Introduction
As I begin to shift into a career in the tech sector, I have found the more day-to-day considerations of the working environment to be of interest. While the ‘tech sector’ is arguably too broad of a grouping, generally these fields are undeniably cushy, often associated with high prestige or compensation, especially compared to working in a number of other fields. However, ultimately, tech workers are workers whose livelihood depends on performing certain tasks at the whim of an employer. This can often have an understandable impact on mental health. We can take a look at mental health survey data in the tech field at a micro level, before zooming out into a broader industry comparison to get a better idea of this issue.
The Data
Primary Dataset
The primary dataset I will be working with is the OSMI Mental Health in Tech Survey. This is a dataset created based on a survey done by Open Sourcing Mental Illness. It was filtered to only includes responses from those with a primary tech role. The survey is from 2014 and includes 1,259 responses from 48 different countries, though the vast majority are from the United States, the United Kingdom, and Canada.
The following are the main variables used in the analysis:
no_employees (Categorical): The size of the respondent’s employer, ranging from “1-5” to “More than 1000.”
benefits (Categorical): Indicates whether the employer provides mental health insurance coverage (Yes, No, or Don’t Know).
treatment (Categorical): Indicates whether the worker has sought professional treatment for a mental health condition (Yes, No).
wellness_program (Categorical): Indicates whether the employer has ever discussed mental health as part of a formal employee wellness program.
seek_help (Categorical): Indicates whether the employer provides specific resources or documentation to learn more about mental health issues and how to seek help.
tech_company (Boolean): A filtering variable indicating whether the employer is primarily a technology company or organization.
Data summary
| Name |
Piped data |
| Number of rows |
1259 |
| Number of columns |
27 |
| _______________________ |
|
| Column type frequency: |
|
| character |
25 |
| numeric |
1 |
| POSIXct |
1 |
| ________________________ |
|
| Group variables |
None |
Variable type: character
| Gender |
0 |
1.00 |
1 |
46 |
0 |
47 |
0 |
| Country |
0 |
1.00 |
5 |
22 |
0 |
48 |
0 |
| state |
515 |
0.59 |
2 |
2 |
0 |
45 |
0 |
| self_employed |
18 |
0.99 |
2 |
3 |
0 |
2 |
0 |
| family_history |
0 |
1.00 |
2 |
3 |
0 |
2 |
0 |
| treatment |
0 |
1.00 |
2 |
3 |
0 |
2 |
0 |
| work_interfere |
264 |
0.79 |
5 |
9 |
0 |
4 |
0 |
| no_employees |
0 |
1.00 |
3 |
14 |
0 |
6 |
0 |
| remote_work |
0 |
1.00 |
2 |
3 |
0 |
2 |
0 |
| tech_company |
0 |
1.00 |
2 |
3 |
0 |
2 |
0 |
| benefits |
0 |
1.00 |
2 |
10 |
0 |
3 |
0 |
| care_options |
0 |
1.00 |
2 |
8 |
0 |
3 |
0 |
| wellness_program |
0 |
1.00 |
2 |
10 |
0 |
3 |
0 |
| seek_help |
0 |
1.00 |
2 |
10 |
0 |
3 |
0 |
| anonymity |
0 |
1.00 |
2 |
10 |
0 |
3 |
0 |
| leave |
0 |
1.00 |
9 |
18 |
0 |
5 |
0 |
| mental_health_consequence |
0 |
1.00 |
2 |
5 |
0 |
3 |
0 |
| phys_health_consequence |
0 |
1.00 |
2 |
5 |
0 |
3 |
0 |
| coworkers |
0 |
1.00 |
2 |
12 |
0 |
3 |
0 |
| supervisor |
0 |
1.00 |
2 |
12 |
0 |
3 |
0 |
| mental_health_interview |
0 |
1.00 |
2 |
5 |
0 |
3 |
0 |
| phys_health_interview |
0 |
1.00 |
2 |
5 |
0 |
3 |
0 |
| mental_vs_physical |
0 |
1.00 |
2 |
10 |
0 |
3 |
0 |
| obs_consequence |
0 |
1.00 |
2 |
3 |
0 |
2 |
0 |
| comments |
1096 |
0.13 |
1 |
3548 |
0 |
159 |
0 |
Variable type: numeric
| Age |
0 |
1 |
79428148 |
2818299443 |
-1726 |
27 |
31 |
36 |
1e+11 |
▇▁▁▁▁ |
Variable type: POSIXct
| Timestamp |
0 |
1 |
2014-08-27 11:29:31 |
2016-02-01 23:04:31 |
2014-08-28 02:30:00 |
1246 |
Supplementary Data
To complement this data, I used data from the Current Employment Statistics (CES) and the Job Openings and Labor Turnover Survey (JOLTS) accessed through the BLS API. It should be noted that in this data, I have used the Information sector as the counterpart to Tech.
