Nonprofit Program Classification (NPC)
The National Center for Charitable Statistics (NCCS) has developed the Nonprofit Program Classification (NPC) system for classifying program services and beneficiaries for the NCCS/GuideStar digitized database of IRS Forms 990. The system should serve all potential users of the NCCS/Guidestar database, ranging from researchers to donors.
Download a complete listing of NPC codes(PDF)
Currently, codes from the National Taxonomy of Exempt Entities (NTEE) system are used to classify nonprofit organizations according to their organizational purpose. Although this is essential to research, an organizational category does not adequately describe the activities and the beneficiaries of these activities for many organizations. Often, the questions that are asked by researchers, nonprofit organizations, and the public are focused on what the organizations actually do and who is served, rather than the types of organization. Organizations may have several types of activities that cannot be adequately described using one organizational classification.
The Nonprofit Program Classification System will classify the actual activities of each organization, as reported on Part III of its Form 990. For the first time, such a system is feasible because of the creation of the NCCS/GuideStar digitized database, with Part III information that can be electronically searched and categorized. The task of implementing a system that involves reviewing the Part III text of each organization=s Form 990 each year (as major activities and beneficiaries may change from year to year) is daunting. Implementation plans call for the use of artificial intelligence to the maximum extent possible, the creation of a Web-based interactive program for self coding by organizations, and the merging of NCCS data and information from Information and Referral Systems to maximize data collection efforts. As this information becomes more widely available, the nonprofits themselves should have a greater interest in accurate coding, as this is a way that their organizations will be introduced to potential donors.
The process of creating the Nonprofit Program Classification (NPC) system includes the following components.
Creating subject hierarchies. These are lists of standardized terms, displayed in hierarchical arrays, using the same major types as NTEE. Both the NTEE and actual Form 990 program services descriptions serve as guides to the appropriate terms, supplemented by various sub-sector thesauri in the different disciplines. Some of the thesauri examined include the Educational Resources Information Clearinghouse (ERIC) Thesaurus, the Justice Department?s Criminal Justice Thesaurus, the Art and Architecture Thesaurus, the National Association of State Arts Agencies? National Standard and INFOLINE?s Taxonomy of Human Services. The terms in the hierarchies group like concepts (the term "senior housing" groups organizations that provide senior housing, housing for the elderly, and assisted living facilities). Care must be taken to ensure that activities described with terms that fit but with different definitions (for example, organizations that renovate older housing must be excluded from the term "senior housing") are appropriately placed.
Defining terms, adding broader and narrower terms, and developing extensive cross-references. As users will access the classification system by searching for terms that they enter, a way to take random terms and convert them to terms used in the hierarchies must be devised. Extensive cross-references will complete this task. In addition, lists of broader and narrower terms will allow users to choose terms from "drill down menus" in a logical matter. As each hierarchy is completed, the lists of broader and narrower terms and cross-references are also completed.
Classifying the activities of nonprofits. Based on the IRS Form 990 Part III program service descriptions in the NCCS/GuideStar database, key words are used to place the activities in the hierarchy. Different key words are given rating weights, based on testing how well the machine coding matches manual coding of the same activities. The results of the machine matching are used to identify missing key words and modify rating weights to increase the match rate. This is an iterative process with each revision of key words and weighting leading to a higher match rate. Two hierarchies now have match rates of 85 percent or better, and four others are being fine tuned. The test files of Part III program service descriptions for the other hierarchies have been manually classified and keywords are being compiled.
These data will be an immensely useful tool for researchers, practitioners and policy makers who are concerned with the role of the nonprofit sector at national, regional, and local levels and the various service sub-sectors.