Monday, January 25, 2010

Plugins Needed in Ubuntu 9.10

I was trying to play a .wav file in 64-bit Ubuntu 9.10 (Karmic Koala).  (This was a compressed .wav in 4-bit 22 kHz IMA ADPCM format.)  I double-clicked on it in Nautilus.  Movie Player opened up and gave me this message:

Search for suitable plugin?

The required software to play this file is not installed.  You need to install suitable plugins to play media files.  Do you want to search for a plugin that supports the selected file?

The search will also include software which is not officially supported.

I went with that.  Unfortunately, the next message said this:

No packages with the requested plugins found.

The requested plugins are:

image/ decoder

I clicked OK.  That gave me another message:

An error occurred

The playback of this movie requires a image/ decoder plugin which is not installed.

A search for relevant terms led to a thread in which someone asked whether the user had "the w32codecs" installed.  I didn't see that package in Synaptic.  But then someone else in that thread said maybe I wouldn't need it for my 64-bit Ubuntu.  One person pointed toward an extended tutorial in setting up multimedia in Ubuntu.  There was some discussion on whether 32-bit codecs (e.g., w32codecs) were necessary in a 64-bit system; the consensus (supported, of course, by the actual error message on my system) was that they might well be.  The same opinion emerged in another discussion.  The way to get those 32-bit codecs seemed to be, first, to add the Medibuntu repository to my Ubuntu installation.  The simple way to do this was to cut and paste this command into Terminal:
sudo wget --output-document=/etc/apt/sources.list.d/medibuntu.list$(lsb_release -cs).list && sudo apt-get --quiet update && sudo apt-get --yes --quiet --allow-unauthenticated install medibuntu-keyring && sudo apt-get --quiet update

all on one line.  For additional multimedia options and capabilities and such, it was also recommended that I enter these commands:

sudo apt-get --yes install app-install-data-medibuntu apport-hooks-medibuntu

sudo apt-get install libdvdcss2
sudo apt-get install w64codecs

So I did that.  This all went smoothly.  I was now able to play other .wav files, but I was not able to play that particular one.  I tried playing it in Windows, using IrfanView, and got these error messages:

[filename]:  Can't read file header !

Unknown file format or file not found !

IrfanView: i_view32.exe - Corrupt File

The file or directory [filename] is corrupt and unreadable.  Please run the Chkdsk utility.

So possibly that was why Ubuntu had been unable to play it.  I had checked its properties in Ubuntu, but had not seen any such message.

Monday, January 18, 2010

Configuring 64-Bit Ubuntu 9.10 (Karmic Koala)

In a previous post, I described the steps I followed in setting up x64 Ubuntu 9.04 (as refined in a later post).  This post does the same for Ubuntu 9.10.  I won't re-describe steps that are already spelled out in some detail in that previous post and in the other webpages to which it refers.  I'll still provide most of the details here, just not as much in-depth explanation.

Deciding on a Clean Install.  I was installing on a machine where I had previously been running 9.04.  I installed 9.04 on top of 8.10 on that machine, and this seemed to lead to some unusual problems -- you know, the kind of issue that brings up almost nothing in a Google search.  The better approach was apparently to do a clean install.  So now it was time to do that, replacing 9.04 with 9.10.

My first question was, what should I preserve from my previous setup, and how can I preserve it?  One suggestion was to try this:

tar cf /media/[backup drive]/[backup folder]/[backup filename].tar /etc /home
That didn't work for me, as my /home directory contained 32GB.  Another post in that same thread suggested installing your stuff to a separate /home partition, so that your settings wouldn't be wiped out by future clean installations.  I found a guide to moving the /home partition.  I started to use GParted to make a 50GB /home partition.  I chose ext3 because ext4 still seemed to have some issues.  I discovered that Ubuntu's manual installer has an option not to format the partition you're installing to, so as to preserve your old settings.  I decided against both of these approaches because I did want to wipe out my old /home partition, with its mistakes and assorted junk.  So I installed Ubuntu as described in the previous post, working from a CD.  I went into System > Administration > Update Manager > Check.  This gave me 184 updates and a reboot.

Software Sources.  Next, I went into Applications > Ubuntu Software Center > Get Free Software > search for "restricted extras" > select Ubuntu restricted extras > Install.  Next, System > Administration > Hardware Drivers > NVIDIA accelerated graphics driver (version 185) (because I had an NVIDIA video card) > Activate > Close.  Next, System > Administration > Software Sources > Ubuntu Software tab > Download From > Other > Select Best Server > Choose Server (whichever one it highlights) > Close.  Also, in Software Sources, go to the Other Software tab, select the two entries that are already there, and click Add to add more.  The additional lines come from the X-Updates website.  There, click on "Technical details about this PPA," specify Karmic, and copy the two deb lines there, one at a time, into the APT line, clicking "Add Source" after each.  Next, on that same webpage, follow the Signing Key instructions, just below the deb lines.  Click Close.  This will bring up "The information about available software is out-of-date."  Click Reload.  If you don't get a "Reload" option when you click Close, go back into Third-Party Software and unclick and then re-click some item and try again.

