The Accuracy of the Core Files for Studying Nonprofit Location

Abstract

Studies show that the distribution of nonprofits varies considerably across communities. Affluent communities tend to have ample nonprofit resources and highly diverse nonprofit landscapes, whereas low-income communities often lack the variety of nonprofits found within wealthier areas. As a result of these differences, scholars have suggested that geographic unevenness in the presence of nonprofits may lead to extreme inequities and inefficiencies in how nonprofit services are accessed and administered. Although these concerns certainly warrant serious attention, several limitations have been acknowledged with the National Center for Charitable Statistic’s (NCCS) Core Financial Files—which have been the primary data source used to generate findings on geographic dimensions of the nonprofit sector in the United States. This research note examines the accuracy of the information in the Core Files after adjustments for each of these limitations.

Keywords

nonprofit location nonprofit data nonprofit organizations

Studies show that the distribution of nonprofit organizations varies considerably across communities. Affluent communities tend to have ample nonprofit resources and highly diverse nonprofit landscapes, whereas low-income communities often lack the variety of nonprofits found within wealthier areas (Bielefeld, 2000; Grønbjerg & Paarlberg, 2001; Joassart-Marcelli & Wolch, 2003; for exceptions, see Corbin, 1999; Peck, 2008). As a result of these differences, scholars have suggested that geographic unevenness in the presence of nonprofits may lead to extreme inequities and inefficiencies in how nonprofit services are accessed and administered. Mohan, Twig, Jones, and Barnard (2006), for instance, have pointed out that “the safety net represented by the nonprofit sector ha[s] a ‘mesh of varying size,’ so that the probability of slipping through it varie[s], depending on location” (p. 267); and Eikenberry (2005) has claimed that due to geographic unevenness in the nonprofit sector, heavy reliance on nonprofits in the delivery of public services could “exacerbate rather than ameliorate social and economic inequalities” (p. 1).

Although these concerns certainly warrant serious attention, several limitations have been acknowledged with the National Center for Charitable Statistic’s (NCCS) Core Financial Files—which have been the primary data source used to generate findings on geographic dimensions of the nonprofit sector in the United States.¹ Specifically, the Core Files, which include approximately 60 financial variables from a nonprofit’s Form 990, are based on the Internal Revenue Service’s (IRS) Return Transaction Files (RTFs); and as the RTFs are mainly used for regulatory (and not research) purposes, Lampkin and Boris (2002) have argued that despite rigorous data-cleaning protocols, errors exist in the Core Files due to the nature of how the data are constructed.

Perhaps the most challenging of the limitations associated with the Core Files for studying nonprofit location include (a) the validity of the address information, (b) the presence of post office (P.O.) boxes, and (c) the use of headquarter addresses to account for nonprofits operating at multiple service locations.² Each of these issues poses serious problems for researchers attempting to obtain an accurate estimate of the number of nonprofits located in an area. This research note, therefore, examines the accuracy of the information in the Core Files after adjustments for each of these limitations.

Challenge of Using the Core Files for Locational Analysis

When conducting locational analyses of the nonprofit sector, it is important to have complete address information, or at the very least, some sense that the information contained in the files is accurate. In the absence of such information, it can be difficult to determine if research findings truly present an accurate picture of the nonprofit sector in an area. This research note focuses on common limitations associated with the Core Files for assessing nonprofit location.

Incorrect Address Information

One of the most frequently acknowledged limitations of the Core Files for locational analyses is the accuracy of the address information (Lampkin & Boris, 2002). A 1994 study by the IRS found that 27% of nonprofits in the Core Files had incorrect addresses listed on their Form 990 (NCCS, 2006). The NCCS, therefore, has recommended that researchers incorporate data-cleaning strategies when relying on the location information contained in the Files. Yellow Page listings in particular, NCCS (2006) has suggested, are one source that can be used to verify a nonprofit’s location. Joassart-Marcelli and Wolch (2003), however, found that 15% of nonprofits in the Core Files for southern California could not be verified even when searching local public directories, such as Yellow page listings, thus providing evidence that some nonprofits in the Core Files are not located in the associated area.

