Analyzing Culture

The average person spends more than 90,000 hours of their lifetime at work. Where we spend that 1/3 of our lives matters. Company culture is important to employees, employers and even customers.

Author Simon Sinek tweeted, “Customers will never love a company until the employees love it first.” A healthy corporate culture delivers better on expectations and achieves greater competitive advantages.

Internal audit is uniquely positioned to provide a holistic, in-depth, and independent view. We asked an internal auditor of a large manufacturing company to share how he used CaseWare IDEA® to analyze culture.

Defining Culture

Culture is defined as the ways of thinking and acting that shape an organization. There are two types of culture, material and non-material. The company focused on its non-material culture values, which include:

Integrity, safety, honesty, ethics, engagement, responsibility, and performance.

Step 1: Data Collection – Analyzing Values Using Existing Data

 
Auditors aimed to conduct their analysis using data the company was already collecting and storing.

Values were grouped into categories:

• Engagement – Measuring the active participation of employees
  • Data Source: Internal and external surveys including a Gallup Engagement Survey the company conducts each year.
• Ethics – Measure the integrity levels of employees
  • Data Source: Defense industry initiative (DII), a survey conducted by the Ethics and Compliance Initiative
• Performance – Earned value management including the time to complete tasks, production rates, budget vs. actual costs, etc.
  • Data Source: Cost performance index (CPI)
• Responsibility – A bit more difficult to measure, so the company looked at willingness to work overtime and/or willingness to limit time off
  • Data Source: Payroll department

They also used safety data collected for OSHA reporting including incident rates, lost workdays, and workplace accidents.

Data Collection Hurdles & Sources

• Not all data was available for integrity, honesty, and ethics. Departments with fewer than 5 responses were withheld.
• Not all departments had performance metrics
• Time off and overtime data was provided by the payroll department and included dollar values
• Safety data came in three variables: total case rate, lost time case rate and lost workday rate
• Department headcount included changes due to retirement, termination and voluntary transfer

Step 2: Cleaning Data & Missing Data

Describe/Summarize/Tabulate the Data

 
They determined early on the data consolidated at the company level was too broad and generic. Data at the individual level was too discrete and raised concerns about protecting the privacy of employees. The balanced medium was to use departmental-level data. While some divisions were structured differently, they all contained departments, which was similar enough to serve as the common denominator.

As people enter and exit a company, the culture changes. Auditors used annual averages spanning three years to normalize their data. They also applied variance shifts based on the size of the departments. Each department had a unique population, and not every metric was reported by every department. Auditors used summation to help identify this using the one-click Field Statistics function in IDEA.

Defining Zero

 
While auditors had data from several hundred departments, not all data existed for each data type and each department. Some departments were formed during the review period, while others were disbanded.

A key field was established by combining the department and year; the databases were joined in pairs based on that key. They only considered matching pairs in their analysis. Since they could not validate whether data was missing for no reported occurrence (a valid zero), or the absence of data altogether (an invalid zero), they had to omit those from review.

Step 3: Analysis

With data in hand, it was time to crunch the numbers, but they were somewhat unsure about where to start. They referred to an online resource to gather different types of analysis including:

• Describe/summarize/tabulate data (Step #2)
• Test hypotheses (predictions) about the data
• Explore data and search for structure, patterns, factors, and clusters
• Compute statistics for industrial quality control/improvement
• Perform
  • (a) statistical power analysis
  • (b) sample size estimation
  • (c) Confidence interval estimation
• Find out the meaning of statistical term or concept
• Explore distribution tables (Z Table, T Table, Chi-Square Table, F Tables for alpha)
• Run analysis on enterprise-wide data stored in one or more complex databases
• Explore large amounts of (typically business- or market-related) data to search for consistent patterns and/or systematic relationships between variables

[Source: http://www.statsoft.com/textbook/statistical-advisor]

Regression Analysis to the Rescue

 
They focused on linear regression analysis, using both Excel and IDEA. The IDEA Help Desk provided a regression analysis executable to examine the relationship between the variables of interest, including definitions for each statistical result. It also performs predictions and extracts outliers into a separate IDEA database for further analysis.
 
Download the Regression Analysis Executable here!
 
While they considered using non-linear regression, the team could not come up with a rationale as to why a non-linear relationship would be expected, so they used hypotheses to identify relationships within the data. For example:

• Departments working more overtime might have more safety incidents
• Higher employee engagement might be reflected in higher productivity levels

Data does not have to be one-dimensional, nor was their study of it. Auditors used regression analysis on every dataset combination, both as a hypotheses test and exploratory analysis. They compared each section to find relationships between categories, such as engagement and ethics, or safety and time off. Look at what impacts culture, and keep in mind norms vary by location, diversity and other underlying factors.

Explore Data & Search for Structure/Patterns/Factors/Clusters

Data was converted to a five-point scale, which sparked some debate about how to rank things like overtime and time off. For example, is it considered “better” that an employee takes regular vacations or sick time?

They developed a composite chart of all the factors as an exploratory exercise, then used separate charts for each year. Holding the departments as a constant, auditors looked for any obvious trends, outliers, and other anomalies.

Step 4: Bringing the Results to Life

 
Advancements in data visualization tools now give analysts multiple options for sharing findings. Auditors referenced an online resource, Data Visualization Catalogue to help them narrow their selection by determining what they wanted to show. They used a variety of scatter plots to track trends over time. Because correlations do not imply causation, the auditors looked at cause-and-effect relationships between two variables.

For example, the analysis uncovered a strong negative correlation between employee engagement and the total case rate for safety. Could this mean engaged employees are safer? Or are safe employees more engaged? They found a negative correlation between integrity/honesty and ethics with net attrition. Could that mean unethical employees are more likely to leave the company? Does PTO generate more employee engagement?

Are employees who work more overtime more or less engaged? Looking at engagement scores compared to time off data, are employees who take more time off more engaged? Does performance increase with a more engaged workforce? Is there a correlation between safety and overtime? One might assume an employee working longer hours would be more likely to make more mistakes and have more accidents, but that was not the case.

Using data visualization tools within IDEA and Excel, the auditors were able to tell a story with their numbers, table, and charts that management readily understood. It was just as important to tell management how they conducted their analysis as it was to inform them about what they found and what it meant about the current state of the corporate culture.

Quick Tips

• Have a clearly defined approach before starting your culture analysis.
• Gain the support of management about the importance of analyzing culture. You will need to articulate the value and build a case for spending time and resources to conduct your audit. There are plenty of statistics out there to help, for example:

   
  Recruiting & Retention: Highly educated and qualified job candidates are prioritizing culture, including how the company’s values align with their own and growth opportunities. Engaged employees are 87% less likely to leave the organization. Turnover costs range from tens of thousands to 2x the candidate’s annual salary and impact morale. Yet, just 36.7% of employees are engaged at work, according to a Gallup survey. Disengaged employees can impact profitability, brand reputation, and even safety.

• Determine what you want to assess and what data already exists that can be used.
• Consider analyzing high-risk areas such as departments with high turnover rates, locations with exceptionally high or low sales, departments with below-average engagement scores, etc.
• Give yourself some flexibility along the way, just as this company did when working to address missing data.
• In formulating your conclusions, look for relationships among the data and apply exploratory analysis. Ask the right questions to dig into underlying issues and what they mean.

As the modern workplace continues to evolve, organizational culture is becoming increasingly important to internal auditors. Every company is different. While there is no single rubric for “ideal” company culture, IDEA can help you turn “gut feel” observations into meaningful insights for management using data you’re already collecting.

Need some help acquiring, preparing, or analyzing your data? Contact our Technical Services team at [email protected].