Synaptic.  I had read an upgrade tip that said I could automatically reinstall my installed Synaptic applications by using System > Administration Synaptic Package Manager > File > Save Markings As > Save full state and then, after the upgrade, use Synaptic > File > Read Markings.  I had saved as advised; now I tried the read (restore) step.  It did not work well for me.  After this and a few more misadventures, I wound up reinstalling Ubuntu from scratch.  For posterity, the next few paragraphs describe the failed approach.  After that, I describe the alternate approach, the one that I had used in previous installations. 

To try the Read Markings approach, I started by going to Places > Computer and double-clicking on the drive where I had saved the backup.  Then Apply.  But that gave me an error:  "Could not apply changes!  Fix broken packages first."  So I went into Synaptic > Edit > Fix broken packages.  That, in turn, generated this message:  "An error occurred.  The following details are provided:  E: Error, pkgProblemResolver::Resolve generated breaks, this may be caused by held packages."  Adapting some older advice, I closed Synaptic and then typed this:

sudo -i
apt-get clean
apt-get autoclean
apt-get update
apt-get upgrade
apt-get dist-upgrade

Now I tried Synaptic again.  The steps just taken did not seem to change anything; but after tinkering with the same options, Edit > Fix broken packages seemed to work, and Apply proceeded to download 259 files.  Since the more automated Markings approach seemed to work, the advice for the next installation would be to uninstall any unwanted packages before running the Save Markings step.

Instead of all that, on the second installation I selected and installed these packages from within Synaptic:  boinc; boinc-manager; fdutils (if you expect to be using a floppy drive); gparted; ntfs-config; ntfsprogs; p7zip-full; sysinfo; and webhttrack. (If numerous items come up in response to your search, click on the Package heading to sort them alphabetically. Also install other related packages, if given the option.) (If some of my later descriptions don't work for you, it may be because you didn't install one of these.)  For my e-mail, I preferred thunderbird, so I added that, and uninstalled evolution.  Some of these may ask if you want to "Mark additional required changes?" Click "Mark" and go on to the next one. Then click Apply. Some of these programs may already be shown as being installed on your system. If so, no problem.

Other Programs.  To install Google Earth, I didn't use a previous download (see above). Instead, I typed these two lines:  First, "wget," and then "sh GoogleEarthLinux.bin" (as always, without quotation marks).

I didn't have to install Firefox -- it came installed with Karmic -- but, as in the past, I hoped to speed up my customization by using the FEBE add-in.  I had made a FEBE backup before this reinstallation.  Now, to restore my previous settings, I installed FEBE.  Unfortunately, Firefox did not seem to be functioning properly. I address this set of problems in a separate post.  (I think the problem may have been that I tried to use a FEBE backup of a Windows XP Firefox installation.  Firefox on Ubuntu does not use the same add-ons.)  To have a browser, I downloaded 64-bit Opera.  (At this point, 64-bit Google Chrome was apparently unstable.)  Opera came as a .deb file, so I just double-clicked to install it.  On the reinstall, I didn't even bother with FEBE at this point; for the time being, I just used Opera in its basic form.

VMware Workstation 7 came as a .bundle file, which required the same installation steps as .bin files.  First, I typed "chmod +x" [filename] and then "./"[filename]. I designated "/home/[username]" as the installation directory.  (In all cases, fill in the bracketed names with your actual names.)

I didn't have any .tar files to install at this point.  If I had, my notes said I should have used tar -vxf filename.tar.gz (or possibly tar xvfz instead), tar xvf filename.tar, and tar yxf filename.tar.bz2.

Drive Automount.  I wanted some partitions to be mounted automatically at startup.  In the past, I had manually edited /etc/fstab to do this.  This time, I decided to try PySDM, which was apparently short for Python Storage Device Manager (System > Administration > Storage Device Manager).  Unfortunately, my efforts suggested there were serious bugs in PySDM, so I uninstalled it and edited fstab manually.  I began by typing "sudo ntfs-config" and then "sudo gedit /etc/fstab."  I also ran System > Administration > GParted for a GUI reference, to help me see what I was supposed to be doing.  I plugged in all of my USB drives, typed "sudo blkid" to get the universal identifier (UUID) for each partition, and copied and pasted that into fstab.  I refreshed GParted (Ctrl-R) and created a comment line for each partition shown in GParted.  This was about the point when things seemed so grotesquely screwed up (because of PySDM, it seemed) as to warrant a complete reinstallation.  When I rebooted, it looked like everything was getting automatically booted without a problem.  I saved a copy of my resulting fstab in case I had to reinstall again.  I noticed that the booted partitions were all represented by icons on the desktop.  I wanted to remove those, so I typed "gconf-editor" and went into apps/nautilus/desktop, unnclicked volumes_visible, and closed the Configuration Editor, and the icons were gone.

