innovation culture, and prosocial motivation as control variables in order to test their application to nonprofit executives and better investigate the hypothesized effects of stakeholder involvement and networking behavior . Goal clarity is included as a control to help mitigate the effects of sampling from a profession that lacks standardized performance indicators and change models. This is a fact made more complicated by the variety of youth-serving organizations included in the sample.
Studies have shown that the success of performance measurement and management systems is dependent on the level of support, often in the form of time, personnel, money, and information technology, that is extended during adoption and implementation (Kroll, 2015, p. 12). This 'support capacity' enables organizations to make the most of their performance measurement system by ensuring adequate training and access to employees. Prior research demonstrates that the items used to measure support capacity in this study have conceptual validity Kroll, 2015). The present study, however, is the first study to use these items in a scale. The Cronbach's alpha for this scale (α=0.85) performed well above the 0.70 cut-off for confirmatory use (Garson, 2012).
Organizationally, innovation culture is thought to help establish low stakes learning environments and enhance a group's natural proclivity to improve (Kroll, 2015;. Innovation culture and support capacity are measured using validated scales scored with average indices (for full details, see Table A1 in Appendix).
Public service motivation (PSM) has been defined as an extra role behavior, where employees make gifts of time and effort without expectation of individual reward Saliterer & Korac, 2014). Individuals with high PSM are believed to care about publicly minded organizational goals. They are, therefore, more likely to use performance information as a tool to achieve those goals Perry & Wise, 1990;Saliterer & Korac, 2014;Waterhouse, 2008).
PIU scholars believe that employees who sense the public benefit of their work (i.e., individuals with heightened perceived social impact (PSI)) are also more likely to use performance information to achieve the goals they value. Likewise, PIU scholars have suggested that employees who see the value of their work are more likely to seek support from external stakeholders . For nonprofit executives who operate in an environment where competition for contracts and grants has increased "performance pressures and expectations for measuring outcomes" (Mitchell & Calabrese, 2018, p. 2) and where external engagement strongly influences financial performance, performance information may be considered a weapon capable of legitimating services and assisting the organization in obtaining resources . Thus, I include PSM and PSI as control variables to account for prosocial motivation, measured in this study through the use of validated scales (see Table A1 in Appendix).

Addressing Common Source Error Concerns
Self-reported responses like the ones used in this study have come under increasing scrutiny given the likelihood of reporting bias, including common source bias (Meier & O'Toole, 2013). While there is debate over the degree to which these concerns are exaggerated (George & Pandey, 2017), this investigation does incorporate mitigation techniques into data collection.
First, the survey items were designed to mitigate the unfavorable effects of common source bias by focusing on observable behavior over a specific time period (Meier & O'Toole, 2013;. Second, the survey included passages assuring participants that their Leadership Support Goal Clarity R 2 (Imputation 1) Adj. R 2 (Imputation 1) n=260

Results
This investigation leveraged existing evidence on the drivers of purposeful PIU by public managers in order to examine whether the same patterns of information use emerged for nonprofit executives. The results of the purposeful and political use partial and full regression models are summarized in Table 6. The results are presented with pooled coefficients and standard errors.
The primary goal of this study was to assess whether different considerations drive nonprofit executives to use performance information than public managers. This analysis supports this notion. When considering the control variables included in the analysis, as shown in Table 6, only leadership support was significant in driving purposeful PIU. This likely indicates that there are contextual differences influencing nonprofit executives compared to public managers in government. Leadership support, goal clarity, and networking behavior, meanwhile, were shown to significant drivers of political PIU by nonprofit executives.
Considering the hypotheses related to stakeholder involvement, the results (in Table 7) show that external stakeholder support of performance measurement is not significantly associated with nonprofit executives' use of performance information, whether politically or purposefully. Thus, Hypothesis 2 (i.e., External stakeholder support of performance measurement is not associated with political PIU) is supported. Hypotheses 1 (i.e., External stakeholder support of performance measurement is positively associated with political PIU) and Hypothesis 4 (i.e., External stakeholder involvement is a stronger predictor of political PIU than internal stakeholder involvement is of purposeful PIU), however, are not supported.
Internal stakeholder support, which I hypothesized was positively associated with purposeful PIU, lacks significance in the full model. Thus, as shown in Table 7, Hypothesis 3 (i.e., Internal stakeholder support of performance measurement is positively associated with purposeful PIU) is not supported. Increased network activity, however, is significantly and positively associated with political use in support of Hypothesis 5 (i.e., Networking behavior is positively associated with political PIU).