P.O. Box Listings

It has also been acknowledged that some nonprofits do not include their operating address on their annual tax return (Form 990). Instead, many provide address information associated with a P.O. box. However, P.O. boxes provide location information for a centralized delivery system (such as a postal provider), and not the location of the organization. In recognition of this limitation, some scholars have begun locating the operating addresses of nonprofits with P.O. boxes listed in the Core Files. Peck (2008), for instance, located the operating addresses of nonprofits with P.O. boxes in Phoenix, Arizona, and Twombly, De vita, and Garrick (2000) conducted surveys of nonprofits in Philadelphia to identify operating locations of nonprofits with P.O. boxes listed in the Core Files. In general, though, such attempts are rare.

Multiple Service Locations

It has also been acknowledged that larger nonprofits often operate at multiple service locations, and for tax purposes these organizations tend to file aggregate tax returns. Although these aggregate returns are efficient from an organizational standpoint, they usually only include the address of the organization’s headquarters location (Froelich & Knoepfle, 1996). In fact, it is rare that a “parent” nonprofit will file a tax return for an affiliate site. Thus, any nonprofit organization operating as an affiliate, satellite, or subsidiary of the parent organization will typically not be included on tax returns or listed in the Core Files.

Reliance on Core Files

Despite widespread recognition of these limitations, nonprofit researchers have often relied on the Core Files—with little or no modifications to the data—as a primary source when conducting studies assessing nonprofit location. Stater (2009), for instance, used the Core Files, with no modifications, to examine the permeability of the nonprofit sector across U.S. counties. Bielefeld, Murdoch, and Waddell (1997) used the Files, with no modifications, to examine how distance and demographics influenced the presence of nonprofits in Dallas County, Texas. Joassart-Marcelli and Wolch (2003) used the files, with some modifications (e.g., searching local public directories to verify organizational locations for a sample of the organizations), to examine the locations of nonprofits in southern California. They, however, did not address each of the location-related limitations identified here.

Without an attempt to address each of these limitations, studies examining nonprofit location may fail to capture the full extent of nonprofit activity in an area. Ultimately, such oversight may create issues for generalizability and could increase the possibility of making incorrect inferences regarding the importance of nonprofit location. This study explores this possibility.

Data

This research note focuses on Core File data in San Diego County (California). The county of San Diego is particularly well suited to undertake the research in this study based on, both, its size as the fifth most populous county in the United States and its substantial racial and ethnic diversity (U.S. Census Bureau, 2013). As a result of these two factors, nonprofit organizations in San Diego County tend to fill important community roles (Schumann et al., 2014). Indeed, previous research has shown that San Diego is an area highly variable in several dimensions believed to influence nonprofit functioning. Bielefeld (2000) found that although San Diego had fewer nonprofits per capita than other major metropolitan centers, the area had a considerably higher quantity of amenity-type nonprofits (e.g., arts and cultural, education) than other areas. Moreover, he found that even with fewer nonprofits per capita, San Diego directed a relatively large percentage of gifts and grants to human services nonprofits—potentially an indication of the extent to which these nonprofits are relied upon to address social and welfare needs in the area.

The primary data used in this study come from the 2007 Core File of public charities for San Diego County (n = 3,199).³ Findings are reported at the ZIP code level, as exact location information for some nonprofits in the file could not be verified. A ZIP code level analysis should provide a good indication of the usefulness of the Core Files for community-level research. Indeed, previous studies on organizational location, more generally, have shown that ZIP code level data can be a useful geographic unit of analysis for community-level studies (e.g., Small & McDermott, 2006).

Verification Methods

From June to August 2010, attempts were made to verify the address information (as of year-end 2007) for each nonprofit in the file.⁴ Verification consisted of querying several sources—including organizational websites (if available), GuideStar listings, local public directories, and other administrative data sources (e.g., California Secretary of State’s registry of charitable organizations, California Department of Public Health). After verification, all incorrect (or missing) ZIP code information was corrected. When location information could not be verified, nonprofits were contacted directly.⁵ When contacted, deference was given to information provided by the contact, as it was assumed that an individual working for (or volunteering with) a nonprofit would provide more accurate information. Based on these same methods, attempts were also made to identify nonprofits in the area that were operating at multiple service locations. When additional service locations were identified, they were included in the data set as separate nonprofit entities.

The following nonprofits were excluded from the analysis: those with a P.O. box address but no identifiable operating address, those located outside of San Diego County, those with an inaccurate Employer Identification Number (EIN), those with a rule date outside the time period under study, and those that could not be verified due to lack of available information. In total, 426 nonprofits were excluded resulting in a final data set of 2,773 organizations.

