According to the U.S. Census Bureau, the unemployment rate on the Navajo Nation and its trust lands is 22.1 percent. Source: 2013 American Community Survey

The job market remains a disaster zone for Indian workers in reservation areas. Even as the Obama Administration was touting progress in Indian Country in a summit meeting with tribal leaders, the Census Bureau reported that the unemployment rate for the Indian workforce in federal reservation areas is 22.6%, close to two-and-a-half times that for all workers nationally.

The number would be even worse if all on-reservation Indian workers who don't report "actively" looking for work they know is not available were to be counted among the unemployed. If such "discouraged workers" and others who could be working if work were available were included in the computations, the unemployment rate would be a staggering 37%.

This new data on the employment status of the on-reservation Indian population comes from the Census Bureau's American Community Survey (ACS). The survey is an ongoing sample of American households that produces data on the socio-economic characteristics of the population down to the neighborhood level.

Data for smaller communities, including all reservation areas, is aggregated over five-year periods to produce what Census regards as reliable numbers. The ACS data released on December 4th covers information from survey questionnaires collected during calendar years 2009 through 2013.

In analyzing the labor market data for the on-reservation Indian population it is necessary to look beyond just the unemployment figures. To be counted as unemployed a person must have no job, be available for work and must have been "actively" seeking work during the four-week period before responding to a questionnaire.

In reservation areas, this "actively" seeking work test ignores those who don't look for jobs they know are not available to them. Such persons are counted as not even part of the labor force. A broader, though still inadequate test of labor market problems is contained in the labor force participation rate. This counts those in the labor force as a percentage of the total population age 16 and over. In many reservation areas the labor force participation rate is substantially below that for the workforce of all races nationally, reflecting the problem with the "actively" seeking work definition.

Labor force problems affect all Indian workers nationally even if not quite to the same degree as Indian workers on reservation. The graph below contrasts the situation of all workers, of all races, nationally with that of Indian workers nationally and of Indian workers in federal reservation and trust land areas. The data source is the new ACS data set for 2009 to 2013.

Data Reliability
The ACS is a survey. Even though it's larger than other federally administered surveys of the general population, it's a relatively small sample of the characteristics of all individuals. The smaller the population and the smaller the geographic area, the larger the amount of sampling error. In other words, the numbers extrapolated from the sample may not accurately represent the real characteristics of the population.

There is a continuing problem with the reliability of the ACS data on the Indian population.

At the national level, the total size of the Indian (AI/AN alone) population shown in the ACS is far less than the number actually counted in the 2010 Census. The 2009-2013 ACS data set shows a total Indian count of a little over 2.5 million. However, over 2.9 million people said their only race was Indian in the 2010 Census. The Census Bureau simply dismisses this undercount of 400,000 by saying that people switched their identity from being only Indian in the 2010 Census to being Indian and also members of another race when they answered an ACS questionnaire.

In some respects the ACS appears to be getting a little more reliable, at least from a sampling error perspective.

The reliability test in this analysis uses a statistic that measures relative sampling error called the "coefficient of variation." There is no universally accepted standard of when a coefficient of variation (CV) shows an estimate from a sample survey to be reliable. The standard used in this analysis considers an estimate statistically reliable if the CV is 15% or less. An estimate is considered as of questionable reliability if the CV is between 15% and 30%, and unreliable if the CV is over 30%.

The data from the 2009 to 2013 ACS data set for the number of unemployed Indians between the ages of 16 and 64 met the reliable standard for 28% of the 86 largest reservations, those with an Indian population in 2010 of 1,000 or more. This was an improvement over the 2007 to 2011 ACS data set, in which only 16% of the same 86 large reservations met the reliable test.

At the other end of the spectrum -- the percentage of the 86 largest reservations with a count of unemployed Indians with a CV of over 30% -- only 14% of these reservations had data in this unreliable range in the ACS 2009-2013 figures. In the 2007 to 2011 data set, 27% of reservations had unreliable data.

Improvements made in ACS coverage of reservation areas and other small rural communities in mid-2011 may have had some effect in reducing sampling error in the results for a number of the larger reservations.

But the bad news is in the middle of the spectrum. The percentage of the 86 largest reservations with data on the number of unemployed Indians in the potentially unreliable range has remained steady in a 57% to 59% range for the last three ACS 5-year data sets.

The analysis focused on the large reservations where there is a better chance that the questionnaires collected will be a truly representative sample of the Indian population. The picture for the smaller reservations, a much larger percentage of all 325 reservations counted in the Census data, is much worse. In general, the smaller the reservation, the less likely it is that the ACS figures represent an accurate picture of the conditions among the Indian population.

In addition to sampling error, survey data is also susceptible to non-sampling error. This may be a factor in reducing the reliability of some of the reservation unemployment data, even for the larger reservations where the level of sampling error is relatively low.

Additional Information Available
A table with figures on Indian unemployment for each reservation in the country, former reservation areas in Oklahoma (called OTSAs) and Alaska Native villages from the ACS data covering the period from 2009 to 2013 is available from the author on request. See the e-mail address below.

Subsequent data notes on the characteristics of the on-reservation Indian population from the ACS 2009 to 2013 data will focus on poverty status and educational attainment.

Norm DeWeaver is an economic development and data expert and the former director of the Center for Community Change. His email is: norm_deweaver@rocketmail.com