Why they ‘Quit You’

Methodology

To determine the likely reasons why someone would quit their job, PayScale developed a multinomial logistic model to calculate the statistically significant probability that someone in one group would leave their job in comparison to another group (e.g. Millennial vs. Boomers.) In this type of modeling, the focus is typically on the idiosyncratic (peculiar or individual) answers as an overarching control group. As such, all called out comparisons are statistically significant in comparison to an idiosyncratic “Other” response that is inherently random in nature.

Definitions:

Effective Annual Compensation (EAC): EAC combines base annual salary (or hourly wage), commissions, tips, bonuses, and profit sharing. It does not include equity (stock) compensation, cash value of retirement benefits, or value of other non-cash benefits (e.g., healthcare).

Generations: Observations are grouped into the following generations based on their reported age at the time they took the survey.

– Boomer: Those born between 1946 and 1964
– GenX: Those born between 1965 and 1981
– Millennial: Those born between 1982 and 2002.

Income Levels: Income levels are based on the single-person earnings levels used by Pew Research Center. An observation’s income level is determined based off of their EAC.

– Low income – $0 to 26,092
– Middle income – $26,093 to $78,280
– Upper income – $78,281 and up

Industries: Industries are custom aggregates of the North American Industry Classification System (NAICS) (https://www.census.gov/cgi-bin/sssd/naics/naicsrch?chart=2002).

Job Levels:
Director Level: Workers with a Director title or a title with a comparable level or responsibility, years of experience, and management scope. These are managers of managers.
Executive Level: Workers with a Chief Executive title (CEO, CFO, etc.), a Vice President title or a title with a comparable level or responsibility, years of experience, and management scope.
Individual Contributor Level: Workers who do not supervise people and do not have a higher level title.
Manager or Supervisor Level: Workers with supervisory responsibility who do not have a higher title level.

Occupations: Occupations are defined, in part, by the Standard Occupational Classification (SOC) system

Survey Questions

In our Job Offer survey we asked, “Are you currently employed”.

Respondents could answer:
– Yes, at the same company;
– Yes, but at a different company;
– No

This question was asked between November 2017 and January 2019.

Follow-up Questions:

People Reporting They Were Employed but at a Different Company

1. What is the primary reason you sought employment outside of your current organization?
– I want a promotion
– I want higher pay
– My current position is not full-time
– I want to work at an organization more aligned with my values
– I am unhappy at my current organization
– I want a more flexible schedule
– I am relocating
– Other

Over 38,400 people answered this question between November 2017 and January 2019

2. How long have you been employed at your current organization?
– Less than a year
– 1-2 years
– 3-5 years
– 5-10 year
– More than 10 years

Over 15,700 people answered this question between November 2017 and January 2019.

3. What is the main thing that attracted you to this new organization?
– Increased responsibilities in this role
– Increased pay for this position
– The opportunity to do more meaningful work
– Better benefits/perks (more PTO, better healthcare, etc.)
– I wanted to work for a smaller organization
– I wanted to work for a larger organization
– Nothing in particular – this is just another job
– Other

Over 14,100 people answered this question between November 2017 and January 2019.

Statistical Method:

We conducted two multinomial logistic regressions on the responses to these survey questions. One with the reasons why people quit their current org as the dependent variable, and one with why people chose their new org as the dependent variable. In both cases “Other” was the omitted/referent group.

To ease understanding, we do not report the coefficients from our regression analysis. Instead we transform the coefficients and report the increased/decreased likelihood of an outcome.’