Findings

For ease of interpretation, the findings in this research note are presented in sections. Specifically, the findings focus, first, on the distribution of location-related issues across ZIP codes in San Diego County. Next, the findings focus on differences between quantities of nonprofit organizations listed in the initial Core File and those in the adjusted file. Finally, the findings focus on differences between revenue and expenditure values in the two files.

In the final data set, approximately 11% of nonprofits had a ZIP code listed associated with a P.O. box—even though the organization had an operating location in the county (n = 302). Approximately 9% had incorrect address information listed at the ZIP code level (n = 242); and an additional 11% (n = 311) of nonprofits that were not included in the initial Core File were identified as multiple service sites.⁶

Distribution of Location-Related Issues

Tables 1 and 2 show the distribution of nonprofits in each of these three categories across ZIP codes in San Diego county and by National Taxonomy of Exempt Entity (NTEE) code. Previous research has shown that the distribution of nonprofits often differs across the class and race composition of areas (Bielefeld, 2000; Grønbjerg & Paarlberg, 2001; Joassart-Marcelli & Wolch, 2003; for exceptions, see Corbin, 1999; Peck, 2008). Therefore, in Table 1, ZIP codes are sorted by average median household income level and divided into quintiles.⁷ Median household income information was obtained from 2008 San Diego Association of Governments (SANDAG) population estimates. These estimates are created using U.S. Census data and are based on 1999 dollars. The top quintile (Quintile 5) represents the total number of nonprofits located in the most affluent ZIP code areas of the county.

Table 1.

Distribution of Nonprofits With Locational Limitations, by Average Median Household Income.

	Q1			Q2			Q3			Q4			Q5
	Average median household income
	US$35,287			US$46,702			US$53,159			US$64,914			US$84,910
NTEE category	P.O. box	Inc. ZIP	Multi. Loc.	P.O. box	Inc. ZIP	Multi. Loc.	P.O. box	Inc. ZIP	Multi. Loc.	P.O. box	Inc. ZIP	Multi. Loc.	P.O. box	Inc. ZIP	Multi. Loc.
Arts, culture, humanities	23%	17%	5%	12%	14%	—	15%	7%	—	21%	7%	3%	20%	15%	3%
Education	4%	17%	6%	12%	9%	5%	10%	7%	9%	17%	19%	8%	21%	23%	6%
Environment, animal	17%	2%	1%	3%	11%	—	6%	7%	5%	6%	7%	—	7%	3%	3%
Health	6%	9%	35%	10%	11%	47%	4%	18%	23%	4%	19%	35%	4%	10%	36%
Human services	30%	35%	45%	42%	32%	39%	35%	43%	50%	36%	30%	38%	24%	33%	42%
International	1%	2%	—	2%	5%	—	—	4%	—	—	2%	—	1%	—	—
Public, societal benefit	14%	11%	7%	10%	14%	9%	27%	7%	14%	13%	7%	16%	19%	13%	9%
Religion	6%	9%	1%	10%	5%	—	2%	7%	—	4%	7%	—	3%	5%	—
Mutual, membership benefit	—	—	—	—	—	—	—	—	—	—	2%	—	—	—	—
Unknown	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
Total %	24	27	35	20	27	28	16	12	14	18	18	12	23	17	11
n	71	66	110	60	65	87	48	28	44	53	43	37	70	40	33

Note. Totals may not add to 100 due to rounding error. NTEE = National Taxonomy of Exempt Entity; Inc. ZIP = Nonprofit organizations with incorrect ZIP code information; Multi. Loc. = Nonprofit organizations operating at multiple service locations.

Table 2.

Distribution of Nonprofits With Locational Limitations, by % Blacks and Hispanics.

