Unfairness Highlights the Fickle Nature of Fiscal Statistics
The Department of Labor is suing Oracle, a software development company, for systematic discrimination against female employees and employees of African American and Asian American decent. On its face, the idea that the Department of Labor (DOL) is suing anyone over discrimination should be seen as a good thing because we want a nation where every person has an equal chance to succeed regardless of their race or biological sex. However, in this case there are several issues that should cause us to take pause.
While companies such as Oracle, Google and Intel are working to hire more female and minority workers, they are also faced with a field that has less workers in those demographics than the actual population. This forces them to make a difficult decision whether to hire someone for demographic purposes or to hire the most qualified worked.
Oracle has long been seen as a stalwart in the fight to provide equal opportunity to all employees. When notified of the lawsuit, Oracle stated it was “in compliance with our regulatory obligations, committed to equality and proud of our employees.” Within the highly complex environment of tech companies, there are increases in compliance/non-compliance issues because of the make up of the field. The concern within the Oracle case is the demographic pool of the tech industry field and not the people working in the field, but within the hiring pool.
On the Forefront of Racial Equality
Demographics is really a numbers game and companies want to keep the best possible employees while giving all employees a chance to advance. Advancing too few of the majority will result in moral problems (and discrimination suits). Advancing too few of the minority could result in being sued by the DOL (and discrimination suits). The make up of the legal regime is devastating to the industry because of the make-up of the field.
According to Gigaom.com, a tech research company, the racial gap in technology is not in hiring, but in the skilled labor pool. Gigaom notes 80% of qualified workers in the United States tech industry are white, which is far above the national average of 60% (white/non-Hispanic). This tends to indicate a training rather than hiring gap. Of the major tech companies surveyed (HP, Ebay, Google, Microsoft, Twitter, Facebook, Intel, Apple, Cisco, Linked In, Pinterest and Yahoo), all had a minority hiring ratio of higher than 20%, which means they are hiring more people from minority groups than the adjusted average in the labor pool. This shows the tech industry is on the forefront of ensuring racial equality in the workplace.
The relationship in the area of gender is likewise skewed. According to Gigaom, the ratio of men to women in the major tech companies is generally about 3:1 with the rates being between 23% female (Cisco) to 42% (Ebay). It should also be noted that the only company reporting transgender employees was Yahoo, whose 1% is nearly 10 times the national population of transgendered persons. With a labor pool that is approximately 25% female, 10 of the 12 companies in the report were showing greater numbers of women employment than the percentage of the labor pool. These augmented numbers tell a much more detailed story than the DOL numbers, which are based on the national demographics (not industry or skill specific). According to Statista, a demographics website, the main causes for dissent in tech stem from being passed over for promotion (22%) and less educated co workers being promoted (22%), the majority of these were from alleged “affirmative action” hires. Conversely, minority workers felt they were stereotyped on the survey questions presented to collect these demographics!
The issue here is not the hiring practices of big tech companies, rather it is the statistical analysis and training of the DOL professionals who are “looking for a case.” Statistics are a great tool for looking at injustice, but only when using the right statistics. In the Oracle case, the DOL Is looking at the wrong statistics and making a case where there is none. The best way to deal with the imbalance of positions in the tech sector is to recruit more heavily into colleges and skill programs for under-represented people. One of the main reasons more white men get promoted is because there are more white men, which just creates a statistical likelihood that white men will have a higher skill level. It is a numbers not a racial game at Oracle. This is why is it important to look at polling, which is one of the government areas where statistics are the most lopsided, and why they fail.
Statistics are an interesting sub-genre of mathematics as seen in a Rasmussen poll of President Trump’s Job Approval released on December 6th . This poll suggested 51% of Americans believed President Trump was doing a good job, 48% believed he was not doing a good job and 1% who were unsure. If you are a Republican or Independent hearing this report, you are overjoyed that President Trump is one of the more popular presidents in recent American history.
If you are a Democrat, then you may have a problem with this statistic. Either way, it is important to look at the methodology used to determine the results. In the case of the Rasmussen report, there were 1,500 likely voters polled to determine the President’s popularity. According to Informary.com, the United States Census Bureau Reports there were 146,311 registered voters from the 2010 Census. Statista.com reports this number grew to 157.6 million voters during the 2016 race. As voter rolls were purged of illegal, decease or felon voters, the number dropped to 153.7 in 2018. Looking at the simple math in this situation, 1,500 respondents in a population of 157,000,000 voters gives us a rate where 1 in every 104,666 voters was polled, which is 0.0001% of the voting population. In orders of relevance, the rate of gun death in the United States (which is ranked 28th in the world), was 4.43 deaths per 100,000 voters – a rate of nearly 5 times that of valid statistical studies on elections. This example demonstrates the importance of paying attention to the statistics and ignore the flawed voting system; but, which do we hear more about the problem’s of?
