DATA VISUALIZATION

Reviewing Coded Bias: When Data Meets Politics

When it comes to stories of data usage and how they come in different forms in today’s modern world, I usually am not surprised by what I hear. After watching this film, I now believe that I feel differently. Watching Coded Bias has introduced me to new ways in which we see technology harming us today, especially as these ways are mixed in with human error and human prejudice and biases. As we’ve learned in class, one of the most important lessons we’ve known yet is that data is influenced by what we decide to include in it as well as what we exclude. While many of us have thought as data as an ever-reliable source with no flaws, helping to create and organize data in class has shown that data can oftentimes be inaccurate and also be warped by personal or societal biases.

As the main hero of the film, Joy Buolamwini was an incredible character to watch grow and progress and she questioned the systems around her. When she had discovered that facial recognition systems only recognized her while she wore a white mask, she was able to quickly understand why that was and establish that the artificial intelligence systems we live with are, in fact, corrupt. After speaking with tenants in a Brooklyn neighborhood about their complains about artificial intelligence cameras for recognition, Buolamwini decided to take the complains to Washington, D.C. and form it into an official case to present in from on the Supreme Court. One of my favorite parts of the film was the ending; it was a fantastic comparison to show where Buolamwini had been and where she was looking to go. More than anything, it was fascinating to see data being presented in such a legal way that you would have never expected in the Supreme Court. One of the largest issues researchers like Buolamwini have been attempting to address is that there is barely any governmental regulation on using artificial intelligence, especially when it comes to facial recognition. I was very impressed by how the information was presented to the Court in an understandable way, especially when it affects ordinary people, not just politicians. It also comforted me to know that, at least in the film, both Conservative and Democrat representatives were able to recognize that using artificial intelligence dangerously is bad for Americans. This was a rare moment where both sides of American politics could easily agree on one issue. Overall, I was extremely pleased to see the case being taken to court, to the point where ordinary people being affected have a voice in the issue as well.

When it comes to involving politics with facial recognition, I’ve also noticed that researchers will attempt to involve this topic for when they’re collecting data. While it’s been a pressured issue of governments not properly regulating artificial intelligence, we have also seen examples where the government will take advantage of artificial intelligence tools in order to get what they want. An example Buolamwini had presented was how the government in Hong Kong was using facial recognition tools to track down protestors or opposition to the government. I found this fascinating, especially since protestors combatted this by wearing masks or attempting to dismantle facial recognition systems in solidarity. Buolamwini had used this an inspiration for her own research, especially after also hearing how similar A.I. systems in Hong Kong have earned citizens statuses of being a “good” or “bad” citizen. The less wrong things you do, the better credit you have – it also means you’re less looked down upon in society. It made me interested in how the government was using data to create “identity politics” as we’ve taloked about before. Because of this, we see erasure of unique identities and instead see people falling into normalization of one type of person. While it is incredibly interesting, I’m curious to know more about how Buolamwini used this in her research as it might present some limitations, especially as Hong Kong has faced separate issues that you maybe don’t see as much in the United States.

While I think using data in such a way as used in protests in Hong Kong, I remember hearing about the government using facial recognition and remembered that the police had been using similar methods of tracking down protestors during the height of the Black Lives Matter movement in 2020. While I’m not sure about A.I. being used, I still believe that we see cases like Hong Kong everywhere, and all occasion display using organization of information to track down those whom they deem as the opposition. While protesting myself, I remember having to minimize what I look like, especially by wearing all black and a mask and covering up any tattoos. This information would then be passed around to fellow protestors, and it’s made me wonder about how much we are afraid of facial recognition and how it can be used against us since the government has not regulated it at that time. While this is not necessarily or directly related to data, it does show how we, as a large society, believe that there is fear in technology, as well as in fear in reliance or trust in the government.

All aspects of the film were interesting, but I especially loved getting to see the more activist or human side of it and how facial recognition and using data can come into play. It’s been an opportunity to see where data is used as a weapon, which is unfortunate to see being done when data can be used to truly help people. This film has mainly allowed me to understand how precious but dangerous of a tool data is, which is why it’s important to know about biases that come within data and how a single data point can change your very own perspective.