	Q1			Q2			Q3			Q4			Q5
	Average percent of Black and Hispanic residents
	11%			19%			29%			41%			62%
NTEE category	P.O. box	Inc. ZIP	Multi. Loc.	P.O. box	Inc. ZIP	Multi. Loc.	P.O. box	Inc. ZIP	Multi. Loc.	P.O. box	Inc. ZIP	Multi. Loc.	P.O. box	Inc. ZIP	Multi. Loc.
Arts, culture, humanities	19%	15%	—	13%	5%	4%	25%	12%	—	20%	26%	4%	17%	10%	3%
Education	18%	17%	5%	11%	17%	6%	8%	14%	12%	8%	15%	4%	20%	10%	7%
Environment, animal	6%	6%	3%	9%	10%	4%	—	5%	—	13%	—	—	5%	6%	—
Health	4%	9%	38%	8%	18%	28%	8%	14%	38%	5%	10%	32%	5%	8%	45%
Human services	26%	30%	50%	41%	32%	42%	25%	33%	41%	31%	38%	46%	41%	38%	39%
International	1%	—	—	2%	5%	—	—	5%	—	1%	3%	—	—	—	—
Public, societal benefit	22%	15%	5%	13%	7%	15%	21%	10%	9%	18%	5%	13%	5%	17%	7%
Religion	3%	6%	—	5%	7%	—	13%	7%	—	4%	3%	2%	7%	10%	—
Mutual, membership benefit	—	2%	—	—	—	—	—	—	—	—	—	—	—	—	—
Unknown	—	—	—	—	—	—	—	—	—	—	—	—	—	—	—
Total %	25	22	13	21	25	25	8	17	11	32	16	18	14	20	33
n	77	53	40	64	60	78	24	42	34	96	39	56	41	48	103

As shown in Table 1, of the nonprofits with a P.O. box listed despite having an actual operating location, the operating location for 24% of these organizations was located in the lowest income areas of the county, Quintile 1. The operating location for 23% of these organizations was located in the most affluent areas of the county, Quintile 5. Of nonprofits with incorrect address information (at the ZIP code level), the correct address for 27% of these organizations was located in Quintile 1. Of the multiple service sites identified, 35% were also located in Quintile 1. Human services nonprofits accounted for the largest share of nonprofits in each quintile and in each category (i.e., with either a P.O. box address listed despite having an operating location in the county, with incorrect address information, or having multiple service sites).

In Table 2, ZIP codes are sorted by percentage of African American and Hispanic residents in the county and divided into quintiles. This information was also obtained from 2008 SANDAG population estimates. The top quintile represents the total number of nonprofits in areas of the county with the highest concentration of African American and Hispanic residents. As shown, of the nonprofits with a P.O. box listed, the operating location for 32% of these organizations was in areas of the county with higher concentrations of African American and Hispanic residents, Quintile 4. Of nonprofits with incorrect address information listed, the correct address for 25% of these organizations was in areas of the county with lower concentrations of African American and Hispanic residents, Quintile 2. Thirty-three percent of multiple service sites identified were in areas of the county with higher concentrations of these residents, Quintile 5. Human services nonprofits, again, accounted for the largest share of nonprofits in each quintile and in each category.

Differences Between Files: Quantity

Tables 3 and 4 show differences in the total number of nonprofits in the Core File before and after adjustments were made for location-related limitations, by NTEE code. In Table 3, ZIP codes are again sorted by average median household income and divided into quintiles. As shown, the number of nonprofits located in the lowest income areas of the county (Quintile 1) increased the most after adjustments by approximately 35%. The number of nonprofits located in the most affluent areas of the county (Quintile 5) increased the least after adjustments by approximately 18%.⁸ The greatest subsector difference between the initial and adjusted files was in Quintile 3. After adjustments, there were 6% fewer educational nonprofit organizations in this quintile.

Table 3.

Differences Between Adjusted and Initial Core File, by Median Household Income.

	Q1		Q2		Q3		Q4		Q5
NTEE category	Initial	Adjusted	Initial	Adjusted	Initial	Adjusted	Initial	Adjusted	Initial	Adjusted
Arts, culture, humanities	17%	16%	8%	8%	12%	10%	9%	9%	9%	10%
Education	14%	12%	17%	14%	33%	27%	28%	26%	33%	31%
Environment, animal	3%	3%	3%	3%	3%	4%	4%	4%	3%	3%
Health	11%	14%	17%	20%	7%	8%	12%	13%	11%	11%
Human services	33%	34%	34%	36%	31%	34%	26%	28%	24%	25%
International	4%	2%	3%	3%	1%	1%	2%	2%	4%	3%
Public, societal benefit	13%	12%	12%	11%	6%	10%	12%	12%	10%	11%
Religion	7%	6%	6%	5%	7%	5%	6%	5%	5%	5%
Total % (n)	22 (538)	23 (724)	25 (624)	25 (770)	13 (316)	13 (409)	18 (446)	18 (540)	22 (545)	21 (641)
Average concordance^a,b	0.85		0.88		0.73		0.92		0.91
Average revenues^c (in US$)	213.15	278.44	578.85	586.85	66.05	134.17	120.18	193.19	210.44	320.71
Median revenues (in US$)	19.48	23.73	20.47	22.01	7.05	8.49	9.45	9.56	10.06	11.08
Average expenditures (in US$)	189.89	249.78	551.43	550.66	58.64	117.31	107.66	172.93	180.53	281.36
Median expenditures (in US$)	18.73	23.02	17.93	19.18	6.28	7.69	8.35	9.25	9.14	9.80