Miscellany.  Ubuntu 9.10 used Grub2, which no longer used menu.lst.  I heard it was no longer possible to edit the Grub menu to remove entries for older kernels; instead, you had to remove the whole kernel, and then the menu entry would go away too.  ||  In System > Preferences > Startup Applications, I added Thunderbird, Firefox, and VMware Workstation.

The next steps are going to be to restore my Thunderbird profile backup and finish the automation of the rsync scripts I have been playing with for some months now.  But those steps will have to come later.

PsYDM in Ubuntu -- A Bust

I had installed 64-bit Ubuntu 9.10.   I wanted to configure several partitions to open automatically.  For this purpose, I tried using PySDM instead of editing /etc/fstab as root.

PySDM didn't show the names I had given my partitions, so as a guide I also started GParted (System > Administration > GParted, previously known as Partition Editor).  I used the names from GParted in the Name fields on PySDM.  Go by the names under the Partition List at the left side of PySDM.)  For Options, in place of the original "defaults" entry, I followed the advice to enter this for ntfs partitions:


But then I found that the Assistant button in PySDM would allow me to set these options by clicking check boxes that would explain what I was doing, except that the Owner UID was still obscure.  I changed one or two options in Assistant.  This changed "auto" (in my case) to nls=iso8856-1.  The rest of the line (after "auto") was also changed to this:  users,umask=000,utf8,gid=1000,user,owner,uid=1000.  For drive C in a dual-boot Windows XP system, I wound up with this instead:  nls=iso8859-1,users,noauto,umask=000,utf8,gid=1000,user,owner,uid=1000.

For ext3 partitions, the advice on what to enter varied, according to the type of partition:

root partition ( / ): relatime,errors=remount-ro
home partition (/home): nodev,nosuid,relatime
other partitions (aside from swap): defaults,users

But when I tried entering "defaults,users" for one such partition, PySDM just shortened it to "users."  I didn't have a separate /home partition.  For all ext3 partitions other than root, I wound up using this instead:  "owner,check,errors=remount-ro,users,user."  I clicked Apply after entering the information for each partition.

PySDM seemed to be confused in some regards.  Some of this confusion disappeared when I unplugged USB flash drives and clicked Refresh.  That is, it seemed to work best with partitions on internal SATA drives.  But it was still linking two separate partitions:  when I would make a change in one, it would make the same change in the other.  So I seemed to have two partitions with the same name.  I tried uninstalling and reinstalling PySDM in Synaptic.  (I think I had already tried closing and reopening the program.)  Eventually, I decided there are serious bugs in PySDM, so I uninstalled it and edited fstab manually.

Firefox and FEBE Problems in Ubuntu 9.10 (Karmic)

As noted in a previous post, I was having problems with using FEBE to restore a profile with all of my settings and add-ons preset.  This post describes the steps I took in a failed effort to fix that problem.

I completely uninstalled Firefox, first, by searching for it in Synaptic and marking it (3.5.7) for complete removal and clicking Apply, and then by navigating to Home in Nautilus and deleting the .mozilla/firefox folder.  It was a hidden folder, so in Nautilus I had to set View > Show Hidden Files to see it.  Then I reinstalled Firefox 3.5.7 in Synaptic and reinstalled FEBE.  To run FEBE in Firefox, I selected Tools > FEBE > Restore Profile (kill the pop-up reminders) > Create new profile (I called mine Working).  This gave me a box that offered to let me name a new profile, but that was not taking any text input.

With the aid of the tutorial, I discovered that closing Firefox would actually not close the Restore Profile window.  Now I was able to create my new profile and then Start Profile Restore.  Again, it did what it did before:  "Profile restore in progress ... Please wait" continued for much longer than the "minute or two" recommended by the tutorial, and the hard drive light was not running. I killed FEBE and restarted Firefox.  It did not show FEBE as being installed.  It did show other extensions (Tools > Add-ons) that it had not shown previously.

I installed FEBE again, this time from within the Add-ons window instead of from the Mozilla webpage, and again restarted Firefox.  Now Firefox showed only the two add-ons (FEBE 6.3.2 and Ubuntu Firefox Modifications 0.8) that it had shown at the start.  I went through the profile restore steps again.  To get the "Start profile restore" button to light up, I had to switch back and forth between the default and Working profiles.  Then I started the restore again.  Again, no action.  After a minute or two, I killed it and tried restoring a different profile.  Still nothing.

Another flaky thing that Firefox was doing:  it was opening a tiny window sometimes.  This little window could be expanded, but there was nothing in it.  In Synaptic, I did a Quick Search for Firefox and uninstalled all versions that were installed.  I then reinstalled 3.5.7 and tried again in FEBE.  Still no profile restore.  At this point, I gave up and reinstalled.

