Maven Family Leave Challenge

In this project, I worked as a Data Visualization Specialist at Nomics, an online business journal. My role was to create charts, visuals, and infographics as supporting content for articles.

This particular article focused on Parental Leave Policies across the business world, and this impactful visual has been created as a supporting content, to mark Women’s History Month in the United States.

My aim as a Data Visualization Specialist was to make readers of this article understand the content without having to read the article itself. As they say, “a picture is worth a thousand words”.

The Dataset:

The dataset (provided by Maven Analytics) contains a table in CSV format with 1,601 records, one for each company.

• Each record contains the company's name & industry, as well as crowdsourced information on the paid & unpaid weeks off they offer as part of both their maternity & paternity leave policies (when available)
• Tags: Business, Human Resources, Leave Policies
• Data Structure: Single Table
• Number of Records: 1,601
• Number of Fields: 6

Approach:

The idea I had was to build a news-like page, akin to what you can find on the news pages of Wall Street Journal, Financial Times, McKinsey, etc.

The idea was to make readers of this article understand the content without having to read the article itself, using impactful visuals, charts, and infographics that fully depict the article content.

Step 1: Data Cleaning & Aggregation

• CountBlank function was used in MS Excel to ascertain missing data (if any).
• The data was quite clean. What was needed was to manipulate the data to fit the brief.

Step 2: Data Manipulation

• Missing Data:** Some of the fields (especially “Paid Paternity Leave” and “Unpaid Paternity Leave” contained missing data (N/A).
• This type of missing data is termed Missing At Random (MAR): there is randomness in missing values within the male (paternity) and female (maternity) subgroups.
• The causality of the demography leading to missing data cannot be proved in most cases, as we do not have an exhaustive set of variables affecting the decision to fill the survey.
• The tools and techniques for dealing with “N/A”.

a.Mean: used when distribution is near normal.

b.Median: used in case of significant presence of outliers in the data.

c.Mode: used for categorical variable.

• As a user-reported data, where users report conflicting information, consensus numbers (if any) or the median are shown.
• So, initially, I used the median numbers for “Paid Paternity Leave” and “Unpaid Paternity Leave” fields to replace the “N/A”s (using CTRL + H).
• Median for “Paid Paternity Leave” = 6.
• Median for “Unpaid Paternity Leave” = 6.
• But this approach distorted the distribution.
• After a second look at the distribution, I concluded that since “N/A” means “no information has been reported”, then the companies had zero leave policy. In other words, instead of using the “median” approach, “N/A”s were replaced with “0”.
• Bands were then created for the fields in the following categories:

a.No paid leave (or unpaid leave)

b.Less than 14 weeks

c.14 – 25.9 weeks

d.26 – 51.9 weeks

e.52 weeks or more

Step 3: ETL

Step 4: Data Transformation and Field Categorization in Power Query

Step 5: Data Modelling

Step 6: DAX creation (Maximum and Minimum Weeks for Paid Maternity Leave, Unpaid Maternity Leave, Paid Paternity Leave, and Unpaid Paternity Leave)

Step 7: Analyses and Dashboard (news page) Design

Challenges:

Major challenge was in trying to understand how best to deal with “N/A” as well as how best to create minimal charts and visuals that will adequately tell the story.

Key Insights:

• 3% of the companies offered no paid maternity leave, and just about 1% offered 52 weeks or more paid leave. 73% of companies offered less than the 14 weeks advocated by the ILO.
• 84% of the companies offered no paid paternity leave, and only 3% of companies offered 14 weeks or more paid paternity leave.
• 373 (23%) out of 1,601 companies are compliant with the ILO paid maternity leave policy. Only 20 (5%) of these 373 companies offered 52 or more weeks of paid leave.
• 140 industries (75%) out of total 186 industries offered between 0 and 12 weeks paid maternity leave. Bus Transport, Telemedicine, Philanthropy, HR, and County top the list of family-friendly industries.
• Telemedicine, Venture Capital, Gaming, Tobacco, and Education top the list of father-friendly industries, and 80% of these offered less than 10 weeks paid paternity leave.
• Only 17% of companies provide paid paternity leave which is 88% lower than maternity leave. 82% of companies had no data regarding paternity leave.

Recommendations:

· Infant heath is best supported by exclusive breastfeeding for six months (WHO recommendations based on research evidence). The best way to support this is to provide at least six months of paid maternity leave.

· Where this is not possible, a minimum of 18 weeks of paid maternity leave should be provided (consistent with current ILO recommendations).

· Fathers should be provided with paid paternity leave of adequate length to support bonding with the infant, establish a role for the father in the care of the child, and support children’s health development and gender equality. There is substantial evidence that paternity leave increases a father’s involvement, reduces gender inequality, and benefits both infant and maternal health.

· Total paid parental leave (maternity, paternity and parental) should be long enough to ensure access to all preventive care and to ensure high-quality infant care at least until the age at which affordable, quality non-parental care is available. This should be at least 6 months and, in many settings, should total 9–12 months.

·Paid parental leave should be structured to better cover the informal sector.