The MBTA Salaries Data User Guide

By Eva Levin, Nena Hall, and Natalie Vasileff

Original purpose and application

This User Guide for the Massachusetts Bay Transportation Authority (MBTA) data is meant to provide the audience with information about the large pay discrepancies within the MBTA between workers, managers, and directors. In 2021, 350 full-time employees made less than the Federal poverty line, and 170 of these employees were train operators and bus drivers.

The data set that will be used is collected from the Statewide Payroll, which collects the payroll data each year for various departments within the state. It is used to document state workers salaries through paychecks and the total amount of money they earned that year. This User Guide will be looking at the payroll information for 2021.

History, standards, and format

The Commonwealth's Financial Records Transparency Platform (CTHRU) is an open records platform that offers transparency within statewide spending, payrolls, and revenues. It created the Statewide Payroll, which has collected this version of the data format of payroll data from the calendar year 2010 to current, and is updated bi-weekly during the current payroll year.

With the structure of the data, the website allows the viewer to look at the information through paychecks, or through a graph that allows you to toggle employment type and pay type. The graph provides different ways to visualize the data depending on what the viewer is looking for.

The dataset is based on the fact that not all payroll expenses are fulfilled solely through public funds, but private funds as well. When looking to see which organization collected the payroll data before 2010 was difficult to find, and no information was available.

In 2018, software developed from Tyler Technologies was implemented into the Statewide Payroll system when it was moving the Boston Logan Airport payroll into the system. Since then, Tyler Technologies has still been the software used for the Statewide Payroll data system.

Organizational context

CTHRU is a program run by the Office of the Comptroller for the Commonwealth of Massachusetts (CTR). The current Comptroller is William McNamara, appointed by former governor Charlie Baker in February 2021. His term ends in 2027. CTR oversees $60 billion dollars in funding every year, mainly from governmental agencies. CTR employs 135 full-time employees, only some of whom work on the CTHRU program.

The most recent audit for CTHRU was completed in 2022, and found that there were no significant instances of noncompliance in the office of the Comptroller. CTHRU was launched in 2016. Before then, the data was displayed on a different program called Open Checkbook, which launched in 2010.

Workflow

This data from the CTHRU is sourced from transactions reported by the statewide accounting system, Massachusetts Accounting and Reporting System (MMARS). The data is automatically published to the platform through a set of automated data processes from the Commonwealth’s information warehouse. The payroll data used to supply the data sets is from the Office of the Comptroller’s Human Resources/Compensation Management System. The data is a product of the Office of the Comptroller.

The Commonwealth’s fiscal year runs from July 1st to June 30th, and the data updates annually to reflect the year’s payroll. The data is built on complete payroll data from previous years, but stands alone as individual data sets per year, if the viewer wishes.

Tyler Technologies, the software company used to protect and provide the Statewide Payroll data sets, aids in visualizing the data as well as presenting it in a sheet form to make it accessible to all kinds of viewers. Tyler Technologies reports that CTHRU has 400,000 views per month, with 20,000 unique viewers. In addition, the program has increased media inquiries in statewide payrolls.

CTHRU is used for audits of the state at-large, as well as auditing specific departments and segments of the Commontwealth’s government to determine if the state is properly paying their employees.

Exploratory Visualization/s of the Data.

1. Full-time employees that made under the federal poverty line:

Over 350 full-time employees at the MBTA were paid less than the federal poverty line in 2021. Around 20% of those people were surface operators. This image visualized the top ten positions at the MBTA paid less than $14,000, the federal poverty line.

2. Positions that earned the most overtime pay in 2021:

In 2021, the position to accumulate the most overtime pay was surface operators at a total of $13,102,845.

3. MBTA Employees and the Contracts They’re Most Attached To

Under certain contracts, MBTA employees work for them under their labels as they continue to provide employee funding. This can become important for limits on information regarding workers. Most employees came from Carmen's Non-Miscellaneous contract.

Things to know about the data, including limitations

As previously mentioned in this guide, plenty of payroll expenses are paid through private funds. This means that this information is incredibly difficult to find, especially in earlier years.