Monday, January 4, 2010

American Community Survey: 2008 Update: Defining Disability

In previous posts, I have looked into the prevalence of disabilities in the United States, as measured especially by the American Community Survey (ACS), especially in 2003. This post takes a first step toward updating the ACS measurement of disabilities, in light of changes made between 2003 and 2008.  The focus here is on the definition of disability under the 2008 ACS.

As of 2008, the ACS 2008 Subject Definitions (p. 38) defined disability as “the restriction in participation that results from a lack of fit between the individual’s functional limitations and the characteristics of the physical and social environment.” The Census Bureau asserts that “disability is not seen as intrinsic to the individual”; nonetheless, the Bureau also identifies “four basic areas of functioning” in which respondents might have “serious difficulty.” In that sense, the language still seems to reflect an impression that disabilities are the individual’s problem, rather than a mutual “lack of fit.” For instance, one might expect a different outcome if respondents were presented with a broader inquiry into functional limitations perceived either by themselves or by others in their lives (e.g., employers, neighbors, family members).

In what the Subject Definitions (p. 38) describe as “a conceptual and empirical break from earlier years,” the 2008 ACS still had six questions about disabilities, but now sought to identify “serious difficulty in four basic areas of functioning: vision, hearing, ambulation, and cognition,” and also included “two questions to identify people with difficulties that might impact their ability to live independently.”

These changes followed from a 2006 American Community Survey Content Test, whose results in the area of disability were written up by Brault and Stern (2007) in an Evaluation Report Covering Disability. The 2006 Content Test was driven by a concern with improved reliability in survey results (Brault & Stern, p. 9), not by a concern with capturing the full numbers of people with disabilities. To the contrary, Brault & Stern admit that some of the changes in the ACS disability questions tend to reduce reported prevalence (p. 8) and that the 2006 Content Test used data collection modes that contained “inherent barriers to collecting disability data” (p. 15). The latter point is important because the regular ACS appears to use essentially the same methods.

In a separate report attached as Appendix B to the Brault and Stern (2007) report (above) (i.e., beginning at page 33 of that PDF), Miller and DeMaio (2006) indicated that, for purposes of consistency with the Americans with Disabilities Act (ADA), an unspecified ACS subcommittee determined that disability would be defined as “a mental or physical impairment that substantially limits at least one major life activity” (p. B-2).  That subcommittee appears to have approached that question by exploring “domains of limitation” or “limitation domains” (pp. B-2, B-3), instead of taking the seemingly obvious step of deciding what counts as a major life activity.

To cite some rather choice examples, Miller and DeMaio did not indicate that the subcommittee considered sleep, sex, or socializing to be major life activities.  All three of those examples seem to involve the Census Bureau’s alleged orientation toward the two-sided question of fit between person and society.  It is society that supplies the people, the stresses, and other key components and determinants of sleep, sex, and socializing.  One would expect, in other words, that these three would be prime candidates for inclusion in the list of major life activities, or as essential contributors thereto.  For instance, there is evidence that sleep deprivation significantly interferes with work and other major daily activities of tens of millions of Americans (Baxter & Kroll-Smith, 2005, p. 39), that it results in substantial numbers of injuries and deaths (e.g., Lyznicki, Doege, Davis, & Williams, 1998), and that one-third of Americans take naps on a typical day (Taylor, 2009).  Although it is presumably not necessary to do so here, one certainly could develop a more extensive argument for sleep, and could make similar arguments for the inclusion of sex, socializing, and other activities that the subcommittee somehow considered not to be "major" for purposes of detecting the existence of a disability.

The subcommittee's decisions in this regard might have been more defensible if the ADA had been construed as taking an interest in major productive activities.  Viewed that way, the subcommittee could reasonably conclude (especially in a traditional industrial economy) that it is all right if workers are tired, as long as they keep working.  Likewise, if industrial-type productivity were the stated focus, attention to sex and socializing would probably seem humorously superfluous.  The full list of impairments that mattered to the subcommittee -- involving vision, hearing, lower body mobility, cognition, activities of daily living (ADLs), instrumental activities of daily living (IADLs), and work (Miller & DeMaio, 2006, p. B-3) would then make sense.  ADLs “generally include self-care type activities such as bathing and dressing,” and IADLs are activities often associated with independent living such as going out alone to shop or visit a doctor’s office” (Brault & Stern, 2007, p. 4), all of which align with an emphasis upon the worker's functionality and his/her ability to remain functional.