Note. “Mutual, membership benefit” and “Unknown” NTEE categories were excluded; percentage values for these categories equaled “0.” NTEE = National Taxonomy of Exempt Entity.

“Mutual, membership benefit” and “Unknown” NTEE categories were excluded from average concordance calculations due to low values.

Total concordance = .94

Dollars in US$10,000 values.

Table 4.

Differences Between Adjusted and Initial Core File, by % Blacks and Hispanics.

	Q1		Q2		Q3		Q4		Q5
NTEE category	Initial	Adjusted	Initial	Adjusted	Initial	Adjusted	Initial	Adjusted	Initial	Adjusted
Arts, culture, humanities	10%	10%	9%	8%	11%	11%	20%	19%	7%	13%
Education	31%	28%	25%	22%	21%	19%	21%	18%	17%	16%
Environment, animal	3%	4%	4%	5%	3%	3%	2%	3%	2%	3%
Health	13%	13%	11%	14%	16%	16%	16%	11%	12%	15%
Human services	22%	24%	34%	35%	29%	30%	30%	30%	37%	36%
International	3%	2%	3%	3%	2%	2%	2%	2%	3%	1%
Public, societal benefit	12%	13%	9%	10%	12%	13%	13%	11%	11%	12%
Religion	5%	4%	5%	5%	5%	5%	5%	5%	10%	5%
Total % (n)	26 (644)	25 (766)	26 (640)	25 (776)	16 (391)	14 (440)	16 (391)	18 (542)	16 (403)	18 (560)
Average concordance^a,b	0.86		0.85		0.95		0.83		0.82
Average revenues^c (in US$)	210.35	283.71	152.28	267.30	832.43	808.76	112.41	173.42	155.28	256.50
Median revenues (in US$)	10.11	11.40	12.09	12.80	16.97	21.75	12.36	15.01	15.27	19.11
Average expenditures (in US$)	182.83	249.43	131.10	235.60	788.72	759.88	99.88	152.63	153.87	242.14
Median expenditures (in US$)	9.40	10.80	10.37	10.77	14.67	17.92	10.91	12.67	14.34	18.05

Note. “Mutual, membership benefit” and “Unknown” NTEE categories were excluded; percentage values for these categories equaled “0.” NTEE = National Taxonomy of Exempt Entity.

“Mutual, membership benefit” and “Unknown” NTEE categories were excluded from average concordance calculations due to low values.

Total concordance = .91

Dollars in US$10,000 values.

Table 4 shows that when sorting ZIP codes by the percentage of African American and Hispanic residents, the number of nonprofits in areas of the county with the highest concentration of these residents (Quintiles 4 and 5) increased the most after adjustments were made, both by approximately 39%. The number of nonprofits in areas of the county with average concentrations of African Americans and Hispanic residents (Quintile 3) increased the least after adjustments, by approximately 13%. The greatest subsector difference between the initial and adjusted files was in Quintile 5. After adjustments, there were 6% more art and culture nonprofit organizations in this quintile.

Next, a series of paired-samples t tests were also conducted to determine if the mean number of nonprofits (total) in each quintile in the initial Core File significantly differed from the mean number of nonprofits (total) in each quintile after adjustments were made (results of the means tests not shown, available upon request). The mean number of nonprofits listed in the two files significantly differed in all areas of the county with the exception of in Quintile 3 when the analysis was based on median household income (Table 3). In this quintile, the mean number of nonprofits in the adjusted File (M = 41, SD = 48) was not significantly greater than the mean number of nonprofits in the initial File (M = 32, SD = 38), t(9) = −2.15, p = .6. Still, in most areas of the county, after adjustments were made the total quantity of nonprofits in each quintile significantly increased.

