Codes for Religious Affiliations

Original author and date: Glenn Pearson, May 15, 2009

The current pull-down control in the MPR module supports a single-level representation for Religion. Default 0.6.x install has, in the field_options table:

We consider here how to be a bit more specific and better reflect the US context. There is also mention about what alternatives might be if multi-levels were used. (Entry using one or more free-text fields is another alternative, often used in the U.S. Census and in religious surveys.)

Every decade, the US Census collects household information. This can include religious affiliation, although response to that is optional by law. Summary information from the 2000 Census (see particularly www.census.gov/prod/2005pubs/06statab.pop.pdf) is now a bit dated.

In 2007, a survey was undertaken by Pew Forum. This was reported in terms of broad categories as http://religions.pewforum.org/affiliations, and also more finely subdivided (down to 3 or 4 levels) as http://religions.pewforum.org/pdf/affiliations-all-traditions.pdf . Looking at just the broad findings (with +/– 0.6% margin of error, based on 35,556 respondents):

The finer subdividing had 146 leaf nodes (plus additional intermediate nodes). This is rather too much for a single-level pulldown.

No encoding for the religions is publicized.

In a healthcare context, the HL7 coding for religions is pertinent (available for v 2.5 from http://scm.chip.org/svn/repos/ping/trunk/model/codes/src/main/resources/hl7-religion-2.5.xml). It constitutes 83 name/code pairs, where a code is 3 alphabetic characters. This is a bit much for a single-level pulldown. Most terms are for specific religious groups, but some are broad. It is possible to infer a 2-level hierarchy from the naming convention, e.g., “Christian: United Methodist”. This is a sparser treatment generally than Pew’s.

Human and machine-readable documentation about an extensive coding for Australian religious affiliation is defined at www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/1266.02005?OpenDocument This has also been adapted by the New Zealand census (from www.stats.govt.nz/default.htm, search for “Religious Affiliation – New Zealand Standard Classification 1999”. Or ask for ”RELIGAFF V1.0” in the hierarchy browser at www.stats.govt.nz/statistical-methods/classification-code-finder/default.htm)

“In the classification, religions are grouped into progressively broader categories on the basis of similarity in terms of religious beliefs, religious practices and the cultural heritage of adherents. This results in those religions and religious groups, which are closely related in terms of their intrinsic characteristics, being closely aligned in the structure of the classification. The number of adherents of a particular religious group has been a significant factor in developing the classification structure. Thus, Christian denominations are extensively identified [including some specific to New Zealand like Maori].… The classification also includes a ‘No Religion’ category…. In general, level one consists of world religions, level two is denominations, and level three is sub-denominations. However, when these classification criteria are not appropriate (for example there are no ‘denominations’ for Hinduism), religious affiliations are broadly grouped at level one and repeated at level two. The most detailed responses are classified at level three.

Religious affiliation is a hierarchical classification with 190 categories. Level one has a one-digit code structure and has 10 categories. Level two has a three-digit code structure and contains 50 categories. Level three has a five-digit code structure and contains 130 categories.”

At this time, we will adapt a system compatible with a single-level control, relying primarily on the Pew categorization. Since Pew provides no encoding, we could map either of the foregoing systems. While the Australian/New Zealand encoding is the more flexible, the popularity of HL7 for information exchange in the US argues in its favor, even if mapping is more awkward in spots. We invent 3 new codes in italics for the three major branches of Protestantism delineated by Pew (the 3-letter values chosen are not used elsewhere in HL7 2.5). The row order shown is the desired display order.

[1] Closest match with HL7 is EVC/Christian: Evangelical Church, but this may be referring to a specific church, possibly the one that Pew calls the “Evangelical Lutheran Church” within the Mainline Protestant tradition. Another possibility would be PEN/Christian: Pentecostal, but Pew considers that a subset of Evangelical.

[2] Closest match with HL7 could be PRO/Christian: Protestant, but the latter is overbroad.

[3] HL7 has no sense of “historically black”. One could code based on picking the largest denomination in this category. The first largest, National Baptist Convention, is not coded at all in HL7. (There is the overbroad category BAP/Christian: Baptist, as well as SOU/Christian: Southern Baptist). Second largest is COI/Christian: Church of God in Christ. Third largest is AMT/Christian: African Methodist Episcopal. Also: AME/Christian: African Methodist Episcopal Zion.

[4] Pew reports Latter-day Saints as a subset of Mormon.

[5] An alternative might be EOT/Christian: Eastern Orthodox.

[6] We are collapsing Pew categories “Other World Religions” and “Other Faiths” into the latter. There is no HL7 code that would separately denote “Other World Religions”.

[7] Other HL7 codes are ATH/Atheist, and AGN/Agnostic, but NOE/Nonreligious seems the best match to Pew’s Unaffiliated.