Ironically, contrary to the apparent purpose of the ADA, a functional orientation can reinforce stereotyping and exclusionism.  If the majority defines functionality in terms of what the majority can do, then those in the minority will tend to occupy an inferior role.  It will be newsworthy when technology enables them to overcome – say, to run faster than those who began with what was, by definition, a functional advantage (e.g., Christie, 2009).  In reality, there may be many ways (involving e.g., sleep, sex, or socializing) in which many putatively high-functioning people are experiencing profound limitations in major life activities; but that will not be recognized in the ACS until a standard other than industrial-type functionality is used to guide disability determinations.  There would seem to be a number of good, currently relevant candidates for such a standard ranging from “life, liberty, and the pursuit of happiness” to social capital to adjustment to or satisfaction with life – that could be used alone or in combination for that purpose.

The apparent functional orientation was not the only way in which the 2008 ACS reduced the count of disabilities even below what one might have expected from the subcommittee’s own stated concept. The survey discarded the subcommittee’s broad interest in mental and physical impairments, in favor of very limited kinds of difficulties. The survey’s only questions about mental disabilities were as follows:

18a. Because of a physical, mental, or emotional condition, does this person have serious difficulty concentrating, remembering, or making decisions?
19. Because of a physical, mental, or emotional condition, does this person have difficulty doing errands alone such as visiting a doctor’s office or shopping?
Bizarrely, the 2008 ACS no longer has even the question about work disability (Brault & Stern, 2007, p. 7). That is, there is no question about the presence of a physical, mental, or emotional condition that would limit the kind or amount of work a person can do. So if someone has no problem concentrating etc. (Question 18a), and does not have difficulty doing errands alone (Question 19), it seems s/he would not have a work disability even if s/he spent the workday standing on his/her desk and screaming.

Apparently the question about work disability was deleted because Miller and DeMaio (2006, p. B-32) claimed there was an “excessive extent of misinterpretation and misreporting in this question.” One objection Miller and DeMaio cited had to do with vagueness in the word “work.” Some people reportedly took it to include work around the house. That raised interesting questions in itself, but in any event one version of the test question specified work “at a job or business” (p. B-31). Despite the alleged breadth and complexity of the work construct, Miller and DeMaio devoted only about 1.5 pages to the question of work before recommending its deletion.  (While I did not think that disabilities should be defined exclusively in work-related terms, it did seem obvious that inability to work would be an important indicator of disability.)  That is, Miller and DeMaio did not seriously wrestle with the matter; yet they also did not claim that the question of work-related disability is unimportant.

Bear in mind that Miller and DeMaio (2006), who balked at what they consider excessive vagueness in the work construct, were the same researchers who pointed out that “Respondents’ answers to survey questions are necessarily based on personal experience and perceptions of that experience” (p. B-3), that “it is unfeasible to develop questions that yield a perfect measure of disability” (p. B-4), and that, in connection with a question that was retained in the survey, respondents were being required “to discern a clear line of yes or no in a reality that, for them, was essentially grey and multi-dimensional” (p. B-7). All of these remarks seem to suggest that Miller and DeMaio were generally comfortable with the hard decisions that arise in the process of quantifying experience – including work experience.

Under such circumstances, one might be forgiven for asking whether political considerations during the Bush Administration had any influence upon the choice of researchers and/or the recommendations they made in such regards. In a report on changes to measurement of disability in the 2008 American Community Survey that was produced after Obama’s election, Brault (2009, p. 5) seems to indicate that the previous administration had not prioritized the ACS generally. He says that, “due to budgetary issues, FEFU [i.e., Failed Edit Follow-Up] operations were scaled back from April through September” of 2008.

In his 2009 report, Brault states that the net effect of the changes in questions on the 2008 ACS, including the deletion of the work disability question, resulted in a marked drop in the estimated number of Americans with a disability, from 41.2 million (2007) to 35.9 million (2008). “Given the differences in the questionnaire,” Brault says, “one should not interpret these changes as real differences in the number or percent of people with disabilities” (p. 16).  Brault emphasizes that ACS disability data from 2008 are not comparable with ACS disability data from prior years.

This is not to say that the 2008 ACS is useless for all disability-related purposes. Brault (2008, p. 8) suggests, for example, that the ACS is more likely to capture instances of hearing difficulty than is the Survey for Income and Program Participation (SIPP), because the SIPP excludes hearing difficulties that are corrected by hearing aids. That particular example is not necessarily good or bad in itself, but it does suggest that the ACS may sometimes provide a useful alternate picture on the prevalence of some kinds of disabilities.

There are some other limitations on ACS data.  First, ACS data of all sorts are not yet available for geographic areas with populations of less than 20,000.  In late 2010, according to the Census Bureau, data will be released “for areas as small as census tracts and block groups, nationwide. For very small towns, 2010 will be the first time that ACS data are published. These data products, referred to as 5-year multiyear estimates, will be based on the data accumulated during the 2005-2009 time period.”  The Bureau also says that information on demographic, socioeconomic and housing characteristics will thereafter be available annually for all areas.  These include annual estimates for all areas of 65,000+ population, three-year averages for all areas of 20,000+ population, and five-year averages for census tracts and block groups.  Note also that, starting in 2006, the ACS did begin to include data for the group quarters population.  These data were not available, however, in 15 states that had a total group quarters population of less than 45,000.