Degree of concordance between files

To determine the degree of concordance (or agreement) between quantities of nonprofits in the initial File and those in the adjusted File, concordance correlation coefficients (CCCs) were calculated with associated confidence intervals. CCC is often used to measure agreement between continuous variables, and is a product of both the Pearson correlation coefficient (PCC) and the bias correction factor (BCF). Given that the PCC ignores differences in the means and standard deviations of two sets of measures, the CCC adjusts for this limitation (Barchard, 2012). It is calculated as

ρ_{c} = \frac{2 s_{x y}}{s_{x}^{2} + s_{y}^{2} + {(\bar{x} - \bar{y})}^{2}},

where $\bar{x}$ and $\bar{y}$ are the mean values of two variables, and s_x and s_y are the associated variances (Lin, 1989). Similar to the PCC, a CCC value of 1 indicates perfect positive agreement; a CCC value of −1 indicates perfect negative (inverse) agreement, and a value of 0 indicates no agreement. In this study, CCC provides an indication of the degree to which quantities of nonprofits (per ZIP code) in the two files align, by NTEE code. (CCC values for each NTEE category by quintile are not provided, but available upon request.)

As shown in the middle of Table 3, when ZIP codes are sorted by average median household income, the average degree of concordance between the two files was highest in the most affluent areas of the county (CCC = .92 in Quintile 4, CCC = .91 in Quintile 5). The average degree of concordance between the two files was lowest in Quintile 3—areas of the county with approximately average median household income (CCC = .73). In Table 4, when ZIP codes are sorted by percentage of African American and Hispanic residents, concordance between the two files was lowest in areas of the county with higher concentrations of American and Hispanic residents.

Differences Between Files: Revenues and Expenditures

Despite overall increases in the quantity of nonprofits in each quintile, it should be noted that greater quantity (in absolute terms) does not imply greater capacity to provide services. In fact, as the number of nonprofits in an area increases, the resources available to support those organizations may decrease. Reed, Lally, and Quiett (2006), for instance, highlighted the concept of “battered agencies” where nonprofits located in low-income communities tend to be hindered by the same risk factors facing the families they seek to help, including lack of economic resources and limited financial stability. Therefore, in an attempt to account for differences in fiscal capacity between the two files, the lower portions of Tables 3 and 4 display differences in total revenues and expenditures (often considered important indicators of organizational capacity; Grønbjerg & Paarlberg, 2002; Snyder & Freisthler, 2011).

As shown in the lowest portion of Table 3, the difference between average revenues and average expenditures in the two files was greatest in the most affluent areas of the county (average revenues: US$320.71 − US$210.44 = US$110.27; average expenditures: US$281.36 − US$180.53 = US$100.83 in Quintile 5). The difference between median revenues and median expenditures was greatest in the lowest income areas of the county (median revenues: US$23.73 − US$19.48 = US$4.25; median expenditures: US$23.02 − US$18.73 = US$4.29 in Quintile 1). Undoubtedly, the variation in these aggregate values is influenced by outlier organizations such as hospitals and universities. In fact, health care and education nonprofits—which tend to have the greatest financial resources—were more often located in areas of the county with higher median household income values.

In the lowest portion of Table 4, the difference between average revenues and average expenditures in the two files was greatest in areas of the county with lower concentrations of African American and Hispanic residents (average revenues: US$267.30 − US$152.28 = US$115.02; average expenditures: US$235.60 − US$131.10 = US$104.50 in Quintile 2). The difference between median revenues in the two files was greatest in areas of the county with approximately average concentrations of African American and Hispanic residents (median revenues in Quintile 3: US$21.75 − US$16.97 = US$4.78). The difference between median expenditures in the two files was greatest in areas of the county with lower concentrations of African American and Hispanic residents (median expenditures in Quintile 5: US$18.05 − US$14.34 = US$3.71).

Discussion and Conclusion

Overall, these findings seem to suggest that when location information in the Core Files is verified and adjusted, the prevalence of nonprofits in an area can change significantly. This change may very well affect our understanding of locational dynamics of the sector.

Although studies highlighting geographic variability in the nonprofit sector have led to concerns about the ability of nonprofits to address social ills effectively, the data used to generate much of these findings, the Core Files, have several well-acknowledged limitations, particularly for assessing locational dynamics of the sector. The purpose of this study was to examine the extent to which location-related limitations in the Core Files existed in San Diego County, and to determine if failing to account for these limitations could potentially impact research findings.