According to the Google Margin of Error Calculator, if there is a 90% assurance that a sample is heterogeneous, then that sample would provide a result with a margin of error of 1.5%+/-. This is enough to change the results in President Trump’s popularity study. However, according to the UN, the highest fractalization rate in the United States is 0.8242 which would result in a heterogeneity of about 0.18. Some accounts list this as fractured as 0.9 with a heterogeneity rate of 0.1. Further, reports are showing the United States is becoming more fractured and diverse, especially within the voting population and workforce numbers. When the margin of error is calculated with this diversity level of the population, the margin or error expands to between 7.5% and 16.2%, which makes the study statistically irrelevant. This is why statistics are dangerous for a population that is not looking at the source data, only the sound bite that they put on the news.
Department of Labor Study
So, how does this relate to the Department of Labor and their report that there is a wage gap in the tech sector? On October, 15th 2019, the Department of Labor released information that a $5 million settlement had been reached with Intel Corp for discrimination against “Women, African American and Hispanic American” employees. This settlement was reached under Executive Order 11246, which was signed by Lyndon B. Johnson to prevent discrimination in Federal contracting.
The premise of the release was that Intel volunteered to be reviewed by the program for its federal contracting needs and that when found to be outside of statistical compliance, they agreed to pay $5 million in back pay and adjustments to correct their policies. In this case, Intel was being a good corporate citizen and should be applauded; however, the question is whether the settlement was just. When we look at the tech sector, was Intel out of compliance with the industry and was industry, as a whole, out of compliance? According to Vox, “The Gender Wage Gap is smaller in tech, but it will hurt the economy more.” An odd statement to be sure, but Vox is pointing out the coming storm of equal pay lawsuits that are going to rock the tech sector following the Intel settlement.
The foresight of Vox is commendable, but their statistical methodology is not. According to Hired (cited in the Vox article), the wage gap in technology is fairly small, 10 cents on the dollar between White males and Black women. Payscale.com reports that this is $0.11 better than the national average for gender pay gap. This is slightly worse than the institute of Women’s Policy Research, which asserts that the pay gap is still at approximately $0.82 cents for woman for every dollar earned by men. Under analysis from either of these groups, the tech sector should be lauded for the progress it is making, but still understand that women doing the same job as men should be making the same amount as men. Thus, under the Vox simple statistical analysis, there may be a point to be made for the Intel settlement.
However, as with most simple statistics, there is more to the story. According to Marketwatch.com, the pay gap in tech is currently shrinking much more quickly than in other industries. Yes, there is a pay gap in some segments of the industry, but by and large there is a greater progress by the companies in Silicon Valley than in any other industry. One of the reasons for this may not be ill intent, but lack of qualified applicants. According to Forbes, in 2017 53% of pools for jobs in the tech sector only had men meet the basic requirements of the position, 6% of the pools had women only and 41% were blended. This ratio got a little better in 2018, where 46% of the pools were men only, 6% were women and 48% were blended. These numbers paint an entirely different picture of the wage gap in the industry.
Looking at the raw numbers, it is possible the industry is not discriminating against women in pay, rather women are simply not applying to tech jobs. Wage gap economics always has taken a look at the highest paid people to the lowest paid people. One of the key reasons that the wage gap is shrinking is because women are being recognized for their skill and being promoted to high paying executive jobs. This results in the average skewing back towards the center. By and large, the wage gaps are not caused by how much women are paid, but by what position they are holding at specific points in their career. According to FEE (Foundation for Economic Education), the disparity in wages can be readily explained by age, hours worked and marital status – thus, it is not always best to be jumping to the conclusion of ill intent.
FEE Noted Several Characteristics Which Enhanced the Wage Gap:
1) Single men were more likely to work longer hours than married women. The wage gap for single women and single men was less than half of the normal gap.
2) Men worked 2 – 5 more hours a week increasing their take home pay (this amplifies the pay gap because these hours are overtime).
3) Millennial (and post millennial) women are all but eliminating the wage gap, especially in the tech sector.