Because of the change in the operational definition of disability in the 2008 ACS, the Census Bureau will not be able to provide three-year estimates for disabilities until the 2008-2010 ACS 3-year estimates are released in 2011.  That is, the change in definitions means that users should not compare disability data from 2008 and afterwards to any year before 2008.

Recap So Far

In this series of posts on the prevalence of disabilities in the U.S., I have been moving toward an estimate of disabilities on the local level.  It seemed logical to begin by choosing a national-level survey that would provide local-level data, and then to understand that survey and, on that basis, to interpret its local data.  Preliminarily, the ACS seemed far superior to other federal surveys, in terms of its penetration to the local level.  It did not define disabilities as comprehensively as some other surveys, but it seemed that it might be possible to adjust its reported prevalence rates upwards by some plausible percentage, so as to reach a defensible ballpark figure for actual local disability prevalence.

Unfortunately, this post's critique of the definition of disabilities, and of its operationalization in the 2008 ACS, suggests that the ACS does not provide accurate data on the prevalence of disabilities.  Hence, before conducting further research into federal disability statistics, it presently seems advisable to pursue some other routes.  Several such routes come to mind.  First, there is surely a great deal of relevant third-party literature on the ACS, as distinct from the literature provided by the Census Bureau itself.  Second, there may be good local-level research on the prevalence of disabilities in various types of cities and counties, sufficient to inform an educated guess on prevalence in other similar local areas.  Third, on a more theoretical level, there is no doubt much to learn about the conceptualization of disability -- on, for example, the validity of viewpoints suggested in this post.  Since the conceptualization seemed likely to influence the interpretation of research, my next step in this investigation moved in the conceptual direction.

Sunday, January 3, 2010

American Community Survey (2003): Disabilities on the State Level

In a previous post, I examined some statistics provided by the American Community Survey (ACS) for 2003, regarding the prevalence of certain kinds of disabilities on the national level.  Now I turn to the interpretation of the ACS on the state level.

Generally, as a Census Bureau methodological report explains, state population estimates are obtained by adding up the estimated numbers of people in each county.  The calculation of county population begins with the decennial census.  At present, that is the census taken in 2000, but of course there will be a new one this year (2010).  The Bureau makes adjustments from the previous year based on a variety of county-level sources, including tax records, estimates of migration, and registered births and deaths.  Previous years' estimates are revised as new information sheds light on actual developments.

State totals are not added up to produce the national population, however.  Instead, national totals are developed independently, and are used as a check upon county-level calculations.  Roughly speaking, if the sum of all state totals (that is, the sum of all county totals) were to exceed the independently developed national population estimate by 3%, then the population estimates for all counties would be reduced by 3% so that the state totals would match the national estimate.

Given this information, a lay reader might suggest that a parallel method of estimating the prevalence of disabilities within a county -- and thus within a state -- would supplement the ACS questionnaire by investigating actual records of disabilities, from such sources as hospital and school records, handicap-plate motor vehicle registrations, and Social Security records.  Presumably the efforts of researchers have uncovered shortcomings in decennial censuses; one might expect that these sorts of disability-related investigations would exert a similar beneficial influence on ACS disability estimates.

According to Weathers (2005, p. 61), the Census Bureau does investigate ways in which its statistical processes may produce errors.  Errors may be random, in which case statistical adjustments can be made or estimates of error can be produced.  Errors may also be nonrandom.  Nonrandom errors are also called systematic errors.  These occur, not because (for example) a predictable percentage of statistical typists will make a predictable percentage of typographical errors as they enter the data, but because some kind of atypical distortion occurs with respect to some particular kind of person or situation.  If, for example, followers of a certain religion were concentrated in a particular county, and if their religion taught that it was wrong to respond to surveys, then there could be a pronounced systematic, nonrandom undercounting of people in that county.  Along these lines, it would certainly seem that people who have physical disabilities that make every task a chore, or mental disabilities that discourage cooperation with governmental surveys they perceive as suspicious, could be systematically undercounted.

Weathers (2005, p. 61) indicates that the Census Bureau maintains information on identified systematic errors in the ACS at a page on its website.  A quick search of that webpage finds, at present, no entries pertaining to disabilities.  Weathers (pp. 68-70) does note certain regards, however, in which a redesign of the ACS disability questions resulted in dramatic and potentially erroneous declines in reported disabilities in 2003.  There have been substantial changes in the ACS measurement of disabilities since then.  A subsequent post will discuss those changes.  First, however, the following paragraphs explore the 2003 ACS in light of Weathers's comments, many of which are still applicable and/or have not yet been revisited.