The findings reveal that when using the Core Files for locational analyses, failure to account for location-related limitations can result in an undercount of the number of nonprofits. Indeed, many of the nonprofits listed in the 2007 Core File of Public Charities for San Diego County with only a P.O. box address had an actual operating location in the area. Thus, while some researchers have simply excluded nonprofits with P.O. boxes from their analyses, such an approach is likely to exclude many active nonprofit organizations.

The findings also reveal that when location-related limitations are accounted for, the number of nonprofits located in specific communities can increase—and, in San Diego County, the number located in lowest income areas increased the most. It is possible, then, that distributional differences in the quantity of nonprofits between affluent and less affluent areas may not be as severe a problem as previous research suggests (Bielefeld, 2000; Grønbjerg & Paarlberg, 2001; Joassart-Marcelli & Wolch, 2003). In fact, the overall degree of concordance (or agreement) between categories of nonprofits listed in the initial and adjusted Files was less in lower, rather than in higher, income areas—suggesting that nonprofits in lower income areas may not be accurately represented in the Core Files. Researchers should, therefore, be diligent in ensuring that Core data present an accurate representation of the nonprofit sector in these areas.

Not only were lower income areas in San Diego County found to have greater quantities of nonprofits than initially listed, but the nonprofits located in these areas were also found to have greater fiscal capacity. In fact, when location-related limitations were accounted for, nonprofit revenues and expenditures in lower income areas increased considerably. It may be possible, then, that while nonprofits located in lower income areas may not have the same fiscal capacity as those located in more affluent areas (Reed et al., 2006), the magnitude of the discrepancy may be less than reported in previous research.

Finally, the findings provide insights into how locational limitations associated with the Core Files may relate to resident concentration. Indeed, after adjustments were made, when ZIP codes were sorted by percentage of Black and Hispanic residents, there were more nonprofits located in areas of San Diego County with higher concentrations of these residents. These results suggest that traditionally disadvantaged areas (e.g., lower income, higher racial/ethnic minority representation) may have more nonprofit resources than suggested by past research.

Practical Implications

Although the location-related limitations examined in this study can pose a number of challenges for researchers attempting to undertake locational analyses of the nonprofit sector, many of these challenges will differ depending on the scale of analysis. For example, studies relying on location data at the street-level may be more affected by incorrect address information than studies relying on location data at either the ZIP code or metropolitan level. In fact, it is conceivable that nonprofit administrators will more often list their street address incorrectly on tax forms than either their ZIP code or city of operation. Still, the findings from this study should provide strong evidence that when using the Core Files to assess nonprofit location, failure to examine the accuracy of the data at any scale of analysis will limit the ability to capture the full extent of nonprofit activity in an area.

It may, therefore, be beneficial for researchers to begin supplementing Core data with other data sources. Reiner (2003) supplemented the Core Files with municipal files on tax-exempt real property, and found that using both data sources provided a more complete overview of nonprofits in Philadelphia. Grønbjerg (1994) and colleagues supplemented IRS data with other data sources, such as organizational surveys and interviews with nonprofit executives; they created a comprehensive database of nonprofits in Indiana. Supplementing the Core Files with additional data will allow researchers to present a more accurate picture of the nonprofit sector.

Limitations and Directions for Future Research

Although this study has presented important findings, like all research it has some limitations. First, relative differences between initial and adjusted Core Files, in some instances, may be minimal. Indeed, even though there were more nonprofits located in some areas of the county after adjusting for location-related limitations, aggregate revenue and expenditure values did not fluctuate much. Thus, adjusting the Core Files to account for financial differences may be less important than adjusting the Files to account for locational differences. More research is needed to make conclusive claims regarding this possibility.

Second, this study focuses only on Core File data in San Diego County, which likely differs from other locales. As such, the findings from this study may not generalize to all areas. The findings, however, can provide knowledge that may be useful to scholars undertaking research on nonprofit location in other regions. Still, more research should be conducted on the accuracy of the Core Files in other areas.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Lindsey M. McDougle is an assistant professor in the School of Public Affairs and Administration (SPAA) at Rutgers University–Newark. Her research interests relate broadly to the areas of nonprofit leadership and management as well as social inequality and community structure.

