Hair Drug Tests Don’t Discriminate
Executive Summary
A University of Central Arkansas study of 112,693 paired urine and hair test results for drug use (UCA Hair Test Study) found no evidence that a person’s race has any bearing on the results. The UCA Study applied two recognized test measurements and found that neither the urine or hair test method had a disparate impact on ethnic groups. Further, the study observed that factors other than the drug test method, seem to underly differences in an ethnic group’s pass/fail rate.
Background
- In 2015, Congress authorized the Department of Transportation (DOT) to utilize hair testing as an acceptable alternative to a urine test, for detecting illegal controlled substances among operators of commercial vehicles.
- Yet, four years later, the Department of Health and Human Services has failed to complete guidelines so hair test failures can be submitted to the Drug and Alcohol Clearinghouse. Opponents have alleged hair testing might be racially biased.
- The Alliance for Driver Safety & Security, also known as the Trucking Alliance, is an industry-based safety coalition of freight and logistics companies and other business partners. The Trucking Alliance mission is to advance safety reforms that can achieve zero large truck crash fatalities.
- The Trucking Alliance asked Dr. Doug Voss and Dr. Joe Cangelosi at the UCA College of Business (Researchers) to assess whether actual hair test results for drug use disparately impact minority groups, when compared to their urine test results.
- Researchers utilized data analytics to compare paired urine and hair test results of 73,176 truck driver applicants by racial breakdown, submitted by companies for all pre-employment drug tests in 2017, 2018, and 2019. The researchers also studied a separate population of 39,517 paired drug test results of truck drivers submitted by one company for 2018. The population of paired urine and hair test results totaled 112,693 truck drivers.
Research Methodology
Researchers utilized two methods to assess whether a disparate impact occurred within the truck driver populations. First, Researchers utilized the “Four Fifths Rule,” a method relied on by the federal government to determine if an ethnic group is being unfairly discriminated in hiring policies. The Four Fifths Rule is defined in the Code of Federal Regulations, Title 29 – Uniform Guidelines for Employee Selection Procedures, as “a selection rate for any race, sex or ethnic group which is less than four-fifths (4/5) (or eighty percent) of the rate for the group with the highest rate will generally be regarded by the Federal enforcement agencies as evidence of adverse impact, while a greater than four-fifths rate will generally not be regarded by Federal enforcement agencies as evidence of adverse impact.”
In other words, disparate impact is assumed to have occurred, if any ethnic group does not pass a standard, at a rate of at least 80% the rate that the ethnic group with the highest passing rate performed.
Second, Researchers utilized “Chi-Square Difference Tests” to assess whether there were significant differences between ethnic groups for each test. Chi-Square results would indicate disparate treatment if all ethnic groups experienced similar urine test results, but certain ethnic groups experienced significantly different hair test results. Alternatively, Chi-Square Difference Test results would indicate equal treatment and no disparate treatment, if significant between-group differences exist for both or neither urine and hair tests. This would imply that the groups pass/fail rates are statistically equivalent or different, irrespective of the testing procedure.
Test Results
- Results for each test in each sample met the required Four-Fifths Rule threshold, which indicated that all ethnic groups were significantly above the 80% required threshold. Therefore, there was no disparate treatment of any racial group.
- The Chi-Square results indicated that the proportion of drug test failures (positive for drug use) were higher with hair testing but higher across all ethnic groups. All ethnic groups failed the hair test at higher rates than the urine test. Given these findings, Researchers were unable to identify evidence of disparate impact among any ethnic group resulting from the use of hair tests.
National Significance of Study
Results of the UCA Hair Test Study can be applied to the US truck driver population, estimated to be approximately 3.5 million.
Researchers noted that given a margin of error = 1%, and a confidence level = 99%, a sample size of 16,641 is required to generalize results across the broader U.S. truck driver population. The sample sizes used in the UCA Hair Test Study exceed this threshold and results can be generalized Nationally.[1]
(Charts Below)
UCA HAIR TEST STUDY FINDINGS
TABLE OF URINE AND HAIR TEST FAILURE RATES
| ETHNIC GROUP | 2018 Urine | 2018 Hair | Difference | 2017-2019 Urine | 2017-2019 Hair | Difference |
| AM. INDIAN | 0.5% | 6.8% | 6.2% | 0.8% | 6.3% | 5.6% |
| ASIAN | 0.6% | 4.1% | 3.5% | 0.4% | 3.4% | 3.0% |
| BLACK | 1.1% | 7.6% | 6.6% | 1.0% | 7.8% | 6.7% |
| HAWAII/PACIFIC ISLANDER | 0.0% | 3.5% | 3.5% | 0.7% | 6.2% | 5.5% |
| HISPANIC | 0.6% | 5.9% | 5.4% | 0.5% | 5.5% | 5.0% |
| MULTIPLE | 2.0% | 8.2% | 6.2% | 1.4% | 7.7% | 6.4% |
| NOT SPECIFIED | 1.9% | 12.5% | 10.6% | 1.7% | 11.5% | 9.8% |
| WHITE | 1.0% | 4.2% | 3.2% | 1.0% | 4.2% | 3.2% |
| AVERAGE | 1.0% | 6.6% | 5.7% | 0.9% | 6.6% | 5.6% |
The table above illustrates that each ethnic group had a higher hair test failure rate. Tables on the next page rank order the failure rate differences for the two samples then draw comparisons.
