1. The MCDC profiles allow the user to select up to 4 geographic areas and will display the data for those areas side by side to facilitate comparison.

2. Statistical reliability of the data are displayed here using a 3-tiered font weighting scheme with MOE-based values (relative MOE's and confidence intervals for counts; MOE values for Percents) accessible by holding the mouse over any data item on the report for 1 second. (See details, below).

3. These profiles will always provide both a count and a percentage measure whenever possible. So (for example) while the Bureau profile shows only poverty rates, our profile displays both counts and percentages of poor persons (or families, etc.)   Note: Since this note was written the Census Bureau has modified their profiles so that they almost always now include both counts and percentages (except for the data on poverty).

4. You can choose to view multiple profiles at once here. You can only display one Bureau profile at a time (i.e. Social, Economic, etc.)

5. The data used in these reports are stored in datasets that can be accessed and downloaded (using our Dexter web extraction module). You can easily access these data for every geographic area in the country in a single query. Profiles on the Bureau web site only provide access to the data one geographic area at a time.

6. These tables provide links to dynamically generated color charts for visualizing the data in a table.

7. "Drill down" links are provided to allow viewing of detailed "parent" tables which contain (usually) more detailed data that what are shown here

8. On the other hand, the Bureau provides a "Narrative Profile" option where you can see highlights of the profile data in a non-tabular fashion, interspersed with a data charts. This option is not available with the MCDC application.

General Layout of Tables

The reports are broken into a Subject column, one to four Number-Percent column groups with the name of the area being summarized displayed in the over-arching first header row. The order of the geographic areas left to right across the page is determined by the order in which you select them (or in which they were selected for you if you came in directly from a web link). A final column at the far right of the report labeled "Base Table(s)". In some rows you will see the name of a base table used as a hyperlink. The table referenced contains more detailed data related to the current data item (and in some cases is the source of that data item). A click on one of these base table hyperlinks invokes an application (called acstabgen) that will display that table for each of the geographic areas currently selected. This amounts then to a kind of demographic "drill down" for more detailed information.

The lines of the table consist of Table Header lines identifying the subject matter and "universe" (i.e. the kind of entities being counted or measured). These are highlighted with a gray background. The first table header line is a hyperlink to the metadata where you can get more detailed information about the variables in that table. The detail lines of the table are divided into columns. The Subject column identifies the data displayed on the line. Indentation is used to indicate subordinate data. For example, in Table 24, "Owner Occupied Units" is indented within "Occupied Housing Units". Typically, the first detail line in a table contains the total universe count and is displayed in bold. All items that count or measure the entire table universe are left justified and bolded.

The Census Bureau, along with most statisticians, prefer to use the Coefficient of Variation (or "CV") statistic as the relative measure of statistical reliability. The CV is defined as the SE (standard error) as a percentage of the estimate. The SE can be easily derived from the 90% level MOE using the formula:
SE = MOE / 1.645
So the difference between the Relative Margin of Error measure that we use and the CV is simply:
CV = (MOE/1.645)/E = RMOE/1.645 (where E is the estimate and RMOE is the relative margin of error)
So the intervals we use to determine our 3-tiered font scheme can be expressed in CV terms as:

• Bold font when CV is < 9.1 (15/1.645)
• Plain font when CV is between 9.1 and 21.2 (35/1.645)
• Light gray when CV exceeds 21.2

Charts

Indentation in the Subject Column

Downloading the Data

Using the Front End Menu

http://mcdc1.missouri.edu/cgi-bin/acsprofiles/acsprofiles.py?period=2005_2009&geoids=16000US2915670&geoids=05000US29019&geoids=04000US29&geoids=01000US&subjects=Demographic&subjects=Economic&subjects=Social&subjects=Housing.

Gobbledy-gook? Perhaps, but not really as bad as it might first appear. The first part
http://mcdc1.missouri.edu/cgi-bin/acsprofiles/acsprofiles.py
just tells the browser where to find the program ("app" for those of your under30) to be invoked.

This is followed by the "parameter specifications" portion. Each of these "parm-specs" is of the form
&<parmname>=<parm-value> where <parmname> means substitute the name of the parameter, such as period or geoids and <parm-value> means substitute some value to be associated with that parameter. So the "parm-spec"
&period=2005_2009
is just the secret code needed to tell the application that we want the 2005 to 2009 5-year period data. If the value had been 2009 instead of 2005_2009 then the program would have understood that what was wanted was single year data from 2009. There are 4 parm specs using the same parameter name geoids. That is how we tell the application that we have 4 geographic areas that we are wanting to see data for. The values of these geoid parameters correspond to the values of the geographic identifier values stored with the ACS data. The geoid value specifying that we want Columbia, MO is 16000US2915670. Breaking this down: "160" is a summary level code that indicates a Place (city) summary; "00" means that it is not a geographic component summary, while "US" just says it is a United State geogrphic area. "29" is the FIPS state code for Missouri and "15670" is the FIPS/ANSI code for the city of Columbia. It is not all that hard to see the pattern for these IDs.

The point of all this is that these URLs are the kind of thing that web developers can take advantage of to create their own custom menu pages that can lead users to custom profile reports without any need to go through making a series of choices off a menu. It might take 15 or 20 seconds to choose the 4 geographic areas used on our sample report. But someone who wanted to provide comparable reports for all cities of a certain size in a certain state could pretty easily figure out how to generate the URLs to do so and put those on a page with city names as the hyperlinks.
To be continued....