References

Barchard

(2012). Examining the reliability of interval level data using root mean square differences and concordance correlation coefficients. Psychological Methods, 17, 294-308.

Bielefeld

(2000). Metropolitan nonprofit sectors: Findings from NCCS data. Nonprofit and Voluntary Sector Quarterly, 29, 298-314.

Bielefeld

Murdoch

Waddell

(1997). The influence of demographics and distance on nonprofit location. Nonprofit and Voluntary Sector Quarterly, 26, 207-225.

Corbin

(1999). A study of factors influencing the growth of nonprofits in social services. Nonprofit and Voluntary Sector Quarterly, 28, 296-313.

Eikenberry

(2005, November). Creating social equity: What role for nonprofits? Paper prepared for the National Academy of Public Administration Standing panel on social equity in governance. Fifth social equity leadership conference, University of Nebraska-Omaha, Omaha, NE.

Froelich

Knoepfle

(1996). IRS 990 Data: Fact or fiction? Nonprofit and Voluntary Sector Quarterly, 25, 40-52.

Grønbjerg

(1994). Using NTEE to classify non-profit organisations: An assessment of human service and regional applications. Voluntas, 5, 301-328.

Grønbjerg

Paarlberg

(2001). Community variations in the size and scope of the nonprofit sector. Nonprofit and Voluntary Sector Quarterly, 30, 684-706.

Grønbjerg

Paarlberg

(2002). Extent and nature of overlap between listings of IRS tax-exempt registration and nonprofit incorporation: The case of Indiana. Nonprofit and Voluntary Sector Quarterly, 31, 565-594.

10.

Joassart-Marcelli

Wolch

(2003). The intrametropolitan geography of poverty and the nonprofit sector in southern California. Nonprofit and Voluntary Sector Quarterly, 32, 70-96.

11.

Lampkin

Boris

(2002). Nonprofit organization data: What we have and what we need. American Behavioral Scientist, 45, 1675-1715.

12.

Lin

(1989). A concordance correlation coefficient to evaluate reproducibility. Biometrics, 45, 255-268.

13.

Mohan

Twig

Jones

Barnard

(2006). Volunteering, geography, and welfare: A multilevel investigation of geographical variations in voluntary action. In Milligan

Conradson

(Eds.), Landscapes of voluntarism: New spaces of health, welfare and governance (pp. 267-284). University of Bristol, UK: Policy Press.

14.

National Center for Charitable Statistics. (2006). Guide to use NCCS data. Washington, DC: Urban Institute.

15.

Peck

(2008). Do anti-poverty nonprofits locate where people need them? Evidence from a spatial analysis of Phoenix. Nonprofit and Voluntary Sector Quarterly, 37, 138-151.

16.

Reed

Lally

Quiett

(2006). Battered agencies: Supporting those who serve low-income communities. The Nonprofit Quarterly. Retrieved from http://www.nonprofitquarterly.org/management/619-battered-agencies.html

17.

Reiner

(2003). Tally of nonprofits. Administration & Society, 35, 629-637.

18.

Schumann

M. J.

Jones

J. A.

Hunt

Mitchell

Krasynska

Deitrick

(2014). 2014 annual report: State of nonprofits in San Diego. Caster Family Center for Nonprofit and Philanthropic Research, University of San Diego. Retrieved from http://catcher.sandiego.edu/items/soles/2014_USD_StateNonprofits_Report_vFinal5_Sept.pdf

19.

Small

McDermott

(2006). The presence of organizational resources in poor urban neighborhoods: An analysis of average and contextual effects. Social Forces, 84, 1697-1724.

20.

Snyder

Freisthler

(2011). Are nonprofit density and nonprofit expenditures related to youth deviance? Children and Youth Services Review, 33, 51-58.

21.

Stater

(2009). How permeable is the nonprofit sector? Linking resources, demand, and government provision to the distribution of organizations across nonprofit mission-based fields. Nonprofit and Voluntary Sector Quarterly, 39, 674-695.

22.

Twombly

De Vita

Garrick

(2000). Mapping nonprofits in Philadelphia, Pennsylvania. Retrieved from http://www.urban.org/publications/410015.html

23.

U.S. Census Bureau. (2013). Annual estimates of the resident population: April 1, 2010 to July 1, 2013: 2013 Population Estimates. U.S. Census of the Population. Retrieved from http://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?src=bkmk