RANKING OF URINE AND HAIR TEST FAILURE RESULTS
|
|
The above table ranks the ethnic groups in each sample by the difference between hair and urine test failure rates. The rank order across samples was very similar.
- The Not Specified group had the greatest difference in both the 2018 (10.6%) and 2017-2019 (9.8%) samples.
- Blacks had the second greatest difference (2018 = 6.6%; 2017-2019 = 6.7%)
- Multiple race persons were third (2018 = 6.2%; 2017-2019 = 6.4%)
- The American Indian failure rate in 2018 was higher than the 2018 sample average. The American Indian failure rate for 2017-2019 (5.6%) was roughly equivalent to the sample average.
- Hispanic (2018 = 5.4%; 2017-2019 = 5.0%), Hawaii/Pacific Islanders (2018 = 3.5%; 2017-2019 = 5.5%), Asians (2018 = 3.5%; 2017-2019 = 3.0%) and Whites (2018 = 3.2%; 2017-2019 = 3.2%) had lower than average failure rates for both samples.
FOUR FIFTHS RULE
TABLE 1
2017-2019 URINE TEST RESULTS
|
ETHNIC GROUP |
PASSED |
FAILED |
TOTAL |
PERCENT PASSED |
PERCENT OF HIGHEST PASSING RATE (ASIAN) |
| AM. INDIAN | 753 | 6 | 759 | 99.2% | 99.6% |
| ASIAN | 1802 | 7 | 1809 | 99.6% | 100.0% |
| BLACK | 28632 | 294 | 28926 | 99.0% | 99.4% |
| HAWAII/PACIFIC ISLANDER | 276 | 2 | 278 | 99.3% | 99.7% |
| HISPANIC | 8191 | 44 | 8235 | 99.5% | 99.9% |
| MULTIPLE | 1777 | 25 | 1802 | 98.6% | 99.0% |
| NOT SPECIFIED | 8327 | 144 | 8471 | 98.3% | 98.7% |
| WHITE | 22664 | 232 | 22896 | 99.0% | 99.4% |
| TOTAL | 72422 | 754 | 73176 | 99.0% | 99.4% |
*Pearson chi-square = 67.52; p = 0.00; n = 73,176
Table 1 details 2017-2019 urine test results. Ninety nine percent (99%) of drivers in the Asian ethnic group passed their pre-employment drug screens. To comply with the Four-Fifths Rule, every other ethnic group must pass at a rate equal to 80% of this figure (99% x 80% = 79%). Drivers who chose not to report their ethnic group (“not specified”) passed at the lowest rate, which was 98.7% of the ethnic group with the highest passing rate. This exceeds the required Four-Fifths Rule 79% threshold.
TABLE 2
2017-2019 HAIR TEST RESULTS
|
ETHNIC GROUP |
PASSED |
FAILED |
TOTAL |
PERCENT PASSED |
PERCENT OF HIGHEST PASSING RATE (ASIAN) |
| AM. INDIAN | 709 | 48 | 757 | 93.7% | 97.0% |
| ASIAN | 1739 | 61 | 1800 | 96.6% | 100.0% |
| BLACK | 26329 | 2215 | 28544 | 92.2% | 95.5% |
| HAWAII/PACIFIC ISLANDER | 258 | 17 | 275 | 93.8% | 97.1% |
| HISPANIC | 7699 | 452 | 8151 | 94.5% | 97.8% |
| MULTIPLE | 1655 | 139 | 1794 | 92.3% | 95.5% |
| NOT SPECIFIED | 7149 | 925 | 8074 | 88.5% | 91.7% |
| WHITE | 21678 | 950 | 22628 | 95.8% | 99.2% |
| TOTAL | 67216 | 4807 | 72023 | 93.3% | 96.6% |
*Pearson chi-square = 624.6; p = 0.000; n = 72,023
Table 2 details 2017-2019 hair test results. Ninety six percent (96%) of drivers in the Asian ethnic group passed their pre-employment drug screens. To comply with the Four-Fifths Rule, every other ethnic group must pass at a rate equal to 80% of this figure (96% x 80% = 77%).
