Is There a Doctor in the Building?
A data exploration and analysis of a company’s hiring practices and trends
Scenario:
I was given a data set by a company in the Denver Tech Center. This company is in the higher-education industry and has a preference for candidates that have graduate degrees. My contact there works in the HR department and they gave me a snapshot of approximately 15 months of their data from 2017–2018. My contact mentioned that their company, which I will refer to AI going forward, have had a problem with Doctoral candidates progressing in the interview process.
I had to do quite a bit of data clean up, mostly with removing columns, turning strings into integers so I could use the salary information and I also change the types of Bachelors/Masters Degrees to just “Bachelors” or “Masters”, etc. For example, candidates who listed they had a Bachelors of Science in XXX was changed to just “Bachelors.”
My initial thought is that the Doctoral candidates required a higher salary in comparison to other applicants with a lesser education level, thus causing them to filter out of the interview process because of their starting price point. Let us dig into the data now!
Here, we have a look at the distribution of candidates and their education level.
Doctoral candidates pale in comparison to candidates that hold Master’s/Bachelor's Degrees. But wait, wouldn’t that make them a more distinguished applicant? At this point, that was my thinking, but I wanted to take a look at the salaries each applicant stated they were seeking when they applied to AI.
There were no surprises here. The Doctoral candidates average requested salary at AI was just slightly ahead of those who hold Master’s and Bachelor’s degrees. We can explore the current status of current applicants to see where they stopped in the interview process. Maybe that will tell us more about Doctoral applicants.
The Orange in each column shows that a candidate was ‘hired’ and I can’t tell from the graph if there have been any Doctoral hires. It appears that all of the hires are from Bachelor’s/Master’s degree applicants. Let’s zoom in on the Doctoral candidates to be sure.
It appears that we have one Doctoral applicant that declined, one that withdrew and the majority were discarded due to ‘not a fit’ or ‘not qualified.’
At this point, it doesn’t make a lot of sense that so many Doctoral candidates are not qualified for the same positions that Bachelor/Master’s degree applicants are getting hired for. I wonder what kinds of positions the applicants are applying for?
Research Analyst is, by far, the most popular type of position that AI. For my own curiosity, I wanted to take a look at which position had the highest desired salary.
It appears that ‘Sales Consultant’ and ‘Program Manager’ have the highest desired salaries by applicants. Again, nothing I have found in AI’s data seems out of the ordinary. I do think it seems a little strange that no Doctoral candidates have been hired, but they are also not as prevalent as Bachelor/Master’s degree applicants. I wanted to dig in further to the salary aspect, as I still think this could be of some importance. I went to Glassdoor.com and found an additional data set of salaries in other metropolitan cities.
What I am hoping to show by bringing in outside data from Glassdoor is that AI is paying below average for their positions. I wanted to do a comparison between AI’s positions and the same positions I found in the Glassdoor data set. AI was nice enough to give me six starting salaries of positions they recently hired for.
Below are the starting salaries for positions at AI:
Below are the national averages of the same positions:
Now, I bring in the salary/position from AI and the same positions I found in Glassdoor’s data set for comparison.
From this bar graph, it is no surprise that San Francisco leads in every category, given the cost-of-living. However, it also appears that AI is very competitive compared to the ten other cities from the Glassdoor data set. Program Manager is the only role where they seem to be below average. This could be due to Program Manager being such a broad term and AI might not utilize their Program Managers similar to other companies, thus not needing to pay as high of salary.
Is it possible that AI is getting a lot of applicants with a variety of desired salaries, but just hiring those candidates who are looking for lower salaries by comparison? Let’s take a look.
From this graph, it does not appear that AI is picking candidates that are requesting a lower salary to fill their positions. In some instances, they are actually taking the candidate that requires more salary than the average applicant. This tells me that AI cares more about finding the right fit for their roles than finding the candidate with the lowest salary requirements.
Conclusion:
From the data that I was given from AI, I do not think I can say with confidence as to why they have not had much luck with Doctoral candidates progressing in the interview process. It is my opinion that there are too many intangibles the data set from AI does not provide. For example, I was not given the details as to WHY a candidate was not qualified or WHY they were not a fit. I understand this was in part due to privacy concerns, but if I would have had access that to that information I could have quantified that information and been able to provide a more thorough analysis.
Bias:
There are positions within the data sets such as Research Analyst and Program Manager that are very broad terms and will vary greatly from company to company. This makes it difficult to make equal comparisons across companies that are made up of multiple industries. I feel it would have been beneficial to have had a data set comprised of multiple companies that are in the same field as AI to make for a more level playing field of salary comparisons.