There is plenty of data in this set which continues to be ambiguous or does not provide the full picture of what it is like to be an MBTA employee. For example, as seen in the chart above, most employees are attached to a certain contract which then assists in providing money. These contracts follow their official labels, often having words such as “alliance” or “miscellaneous.” The lack of clarity in these labels limits the importance of the roles of those that work for the MBTA. Once again, this ties in with large amounts of money coming from private funds instead of public ones.

This data also does not show how employees work. For example, some contract employees have regular schedules, but work intermittently and don't have a "salary." This then labels a large part of salaries from employees as $0.00.

Additionally, there are very few ways to track how payment is received to employees and if their salaries are meeting the correct standard, especially since most payments go through a company like Tyler Technologies. As the company audits specific parts of certain departments, there is little public handling in where funds are going.

When it comes to more personal or communal; details regarding MBTA workers, no data mentions any injuries or accidents which are work-related. This is important as the number of injuries can affect how much an employee might be paid.

Overall, the largest issue which creates a gap in the provided data is that a large number of funds comes from private companies which most would not know about. This limits the data that we can see and also does not provide a full picture as to why MBTA employees earn so little.

Supplementary Information

The links included in this section give additional information to the Statewide Payroll dataset and the organization that collects it, along with more information on the MBTA in regards to its budgets and audited financial statements.

The website of the CTHRU has a frequently asked questions section on the bottom of it, which helps to answer the information source, why some salaries are valued at $0, and object codes included in ‘Other Pay.’
The CTHRU also has a glossary of the terminology used for state finance and payroll.
For the MBTA salary dataset, there is also information about the MBTA budgets and audited financial statements.

Authors of this Data User Guide

Eva Levin
Nena Hall
Natalie Vasileff

Source Log

1. Michael V. Sangalang, 617-973-2672, michael.v.sangalang@mass.gov

2. Lisa Battiston, LBattiston@MBTA.com

3. Joe Pestaturo, JPesaturo@MBTA.com

4. Matthew Petersen, 541-971-8436, mpetersen@transitmatters.info

Examining Data Sets: Employee Earnings in Boston, 2017

Who collects this data set? If it's an organization, which department of the organization? Is there a specific person listed who you could contact?: The Department of Innovation and Technology releases this data set every year along with specific names of employees and what they each earn. In the section of the report, there is a contact email of the department. However, there is no official name that you can contact directly. In this situation, it might be possible to contact specific directors/heads of different departments in order to obtain any employee earning information if they are willing to give it up. Sometimes, some departments may release annual earnings reports which can be used to double-check what employees earn each year.
Why do you think the organization collects this data? Does it specify how it uses the data?: It’s important to know what employees earn and if the amount of money they make changes from year to year. It’s also necessary to know if some departments or organizations earn more than others. For example, I personally think that it’s strange that Boston Police Department employees earn more than most employees, especially by such a large amount.
What time period does the data set cover?: This covers the year of 2017, but it is also possible to look at employee earnings reports from any year.
What are some questions you have about this data set? (Note: they can be basic like "why is this data being collected?" or very specific like "what does the field BUS_LIC_STATUS mean?": I don’t necessarily have any questions, I feel like the data being presented is quite clear and also has good reasoning for why it is being collected.
Who are three types of people you could interview about this data set in order to learn more?: I would interview the department manager/head of Innovation and Technology. I would also see if I could speak with someone from Analyze Boston to see how they have received the data in the first place. I also think it would be interesting to interview someone from the Boston Police Department (someone who earns a lot) and then follow with an interview with an employee from a department like the Office of Tourism where employees earn very little.

My First Chart: Examining Deaths of Bird Species at Logan Airport

Final progress photo of final data chart showing the number of sparrows, owls, and sandpipers species (most common types of birds) which were killed by flight activity at Logan Airport.

While data coming from the Logan Airport regarding bird deaths contained plenty of various categories, I wanted to specifically look at which types of birds were most commonly hit during certain months and what species those types contained. The data set which was first introduced included many different categories, including what section during flight time birds would typically be hit as well as what damages an airplane had faced from hitting a bird amongst many other categories. While all of this data is important, I only wanted to specifically look at bird species.

