[DDI-ADG] Geography and time comments

[Wendy Thomas]

>
> >Geography
> >1) Thinking through and documenting the issue we talked about in line 4
> >(where Description could say that the study covers 3 countries, each of
> >which has its own Geography > GeographyValues > GeographyName), I realized
> >that line 5 needed to be rewritten b/it implies that Geographic Coverage
> >will be described by only one geography.  So I broadened it.  I assume the
> >"which can also be distinguished by a Geography which does not have a
> >ParentGeography" would apply to each of the 3 coverage-related instances
> >of Geography in the 3-country scenario.  This relates to line 13.
> >
> >
> >WLT
> >No. There is only one top level geography. If I have 3 countries that 3
> >country group is my top level. If I combine the three individual countries
> >(level 2) I get the top level geography. The study itself defines what the
> >top level geography is in the same way it defines what the population
> >universe is. There is no .predefined. name or code for this level just as
> >there is not a predefined code or name if my .geography. is an area
> >comprised of a 5 mile circle around a specific point. Don.t change this.
> >It is a concept which ties into FGDC and other geographic descriptive
> >systems. Geographic Cover means the full extent of the geography covered
> >regardless of whether or not it is an contiguous area.
>
> I see what you're saying (that in the given case the top level is a group
> of countries, not a country), but am not sure how it would change the
> spreadsheet. Are you proposing:
> - that I change line 6 back to omit the line "The first occurrences of this
> should be Geography (may be multiple occurrences, as in 3-country example
> described above) that describes the overall coverage area, which can also
> be distinguished by a Geography which does not have a ParentGeography" or
> to simply change the plural to singular?
> - Are you suggesting that in the case of the 3-country example that we
> would not have 3 Geographies?

I may have misunderstood your statement. There should always be at least
one geography and this should be the full extent or top level or
"Geographic Cover" as defined by FGDC. Addiitonal geographic detail would
be at the discression of the creator, but obviously specific
identification of each level is preferred. In the case of the 3-country
example you would have 2 geographies: The 3-country group and county which
would contain 3 GeographyValues (country 1, country 2 and country3).

>
> Also, in your notes, on line 15 you say:
> "Parent should be identified in terms of geographic hierarchies (for
> example a County parent is the State rather than the State and sometimes
> the MSA).  Any geography with 2 or more parents is defined as being the
> intersect of those multiple hierarchies"
> Is this meant to supplement or replace the given description?  It sounds
> like the latter, but as what sounds like a significant change, what do
> others think?; I put it in the attached as a choice and bolded it so we
> knew we had to address it.
>
> Others can maybe chime in on these as well; and/or let's talk about this
> tomorrow.

It is a suppliment. The reason for this is that if it is limited to the
strict hierarchy of identification, it provides clear instructions for
code concatonations required for identifying a specific location. For
example:

Census SUMLEV 040 STATE with parent COUNTY US which is the geographic
cover. Therefore you need the level code 040 and the state code 27 to
identify Minnesota. You don't need the Region or Division code. In
actuallity these are composed areas as are Metropolitan Areas. Some levels
of geography are the intersect of 2 or more clean hierarchies. Example:
080 STATE-COUNTY-COUNTYSUBDIVISION-PLACE/REMAINDER-TRACT is the portion of
the tract within a place, within a county subdivision, within a county,
within a state. 3 clean hierarchies:  STATE-COUNTY-COUNTYSUBDIVISION,
STATE-PLACE, and STATE-COUNTY-TRACT. By identifying the code location of
the county TRACT, and the 3 parents COUNTY, COUNTYSUBDIVISION, PLACE you
are able to construct the unique string of codes needed to identify a
specific polygon.

>
>
>
> >2) Line 18: Hate to bring up a substantive issue at this late date (some
> >are just occurring to me when really reading through the descriptions),
> >but I'm not clear on what's required w/in this group.  Is it true that
> >they could choose among lines 19-25 (have to put at least one, but not
> >all)?  If this is the case, then we'd lose the machine-actionability of
> >the values (b/the author could choose just to put in a GeographyName and
> >not a GeographyCode).
> >I think that I recall that we wanted authors to always insert a code for a
> >geography, but we would (in line 20) give them the option of pointing to
> >an external code list (rather than having to document each code using line
> >24).
> >Therefore I would think that at minimum lines 21 and 23 should be
> >required.
> >Does this make sense?  Is this what's needed to ensure
> >machine-actionability?
> >
> >WLT
> >Not all areas have geographic names or codes (see example of 5 mile circle
> >above). It is correct as it is. You have the option of using as many or as
> >few as you wish. Our job is to make machine-actionability possible. We
> >cannot enforce the amount of information that is available.especially in
> >this case where you would be forcing people to make something up.
>
> This was basically what we agreed in the Friday phone call.
>
> --
> Also, in the note to line 22 Wendy wrote:
> "Should exist outside of the repetionShould point to the location of the
> lowest level code variable (such as BLOCK): At this level it can only refer
> to a specific geographic level."  It sounds like a proposal to change the
> structure, so I'm including it as another option/choice, but I'm not sure
> of the best choice.
>

Ok...this was originally designated as a reference to a variable in the
study that describes the geography (in census SUMLEV). However this
applies to the overall geography and should be moved up and listed the
single time. A variable reference to the location of the code that
identifies the lowest level of code should be used, but it applies to the
geographic LEVEL not the geographyValue. For example:

GeographyLevel STATE has the GeographicLevelCode 040 found in the variable
that describes the geography (SUMLEV). The variable that holds the
specific STATE codes is the same variable for all STATE level polygons, in
this case STATEFP. The GeographyValue GeographyCode:Code is the code of a
specific location such as 27 for Minnesota.

Hope this clarifies what I was trying to hastily pull together Sunday
night.

wendy


> Kate
>
> >Wendy L. Thomas                          Phone: +1 612.624.4389
> >Data Access Core Director               Fax:   +1 612.626.8375
> >Minnesota Population Center              Email: wlt at pop.umn.edu
> >University of Minnesota
> >50 Willey Hall
> >225 19th Avenue South
> >Minneapolis, MN 55455
> >
> >_______________________________________________
> >DDI-ADG mailing list
> >DDI-ADG at icpsr.umich.edu
> >http://www.icpsr.umich.edu/mailman/listinfo/ddi-adg
>
> ___________________________________________
> Katherine McNeill-Harman
> Data Services Librarian
> Dewey Library for Management and Social Sciences
> Massachusetts Institute of Technology
> 77 Massachusetts Avenue, E53-100
> Cambridge, MA 02139
> mcneillh at mit.edu
> 617-253-0787

Wendy L. Thomas                          Phone: +1 612.624.4389
Data Access Core Director		 Fax:   +1 612.626.8375
Minnesota Population Center              Email: wlt at pop.umn.edu
University of Minnesota
50 Willey Hall
225 19th Avenue South
Minneapolis, MN 55455