Why Are Summary Statistics Sometimes Inconsistent?

From NCCS Wiki

NCCS uses data from several sources, all imperfect. (There's an old joke about politics that applies -- at least a little bit -- to statistics, too: What do politics and sausage-making have in common? You wouldn't want to know what goes into either.)

All of our databases represent either the full population of organizations or charitable activity or a statistically-significant sample of them. Both population and sample files have their limitations:

• Population Databases
  • NCCS Core
  • Business Master File
  • Master List
  • Guidestar-NCCS "Digitized" Database
  • IRS 1040 Zipcode Database
• Sample Databases
  • IRS Statistics of Income Division (SOI) Exempt Organization and Private Foundation files
  • Overhead Cost Study sample database

[edit] Which Source Should You Use?

We wish we could provide users with a simple answer, but the answer often depends on the question that you are asking:

• How many organizations are REGISTERED with the IRS? The Business Master File is the best source.
• How many FILE IRS Form 990s in a given year?
  • If the question is to determine the total value of assets, revenues or expenses (or related aggregate numbers) for the sector, we generally recommend using the Core file. However, it will include returns that may be two years old, since some organizations are late in filing or for other reasons.
  • The SOI files will also work, but they are usually a year older than the Core files since it takes the extra year to do the more sophisticated error checking.
  • If you need to get an answer for a specific field (other than hospitals or higher education), state, county or city, then SOI is seldom an option, since the sample size is too small.

[edit] Limitations

• Late or missing Form 990 filers Some organizations fail to file their IRS Form 990s on time. Occasional administrative processing issues may also explain why some organizations' returns are missing. (For example, in a very small number of cases, the federal Employer ID Number (EIN) was entered incorrectly by either the organization itself or by the IRS. Thus, a return appears to be "lost."
  Depending on one's research objective, one may want to ignore these late/missing filers or use their form from a prior year. For the Core files, NCCS has opted to include prior returns to ensure that all nonprofit activities are adequately accounted for. For the "Digitized Database," however, we provide the option to analyze the data using either a strict fiscal year basis or use the "circa year" perspective employed in the Core files.
• Registered but "dead" organizations Organizations are supposed to notify the IRS when they cease operating. However, most appear not to do so. The new IRS Form 990-N "e-Postcard" required of almost all registered nonprofits each year should remedy this problem by 2011. In the meantime, keep in mind that an earlier IRS study found that 21% of a small sample of nonfiling organizations in the IRS Business Master File could not be located and were presumably no longer in operation.
• Unregistered organizations Some nonprofit organizations simply do not register with the IRS either because there's no legal requirement -- if an organization has gross receipts of less than $5,000 or is a religious congregation -- or if there's no necessity -- the organization doesn't expect to receive donations and has no taxable income. Many homeowners and community associations, for example, can be found in state nonprofit corporate records but do not register with the IRS.

[edit] Other Differences

• "Failed public charities" are Private Foundations. Most 501(c)(3) organizations prefer to obtain Public Charity rather than Private Foundation status from the IRS, since the latter requires payment of an annual excise tax, completion of a more comprehensive IRS annual return (the Form 990-PF), and deductibility to private foundations is more limited. (See http://www.irs.ustreas.gov/instructions/i1023/ar01.html.) Some organizations initially obtain Public Charity status but are unsuccessful at soliciting sufficient "public support" (donations from the general public and government grants) to maintain their public charity status. They are then redesignated as Private Foundations by the IRS. Many appear to become inactive. Thus, of the 110,000 private foundations registered with the IRS in early 2007, only 88,000 actually filed Form 990-PF in the previous two years.
• Different definitions of "Year." The Core files are based on a "Core" or "circa" year: We include the latest return for organizations we expect to be alive (as defined by their inclusion in the IRS Business Master File) so long as the return was filed within two years of the "Core" year. Thus, the 2006 Core file includes fiscal years ranging from 2004 to 2007. The NCCS Data Users Guide describes the metholodogy in greater detail.
• Weaknesses in reporting on the Form 990. Medicaid and Medicare have been substantially under-reported on the Form 990s.
• Form 990 vs. GAAP and Other Sources for Financial Data. Definitions of "government grants" and "program service revenue" on the Form 990 are simply different than comparable terms used in most audited financial statements. There are sound regulatory reasons for the differences, but this complicates analysis. In one area, Form 990 data is simply wrong. The NCCS Almanac (chapter 4 and the 2008 Almanac Brief) combines Form 990 data with data from the Centers for Medicare & Medicaid Services since many nonprofits, especially hospitals, fail to correctly report Medicare and Medicaid revenues in Part VII of the form.