Discussion
The primary goal of this study was to investigate the relationship between evidenced drivers of purposeful PIU among executives of nonprofit organizations and political PIU. The findings from this study offer two important contributions.
First, research on purposeful use has primarily been carried out in government settings. As such, the existing stream of PIU literature fails to capture important contextual differences and alternative types of PIU that might lead to greater generalizability. In this study, I find that drivers of purposeful PIU by nonprofit executives are demonstrably different than drivers for public managers. This was evidenced by the number of hypotheses that were not empirically supported as well as the number of control variables that were not significant.
Of the control variables, only leadership support was significantly associated with increased purposeful PIU. Moreover, organizational and individual characteristics (e.g., support capacity and innovation culture as well as prosocial motivation) were not significantly associated with nonprofit executives' PIU. This could suggest that leader support for performance measurement (as opposed to an underlying individual motivation base) in tandem with the influence of professional networking (as opposed to organizational attributes) is ultimately what drives nonprofit executives to use information purposefully and politically.
Second, and relatedly, this study further supports the multidimensional nature of PIU. Many nonprofit executives view performance management as a promotional tool (Carman & Fredericks, 2008). Indeed, given that organizational survival is predicated on securing funding, some of these executives may see political PIU as a way to boost financial performance an ultimately, organizational performance. For nonprofit executives, then, activities associated with political PIU may be rationalized as serving the same overall purpose as those associated with purposeful PIU. This may not be the case for public managers who work with assigned budgets in government. Nonprofit executives, on the other hand, may use different dimensions of performance information simultaneously or sequentially. As such, for these executives political PIU may be a means to improve organizational performance.
The findings in this analysis align with those of Thomson (2010), who found that funders' reporting requirements do not necessarily lead to greater use of information in decision-making (Thomson, 2010, p. 54). Although understanding the mechanisms underlying this finding is beyond the scope of this study, it is probable that external pressure(s) beget a passive form of PIU where data is used to appease, rather than substantively change or lobby.
Given that nonprofits often deliver services through service delivery networks and complex governance structures, nonprofit executives may operate by means of persuasion rather than hierarchy. This can, undoubtedly, make information regarding outcomes a useful tool for obtaining further resources. In this instance, political PIU by nonprofit executives may boost performance as a result of the environment in which the organization operates (i.e., with pressure from funders, clients, and network partners alike). Perhaps it is for this reason, then, that findings from this study indicated that networking behavior, as opposed to external support, was strongly related to political PIU.
For nonprofit executives operating in complex inter-organizational arrangements, networking behavior may be related to PIU through from formalized network membership requirements, competitive pressures for finite funding, and/or through standards of professional practice that foster mutual learning. On the one hand, networks provide an opportunity for formal professional requirements to be disseminated; and, depending on the network, measurement may be a requirement for participation. Combined with the tendency for funding to be tied to participation in networks, the relationship between networking behavior and political PIU may be further spurred by competitive behavior. That is, there may be a desire among partners to demonstrate legitimacy (potentially by demonstrating their superiority relative to other network participants); and, ultimately, that they deserve a greater share of available funding.
On the other hand, considering the importance of networking behavior and the lack of significance around prosocial motivation, it may be that peer engagement (rather than supervisory or subordinate support) is useful for expanding purposeful and political PIU. Previously, scholars have suggested that exploring ways to "connect individuals to the impact of their work" might bring about improved adherence to public sector management reforms (Moynihan et al., 2012a, p. 476). The findings from this study do not refute this, nor does this study directly address the criticisms that networks fail to provide bureaucratic-like accountability. This study does, however, show that network pressures can influence individual behavior to adhere to reform rules and promote accountability (Moynihan & Hawes, 2012a).
In other words, connecting individuals to the impact of their work may not be enough to garner support for performance management reforms from individuals who might otherwise be inclined to resist them. Likewise, funder mandates may be equally prone to failure. Creating a space for nonprofit executives to learn from each other and be exposed to best practices in performance measurement, however, may create favorable conditions for executives to obtain buy-in of performance management reforms.

Study Limitations
There are a number of limitations that should be considered. First, the topic area and single informant research design invite the possibility of social desirability bias and common-source error. Although the survey included explanatory passages about the non-evaluative nature of the questionnaire and also included limited perceptual measures, the extent to which these inclusions biased the results is unknown. Still, the inclusion of behaviorally-focused, temporally-grounded items (e.g., environmental support, observable behaviors, and managing in networks) (Meier & O'Toole, 2013) means that the findings are less likely to be biased. This is an improvement on prior studies in this area.
Second, statistical analyses performed on previously validated survey items and scales align with the items and scales used in this study. Still, I cannot guarantee absolute consistency in participant interpretation.
Third, the survey in this study utilizes data from across the United States, which is a distinct advantage among nonprofit performance measurement and management studies that have primarily only investigated one to a few organizations, are limited to a single state, or use a case study approach (Carman, 2007(Carman, , 2009LeRoux & Wright, 2010). However, the low response rate and focus on youth-serving nonprofits substantially limits external validity claims. Thus, future research should focusing on increasing the response rate across a broader array of services.
Overall, the findings presented here have provided fruitful grounds for future research. Indeed, given the importance of networking behavior in promoting both types of PIU, parsing the relative importance of peer pressure (as a form of accountability) is a valuable line of inquiry. Assessing whether nonprofit executives are more likely to use performance information when operating in networks may have important implications for governance effectiveness and ongoing support for inter-organizational public sector managerial reform initiatives (Kroll, 2015;. Moreover, additional inquiry around the multi-dimensional nature of information use across different contexts will be essential in developing more generalizable theories of PIU.

Conclusion
Results from this study suggest that there are perhaps greater incentives for and pressures on nonprofit executives to use performance information politically than purposefully. In comparing the explanatory power of each model, it is clear that many of the variables included are better predictors of political use. This may represent a departure from increasingly popular outcomesbased models, which identify monitoring as a mechanism for increased efficiency through incremental organizational improvement. The findings from this study suggest that, for nonprofit executives, political PIU is conceptually related to performance and that future research should endeavor to better understand the links between non-purposeful forms of PIU and performance. This finding introduces an important caveat regarding this line of research. PIU has become a proxy outcome measure for performance-oriented reforms. However, this is not the ultimate end goal of performance management doctrine. Understanding drivers of information use is important, but as a milestone, not an end goal. The value in this research is in understanding the mechanisms and contingencies through which information use improves, or does not improve, actual organizational performance.

Disclosure Statement
The author declares that there are no conflicts of interest that relate to the research, authorship, or publication of this article.