The following are the series used in the analysis:
CEU5000000011 (Avg Weekly Earnings — Information): Average weekly earnings for the Information sector, representative of the tech industry
CEU6000000011 (Avg Weekly Earnings — Professional & Business Services): Earnings comparison for a somewhat similar industry
CEU7000000011 (Avg Weekly Earnings — Leisure & Hospitality): Earnings comparison for an industry less similar
JTS510099000000000TSR (Total Separation Rate — Information): The percentage of Information sector employment that separated from their jobs each month, includes quits, layoffs, and discharges
JTS540099000000000TSR (Total Separation Rate — Professional & Business Services): Separation rate for a somewhat similar industry
JTS000000000000000TSR (Total Separation Rate — Total Private): Economy-wide baseline separation rate
Data summary
| Name |
Piped data |
| Number of rows |
720 |
| Number of columns |
6 |
| _______________________ |
|
| Column type frequency: |
|
| character |
3 |
| logical |
1 |
| numeric |
2 |
| ________________________ |
|
| Group variables |
None |
Variable type: character
| period |
0 |
1 |
3 |
3 |
0 |
12 |
0 |
| periodName |
0 |
1 |
3 |
9 |
0 |
12 |
0 |
| seriesID |
0 |
1 |
13 |
13 |
0 |
10 |
0 |
Variable type: logical
Variable type: numeric
| year |
0 |
1 |
2020.50 |
1.71 |
2018.00 |
2019.00 |
2020.50 |
2022 |
2023 |
▇▃▃▃▃ |
| value |
0 |
1 |
19286.94 |
44035.88 |
15.72 |
44.16 |
1470.49 |
15992 |
157950 |
▇▁▁▁▁ |
Data summary
| Name |
Piped data |
| Number of rows |
216 |
| Number of columns |
6 |
| _______________________ |
|
| Column type frequency: |
|
| character |
3 |
| logical |
1 |
| numeric |
2 |
| ________________________ |
|
| Group variables |
None |
Variable type: character
| period |
0 |
1 |
3 |
3 |
0 |
12 |
0 |
| periodName |
0 |
1 |
3 |
9 |
0 |
12 |
0 |
| seriesID |
0 |
1 |
21 |
21 |
0 |
3 |
0 |
Variable type: logical
Variable type: numeric
| year |
0 |
1 |
2020.50 |
1.71 |
2018.0 |
2019.0 |
2020.5 |
2022.0 |
2023.0 |
▇▃▃▃▃ |
| value |
0 |
1 |
3.86 |
1.39 |
1.9 |
2.5 |
3.8 |
4.9 |
10.8 |
▆▇▁▁▁ |
Employer Support, Size and Worker Health
It is important first to verify the connection that might be obvious for some, whether workers even utilize the resources provided by their employer to any significant extent. Generally, we can see that accessible professional care, provided by employers, does have a connections with the workers willingness to seek help. It should also be noted, however, that a major hurdle in this is the frequency of “Don’t Know” responses and the corresponding low frequency of employees who seek help. As such, we can see that the employers actions in terms of funding these programs, but also effectively communicating the existence of the these programs through administrative efforts are tied to the mental well-being of the workers ultimately reliant on them.
Having seen the potential connection between worker mental well-being and their employer, we can start to look at the employers willingness to provide any sort of support. This trend follows what one would predict, even if workers needs and reliance on their company remain the same, smaller employers are less able to provide the necessary protections for their employees. Ultimately, these ties to the size of the employer confirm that the mental health of the worker is another asset to be managed according to the decision making of the employer, rather than a baseline necessity.
Even when certain benefits are provided at the technical level, workplace culture can still be hostile towards properly dealing with these issues. This visualization compares the existence of insurance benefits and how easy it is to take medical leave at a workplace. We find that even in places that do clearly offer benefits, employees still often are uncertain or find it difficult to take medical leave. In the end, it seems that even when employers do offer certain benefits, work culture makes it so that worker mental health is often still at risk and up to the decision-making of the employer.
Industry Level Comparison
It is also critical to consider the general position of the Tech industry to gain a better understanding of the quality of health among the employees. Even among other similar industries, tech stands out when it comes to earnings. However, it is clear from the survey data that this doesn’t exactly translate to better mental health support. In fact, as we will see in the next visual, it can often obscure it.
Clearly, tech consistently has the lowest separation rates compared to other fields. This is despite the data we see from the survey, where there are issues with employer support and benefit awareness. Paired with the previous earnings data, where we see the significant financial incentive tech workers enjoy, we could interpret that wage is negating the negative effects made apparent in the survey, allowing for higher retention rates in the Tech industry.
Conclusion
This analysis has made clear that there are significant gaps with regards to the approaches to mental health in the tech industry. Despite having some of the highest wages in the industry, which would seemingly demonstrate a wealth of resources that could be dedicated to well being, a large number of tech workers don’t know whether they have mental health benefits, struggle to take medical leave even when benefits exist, and report that mental health regularly interferes with their work. Even then, they leave their jobs a much smaller rate in comparison to national averages and other, similar, industries. The most obvious explanation is that the high wages compensate for the lack of any resources or culture aimed towards support. This can illuminate a worrying situation, where workers are overlooking serious underlying issues which are being masked by the understandably glamorous wages. It should be noted that this analysis cannot be comprehensive due to a number of limitations like a mismatch of time periods and the inherent inability to quantify certain conditions that might lead to mental illness. Nonetheless, the data points to tech workers existing in a field that incentives certain poor conditions for mental health but covering these issues by its nature of being one of the higher-paying industries.
