Industry Maternity Leave Analysis

Project Brief and Approach

The objective or brief for this project was to "create an impactful visual" to be used as part of an online business journal on parental leave policies.

My own personal objective was to attempt this challenge using Tableau. It is a tool I have recently become reaccustomed to, and is my first Tableau challenge submission since the Mexican restaurant challenge back in September 2021!

On reading this brief, I focused in on a few things, which I summarised in a LinkedIn post on the subject. I was of the opinion that this was looking for a "visual", which I interpreted as a single or small set of charts that could be used to complement an article being written on the subject of parental leave. This may be similar to visuals you see in publications like WSJ, Time or the Economist.

As such, the I felt the visual should be easy to interpret, succinct and focus on a particular aspect that could be used as a reference point for a wider discussion.

Potential Avenues of Analysis

Looking at the data dictionary, prior to examining the data in detail, I could see there was the potential to examine aspects such as:

• maternity vs paternity leave,
• paid vs unpaid leave
• context of industries or
• individual companies.

I was also thinking whether I could look at the location of individual companies, but later saw that would be difficult to assess as many companies were multinationals and it would be difficult to ascertain whether the leave allowances were global or restricted to particular countries.

Data Review

Reviewing the data, I found that <20% of companies had relevant information on paternity leave. As such, I made a quick decision to not go with a male:female comparison.

Therefore I decided to look at maternity leave, and focus on examining paid leave, as it:

• was the most complete data
• would likely be of most interest to readers
• gave insights into company and industry "real monetary benefits" provided to female employees

Industries

Next, I decided to review the paid leave allowances in the context of "main industries." Individual companies and even the detailed Industry category provided in the data set were too granular for a more high level analysis.

For example, there are 185 Industries listed in the dataset, but when you split out the "subindustries" noted after the colons, it reduces the number of main industries to 51.

No. of Companies

The number of companies in each main industry varied from 1 to 327. Industries with very small numbers of companies would likely provide skewed or unreliable insights when compared against those with large populations, so with my main brief objective of providing a succinct and focused analysis, I opted to look at the top 10 industries by number of companies.

Analysis

As the number of companies in each industry was quite large, and as the paid leave weeks were varying from 0 to 52 weeks, it was not appropriate to simply look at an average or median value for each industry. I wanted to see what the distribution was, looking at the overall spread, IQR and potential outliers, and see how I could rank each industry.

Visualisation Method

In my mind I was thinking of using either a violin plot or box and whiskers plot. A violin plot may have been more aesthetically pleasing, but I feel the box plot is more clinical in identifying the key points of the median value, the IQR (or middle 50% of values) as well as the outer extremes and outliers. As such, I opted for that.

Visual Aesthetic

Graphics

As the visual was for a journal, I opted to include some supporting graphics and structure to give a similar feel to that you would see in a magazine. It was important that the picture was not distracting from the data or impeding it's understanding.

The colour palette to apply to the visuals was selected to complement the background picture and provide some understated accenting.

Form

I initially was thinking to style the box plot almost as a mobile you would see above a baby crib (see below). However on review, personally, I found it easier to see the difference between median and IQR values when reviewing vertically, so I swapped the orientation.

This also allowed a better continuity across the page between the box plot and the BANs on the right hand side.

Layering and Context

I kept individual companies shown in the box plot in the form of a jitter plot, and made the points sufficiently small enough as to not distract, but provide that additional layer of information that you can dig a little further into.

I added a few additional context points such as the number of companies and a % of companies that provided paid maternity leave in excess of 24 weeks (~6months) just to give an additional insight into the propensity of companies with very generous allowances.

Supporting Explanation

The main title and subtitle provided some high-level take-aways of the analysis to point the reader in the direction of what to initially look for.

Separately, as box plots are not a very common visual, I provided an explanation on how to read the main points of the box that were important in the context of the analysis.

Final Review

After a few iterations of the final layout, I was happy that I produced something I feel fulfils the brief and that could be used to support an article written in a journal.