ACS 2003 state-by-state disability prevalence rates (Weathers, 2005, pp. 45-46) raised questions of consistency in data collection procedures.  In the Midwest, for example, most states were fairly similar to one another:  overall disability rates were between 11.2 and 13.3 in Ohio, Indiana, Iowa, Michigan, Wisconsin, Missouri, Kansas, and Nebraska.  Indeed, without Indiana and Ohio, the range among those states would have been in the narrow band of 11.2 to 12.4.  Yet somehow, in the middle of those states, Illinois – right next to Indiana (13.3) – somehow produced a rate of 9.2.  Certainly it is plausible that a city like Chicago would manage to accommodate its persons with disabilities better than a more rural state; but why the presence of a large city would affect disability prevalence itself is not intuitively obvious.  If population density itself were a positive factor, Rhode Island (12.0) would not have had a rate considerably higher than those of its neighbors Connecticut (9.2) and Massachusetts (9.7).

Certainly the 2003 ACS state-level data are interesting.  Weathers (2005, pp. 23-24) compares states in terms of the levels of employment, poverty, and household income experienced by persons with disabilities.  His tables 7-9 (pp. 47-52) also provide relative comparisons of the experiences of people with and without disabilities in those several regards.  Thus, for example, whatever the rate of unemployment in West Virginia as a whole, the data reveal that that state’s people with disabilities are employed at only about one-third (35.8%) the rate of its people without disabilities -- as compared to Wyoming, on the opposite extreme among the lower 49 states, where the rate is more like two-thirds (65.5%).  Among the midwestern states just listed (including Illinois), that relative rate ranges from 48.4% (Michigan and Ohio) to 55.8% (Nebraska).

According to Weathers (2005), the relative experience of poverty for people with disabilities, like their relative level of employment, varies considerably among states.  At the low end, people with disabilities in Utah are only 2.1 times as likely to fall below the povety line as are people without disabilities.  At the high end, people in Nebraska are 4.9 times as likely to do so.  Nebraska aside, the midwestern states listed above are within the range of 3.4 (Illinois) to 3.8 (Kansas).

As with some of the other values discussed here, relative household incomes contrast western states against southern states.  Utah leads with a value of 75.9% – that is, the median household income of a person with disabilities in Utah is 75.9% of the median household income of a person without disabilities – and Louisiana (49.6%) and Alabama (50.4%) are beaten at the bottom end only by Delaware (which, with the slightly lower value of 48.3%, is an outlier in regional terms by several of these measures).  The midwestern range is from 54.2% (Ohio) to 62.3% (Wisconsin).  Relatively large differences among these states’ neighbors (e.g., 61.9% in Indiana, 56.4% in Michigan) raise the question of how state-level policies impact these numbers.

Some caveats are in order.  First, as with most of the observations in this post, it remains to be seen how these data have changed since 2003.  Note, too, that the foregoing analyses provided by Weathers (2005, p. 19) are focused upon the working-age population, ages 25-61.  Also, the ACS presents values on employment and poverty that are in some regards markedly divergent from those reported by most other national studies cited in a previous post in this series (Weathers, p. 29).

Across all age groups, Weathers (2005, pp. 20-21), notes that national counts of disabilities may be influenced by race, culture, gender, and education.  Black people comprise 13.8% of people with disabilities, as compared to 11.7% of the population without disabilities.  Hispanic people comprise 14.0% of the population without disabilities, but only 9.6% of the reported population with disabilities.  Women constitute 52.8% of the population with disabilities (versus 51% without), and are especially highly represented in disabilities involving self-care (58.9%) and going outside the home (63.6%).  People with less than a high school education account for 11.6% of people without a disability, but they account for 25.0% of people with a disability.  Disability rates may be correspondingly affected, in states that vary from the mean in any of these demographic regards.

Having provided an introduction to state-level measurement of disabilities through the ACS in 2003, largely as interpreted by Weathers (2005), the next step is to examine how ACS federal- and state-level measurements and results were changed in 2008.

Saturday, January 2, 2010

American Community Survey (2003): Disabilities Nationwide

In a previous post, I took a quick look at how the total household population figure in the American Community Survey (ACS) for 2003 was calculated.  The numbers involved are not precise in every case, but the general concept appears to be that the "resident population" is divided into people who live in households and those who live in group quarters (e.g., prisons, hospitals, dormitories).  In the 2003 ACS, only households were examined.  The total number of people in the U.S. who were living in households in 2003 was estimated to be 282,909,885.  Of those, an estimated 176,395,446 fell into the traditionally defined working-age population aged 18-64.

The next thing to figure out, in moving toward statistics on disabilities at the county level, was how many of those 176,395,446 people were considered to have various kinds of disabilities.  Here, as before, I relied on the analysis of 2003 ACS data provided by Weathers (2005) in his Guide to Disability Statistics from the American Community Survey.  The previous post notes that the ACS understates disabilities when compared to some other national surveys; however, it has the advantage of being updated on the county level, and thus provides an essential stepping-stone in this investigation.