Drivers represented by the Black and Multiple ethnic groups passed at the lowest rate, which was 95.5% of the ethnic group with the highest passing rate. This exceeds the required 77% Four-Fifths Rule threshold.
TABLE 3
2018 URINE TEST RESULTS
|
ETHNIC GROUP |
PASSED |
FAILED |
TOTAL |
PERCENT PASSED |
PERCENT OF HIGHEST PASSING RATE (HAW/PAC IS.) |
| AM. INDIAN | 367 | 2 | 369 | 99.5% | 99.5% |
| ASIAN | 847 | 5 | 852 | 99.4% | 99.4% |
| BLACK | 15345 | 166 | 15511 | 98.9% | 98.9% |
| HAWAII/PACIFIC ISLANDER | 144 | 0 | 144 | 100.0% | 100.0% |
| HISPANIC | 4022 | 23 | 4045 | 99.4% | 99.4% |
| MULTIPLE | 851 | 17 | 868 | 98.0% | 98.0% |
| NOT SPECIFIED | 6043 | 118 | 6161 | 98.1% | 98.1% |
| WHITE | 11455 | 112 | 11567 | 99.0% | 99.0% |
| TOTAL | 39074 | 443 | 39517 | 98.9% | 98.9% |
*Pearson chi-square = 59.43; p = 0.000; n = 39,517
Table 3 details 2018 urine test results. One hundred percent (100%) of drivers in the Hawaii/Pacific Islander ethnic group passed their pre-employment drug screens. To comply with the Four-Fifths Rule, every other racial group must pass at a rate equal to 80% of this figure (100% x 80% = 80%).
Drivers who chose not to report their ethnic group (“not specified”) passed at the lowest rate, which was 98.1% of the ethnic group with the highest passing rate. This exceeds the required 80% Four-Fifths Rule threshold.
TABLE 4
2018 HAIR TEST RESULTS
|
ETHNIC GROUP |
PASSED |
FAILED |
TOTAL |
PERCENT PASSED |
PERCENT OF HIGHEST PASSING RATE (HAW/PAC IS.) |
| AM. INDIAN | 344 | 25 | 369 | 93.2% | 96.6% |
| ASIAN | 812 | 35 | 847 | 95.9% | 99.3% |
| BLACK | 14163 | 1173 | 15336 | 92.4% | 95.7% |
| HAWAII/PACIFIC ISLANDER | 136 | 5 | 141 | 96.5% | 100.0% |
| HISPANIC | 3780 | 239 | 4019 | 94.1% | 97.5% |
| MULTIPLE | 795 | 71 | 866 | 91.8% | 95.1% |
| NOT SPECIFIED | 5110 | 731 | 5841 | 87.5% | 90.7% |
| WHITE | 11000 | 481 | 11481 | 95.8% | 99.3% |
| TOTAL | 36140 | 2760 | 38900 | 92.9% | 96.3% |
*Pearson chi-square = 438.14; p = 0.000; n = 38,900
Table 4 details 2018 hair test results. Ninety six percent (96%) of drivers in the Hawaii/Pacific Islander ethnic group passed their pre-employment drug screens. To comply with the Four-Fifths Rule, every other ethnic group must pass at a rate equal to 80% of this figure (96% x 80% = 77%). Drivers who chose not to report their ethnic group (“not specified”) passed at the lowest rate, which was 90.7% of the ethnic group with the highest passing rate. This exceeds the required 77% Four-Fifths Rule threshold.
CHI-SQUARE DIFFERENCE TESTS
Chi-square results are presented as footnotes below tables 1-4. Significant differences across ethnic groups were found for urine tests across all years and 2018 in isolation. Significant differences across ethnic groups were found for hair tests across all years and 2018 in isolation.
Chi-square results indicate equal treatment if significant between-group differences exist for both urine and hair testing.
This indicates the groups pass/fail rates are statistically different for urine testing and are also statistically different for hair testing. Irrespective of testing procedure, ethnic groups’ drug test results are significantly different.
CONCLUSIONS
Utilizing independently provided urine and hair pre-employment drug screen data, Researchers were unable to find disparate impacts of hair testing among the ethnic groups analyzed. Results for each test in each sample met the required Four-Fifths Rule threshold.
Chi-square tests independently examine urine and hair tests. Chi-square results indicate that the proportion of drug test failures (positives) are higher for hair testing across all ethnic groups but pass/fail rates are significantly different irrespective of testing method.
Given these findings, Researchers found no disparate impact among ethnic groups by testing method.
[1] Cangelosi and Voss (2019), “An Examination of the Geographical Correlation Between Commercial Motor Vehicle Drivers,” prepared for The Alliance for Driver Safety and Security.