4) The wage gap is greater in married persons, thus we are seeing larger gaps in married versus non-married (regardless of gender) than in the gender based wage gap.
5) Whether a person has children is seen to greatly affect the hours worked, which in turn, affects the take home pay per person. This is an issue with work-life balance in the home rather than pay at work because companies only can pay people for the work they do, and people who work less are less likely to be promoted.
6) Motherhood was the greatest factor that effected the wage dynamic and more than gender because single fathers also took a hit in pay. As being a mother is a full time job in itself, the ratio opens up research to wage dynamics if the second job was seen as a paying job. Based on hours worked, if mothers only earned minimum wage, the ratio shifts from -6% (adjusted) to a reverse wage gap of 18% in favor of the women.
The argument presented by FEE, which has incredible validity, is that comparing women’s wages to men’s wages is comparing apples to oranges IF, and that is a big if, you only look at gender. When marital status, hours worked, age, and time in position are looked at, the wage gap becomes much smaller. While any gap is a bad thing, using an inflated statistic as propaganda is simply a cry for headlines for the media-industrial complex.
Before the government should attack the tech sector for wage discrimination, there may be a call for the government to look at the court cases that created the environment in which the tech companies are operating. In Philips v. Martin Marietta Corp, the court ruled employers could not use different policies for males with small children and women with small children.
While this was to prevent discrimination based on having children, it created an associated problem of preventing employers from setting up special programs to help women with small children find care for the children during the hiring process. This effectively forces mothers to chose between spending their own money for childcare (which is expensive) or not taking the job (which lowers the average pay of women if it is a high paying job). Even more repressive, is the Griggs v. Duke Power Company Test, which states if the employer uses a neutral test and the neutral test has a result of more men than women, the employer must defend the test. This ludicrous case creates a positive burden of proof on companies that are making an effort to comply with federal programs (which is similar to the Intel case of them being punished for being good citizens).
The courts have also created a “quick hire” culture which negatively effects women. Under the McDonnell Douglas Corp. v. Green Case, the court established a four-part test for discrimination:
1) The Applicant was qualified for the position sought;
2) The Applicant was from a Title VII group;
3) The job was not offered to the applicant;
4) The Employer kept looking for applicants.
Under this test, the employer is punished for not choosing someone because of their status. There are thousands of reasons a person may choose not to hire a person, gender or race being one of them.
As long as the employer did not break any laws in declining to hire a person, the government should not be able to claim racism or sexism just because someone did not get a job. This is why the McDonnell Douglas Test is problematic:
A young woman comes in for an interview. She is qualified for the job. She meets parts 1 and 2 of the test. As she is walking into the building, I see her park in a handicap spot (the near spaces were full and it was raining), she bumps into a guy in the hallway spilling his coffee all over him and does not stop to see if he is burned or apologize, and she is rude to my assistant, acting as if she already has the job. I conduct a professional interview and decline to hire her. The other three applicants in the pool also failed their interview. She can sue for discrimination if the job is reposted, which is not right.
Under these cases, along with several others (look at the EEOC’s list of relevant cases), the government makes it more difficult to hire women in the tech sector because anything but a quick hiring process opens the employer up to a discrimination claim. As seen in the Intel and Oracle cases, even companies trying to be good corporate citizens are being forthcoming. Gender discrimination is a bad thing, but to assault the industry that is making leaps and bounds (not to mention bending over backwards) to eliminate discrimination in the workplace is the opposite of what needs to be done to rectify the situation. If the government must interfere, create a scholarship that helps more women enter the industry. More women in the job pool, results in more women hired. More women hired results in more women promoted. As women achieve the higher echelons of tech companies, the wage gap will dissolve. We have seen it in other industries, now it is time to do it in the tech industry through education, not litigation.
Overall, the Oracle case is an attack on an industry for doing the right thing. Oracle is a large tech company with over 100K employees. With an older worker population (approx. 40% over 50), they have a workforce that was hired when the tech pool was almost entirely white/Asian. As the age declines, so does the number of white male employees. Oracle is hiring people in a pro-diversity method; however, they are not firing qualified workers just to meet quotas. When looking at a situation as complex as the Oracle matter, the direct demographic information of the country will simply not cut it. Professionals need to look at professional work pools, marital status, hours worked and other factual statistics before determining if a company is in compliance or not. The Department of Labor does not appear to be compliant with this need, thus their methodology is questionable at best.
New Right Network depends on your support. Donate now