Weathers (2005) divides that total of 176,395,446 working-age people into those with (12%) and without (88%) disabilities, as defined by the ACS.  Weathers further narrows the working-age population down to the 25-61 age range, considering the 18-24 group as being of a "school-to-work transition age" and the 62-64 group as being of "early Social Security retirement age" (p. 19).  Since the 25-61 group accounts for 83% (i.e., 17,146,845) of the total of 20,609,733 people with disabilities in the larger 18-64 group (p. 54), I will tend to focus on that 25-61 group in this post.

To summarize, then, the ACS gives us, in 2003, an estimated 17,146,845 U.S. residents who lived in households (as distinct from group homes), had one or more disabilities, and fell into the 25-61 age group.  These people comprised roughly 10% of the estimated working-age (18-64) population of about 176 million.

According to Weathers (2005, p. 35), the ACS draws upon World Health Organization (WHO) concepts of impairment, activity limitation, and participation restriction, as expressed in the International Classification of Functioning, Disability and Health (ICF) (pp. 4-5).  In the ICF, Weathers says, an impairment is a significant deviation or loss in body function or structure (e.g., vision loss).  An activity limitation is difficulty in executing activities of daily living (e.g., dressing).  A participation restriction is a problem that a person may have in a life situation (e.g., lack of employer accommodation to the person's severe health condition).

As Weathers (2005, pp. 10-11, 35) describes, the ACS uses six questions to operationalize those ICF concepts.  Three of the six ask about impairments:  sensory, physical, or mental.  One asks about activity limitation, in the form of an inability to care for oneself inside one’s home.  Two ask about participation restrictions, in the form of physical, mental, or emotional conditions that prevent the person from going outside the home (e.g., to shop) or from working at a job or business.  The ACS defines disabilities as the presence of any one or more of these kinds of disability. As an example of how these questions may fail to count some disabilities, Weathers (p. 27) indicates that the question of going outside the home, used on the ACS, is narrower than the concept of Instrumental Activities of Daily Living (IADLs) used in some surveys.

The 2003 ACS data indicate that, for the 25-61 group, two of these six questions (pertaining to physical impairment and employment restriction) account for 55% of all disabilities.  Those two plus the next most frequently cited disability, regarding mental impairment, account for 70% of all disabilities in that age group.  Yet within that statement lies an interesting age-related permutation.  First, as one might expect, age is a factor in the calculation of disability under the ACS.  The data show that disability prevalence rates rise from 6.3 in the 5-17 group (that is, 6.3% of people in that age group have a disability) to 39.9 in the 65+ group (Weathers, 2005, p. 39).  At the same time, however, while those top three questions account for about 70% of disabilities in all three working age groups (i.e., 18-24, 25-61, and 62-64), the share due to mental impairments steadily declines.  Mental impairments constitute 30% of all disabilities in the 18-24 group, but only 15% in the 25-61 group and 10% in the 62-64 group.

It is not that mental impairments actually decline with age, though that may be the case for some kinds of mental impairments.  It is, rather, that mental impairments rise only slightly, from 3.7 in the 18-24 group to 5.8 in the 62-64 group.  This is quite different from the ninefold rise in physical disabilities between those two groups.  One explanation for that contrast is that the ACS may undercount mental disabilities in adults.  There is evidence of reduced awareness of mental disability on the part of adults and their primary care physicians (e.g., Surman, Wigal, & Lakes, 2009).  Undercounting of mental disabilities in adults would also be unsurprising if significant mental disabilities are incompletely conceptualized among adults (e.g., Nijmeijer et al., 2008, p. 701).

So far in this series of posts on statistics pertaining to disabilities, I have rarely gone beyond the relatively general descriptive approach provided by commentators such as Weathers (2005).  The preceding paragraph suggests, though, that a look at more critical or focused literature may highlight a number of shortcomings with the ACS, for purposes of counting disabilities.  As an example of another area in which a more refined approach may considerably enhance the statistical picture, one may review Weathers's remark (above) about the refinement that the IADL concept adds to the question of going outside the home.  Unless Weathers meant to contrast IADL against the ACS's question regarding self-care, and simply misspeaks on that point (compare pp. 27 and 39), it seems appropriate to question the finding that 6.9% of people in the 25-61 age group experience a significant employment restriction due to disability, and yet only 2.0% report a limitation in their self-care activity.  Without a boss or other reminder of a set goal, people at home may be more likely to adjust their goals and lives to accommodate their disabilities.

Those examples of potential shortcomings in the ACS may or may not be borne out in the professional literature.  I can't say; I haven't gotten that far yet.  The purpose of those remarks was simply to highlight a few ways in which the ACS may undercount disabilities.  Having achieved that, the next step in this investigation is to move from the national-level ACS data to the state level.