While the number of bird deaths at the airport is, in reality, quite small, there is a large variety of different species. In order to condense the data, I had to filter out certain categories an

in order in order to group certain species together into less repetitive categories. Therefore, I used a wildcard filter through Tableau to look at all of the species in these categories: sparrows, owls, and sandpipers. These are some of the most common categories which experienced deaths each month. The progress photos above show how, eventually, I was able to condense categories and make the data set smaller and more understandable.

Explaining the People’s Economics: Boston, MA vs. Scranton, PA

Boston has always been a prominent city - one that is well-known for its successes, traditions, and unique culture. While a city like Boston has always been around, smaller cities are increasingly growing, including the small city of Scranton in northeastern Pennsylvania. Still, with growing cities come growing statistics and data. Even though Boston is much larger and more economically developed, it is still necessary to look at comparisons between bigger and larger cities.

Both East Coast cities share plenty of similarities when it comes to data collection, but there are still some notable differences which set the cities apart. One of the key identifiers which make up each city’s character is what the people living inside of them are like, especially economically. Boston and Scranton differ largely when it comes to how people work professionally, what they earn financially, as well as the upbringings which have led them to the present economic positions they are in.

1. The Poverty Rate

Generalizations in data are always dangerous in any capacity. For example, when we think of higher poverty rates, we usually associate them with more urban areas. Despite this assumption, Boston’s poverty rate is 18% while Scranton’s poverty rate has jumped to 21.8%. An incredibly important piece of information to include is that Boston’s population is around 670,000 while Scranton’s isn’t even a sixth of that. According to a report made by a website called Welfare Info, the poverty rate in Scranton is around 57-70% higher than that of the national average. The issue here is not that Scranton is smaller, but that poverty exists more inside of it than we might realize, especially when compared to a larger city like Boston.

2. Veterans

Groups of people experiencing poverty and/or homelessness also vary by city. Typically, when Americans associate homelessness or poverty with a specific group of people, minds immediately draw to veterans as a stereotype in being unsheltered. However, after looking at a few reports, including a report made by the National Alliance to End Homelessness, few unsheltered persons in Pennsylvania are veterans. This can appear as surprising to some as the veteran population in Scranton, as well as in the rest of Pennsylvania, is much higher than that of the national average, further mentioned in data collected by DataUSA. Boston has a veteran population of about 13,7000, but it is still lower than that of the national average. In fact, the population of Boston alone is still less than the number of veterans living in the entirety of Pennsylvania which is 800,000 veterans.

3. Education

Assumptions regarding homelessness also can point to factors such as an access to education which can then further an individual’s chances of being offered a job. Boston is regarded as a well-educated city, which makes sense. As education-related data are collected on a state basis, just in 2020, 1.19 million people have graduated with a high school diploma. The same data from DataUSA shows that 1.3 million people then went on to graduate with a Bachelor’s degree, exceeding the state-wide number of high school graduates. As Boston is filled with more prestigious, private institutions, most people in Boston at least are living in an area where there are plenty of schools, therefore Bostonians are already experiencing an easier access to education, even if it’s not necessarily affordable for everyone. Scranton sees a different story; once again, on a state-wide basis, around 3.14 million Pennsylvanians in 2020 had graduated with a high school diploma. This number varies drastically from the statistics of Pennsylvanians earning a Bachelor’s degree which remains at 1.8 million. While Scranton has a few schools in the area, they are still not as prestigious as those located in Boston.

4. Household Income

A last large factor which can help identify a Bostonian or Scrantonian economically is the rate of household income. The national average of household income lies at around $67,000-$70,000. Boston reaches above the national average, ending with a household income rate of $76,298, with a number of almost 300,000 households. While Scranton has the smaller number of households, remaining at around 31,000 households, Scranton’s household income is just above $41,000. This rate is much lower than the national average.

While data cannot show all of the aspects which contribute to economic identities in both Scranton and Boston, what we can see is that there are still large gaps in economic equality, especially when comparing both cities to the national average. We also see large gaps in who has access to certain resources and how individuals can build their economic positions.