Guidance for Making Evidence-based Organizational Decisions

The Scientific Method & Decision-Making

In Part 1 of our organizational decisions blog, we introduced you to the emergence and importance of evidenced-based decision making in organizations and the world of business. Now we’ll guide you through the best practice process for planning and conducting data analysis for decision-making and provide examples for applying these strategies to maximize the potential insights available in your data.

Conducting data analysis yourself may seem daunting, particularly if you don’t consider yourself a researcher or an analyst. It may be difficult to determine where to even start on this endeavor, but the good news is you can use the scientific method to conduct organizational research and use your data to make evidenced-based decisions! In this context, the scientific method is the systematic and deliberate gathering and evaluating of data to generate and test ideas to help you answer questions about your organization and make valid and actionable decisions.

The scientific method follows these general steps:

1. Form a research question that is narrow and focused.
2. Form a hypothesis for your research question that is testable and measurable.
3. Gather/generate data to test the hypothesis. Sometimes the data will already be available to you and sometimes you’ll have to generate it.
4. Analyze data to test the hypothesis. It is important to carefully consider which analytical and statistical processes to select to accept or reject your hypothesis and determine whether the findings show practical differences, and whether those differences require action.
5. Draw conclusions, communicate results to stakeholders, and use the results to act, as needed.

Applying the scientific method to your organizational data allows you to understand the relationships among the data points and to make decisions based on them. Using science to make rational, evidence-based decisions is the best method we have to ensure our decisions are unbiased and sound. However, understanding the scientific method and its potential benefits for your organization isn’t enough to realize benefits. To maximize the value of this method, you must adapt the scientific method to the needs of your organization and specific situation.

Guidance for non-Researchers/How to Use Data to Make Better Decisions

Let’s walk through an example of what this adaption process might look like in an organization! Imagine leaders in your organization recently found out that employee satisfaction is low, so they rolled out a new program in an attempt to increase satisfaction. The program just completed its first year of operation and your organization’s leaders want to know how the satisfaction program has impacted the bottom line. You’re tapped to conduct the evaluation and inform stakeholders of results. Let’s go through the steps of the scientific method with this example in mind to highlight how these questions might be answered to make effective programmatic decisions.

1. Form a research question

We need to translate the business question into a research question or questions. In this case the business question is “What impact, if any, has the satisfaction program had on overall organizational performance?” One research question that could come out of this is, “Is employee satisfaction related to employee performance?”

2. Form a hypothesis

The research question now needs to be changed into a hypothesis that is testable. In this case, our hypothesis could be “Employee satisfaction is positively related to employee performance.”

3. Gather/generate data

Identify the data needed to measure the hypothesis (e.g., what variables, what data format). In our example, needed data will include data related to employee performance (production rates, employee performance ratings, etc.) and employee satisfaction data. Next, determine if the needed data and data analysis technology exists and is available in your organization. In this case we are looking at two variables that would be relatively simple to collect data on, but more complex hypotheses may require more research variables. Performance data is typically stored in organizational IT systems and you can reach out to the employee satisfaction program to get the satisfaction data. If the data is not readily available, identify how you might efficiently collect these data. Each organization is different, and the availability of data will vary, so it’s important to make sure your hypothesis truly can be tested with quality data before going into the analysis phase.

4. Analyze data

This is where you use the data you’ve collected to test your hypothesis. You must decide which method(s) is the most appropriate based on the data that you have collected. It’s necessary to keep in mind that while complex analyses are becoming more popular and visible in the media, the most appropriate analysis may be the simplest. When translating the scientific method into practice, the goal of answering the research question needs to be the most important factor, rather than conducting flashy and potentially inappropriate statistics. In this case a correlation or regression analysis would sufficiently test the hypothesis and answer the research question.

5. Draw conclusions and communicate results

Once you’ve conducted your analysis, interpret the results according to the analysis methodology chosen. For example, if you conducted a correlation analysis, identify the magnitude of your correlation coefficient (i.e., the value the r statistic), and determine whether the correlation is statistically significant (i.e., the value of the p statistic).

Guidance for interpreting analysis results are easily found online. In this example, if your correlation between employee satisfaction and performance is statistically significant (and the coefficient is acceptably large), you can conclude that your hypothesis is correct (or more accurately that the null hypothesis is not correct), which answers your research question. The answer to your research question will greatly contribute toward your ability to use organizational evidence to answer your business question. At this point, you can format your analysis output and interpretation for presentation/reporting (e.g., use data visualization/storytelling to convert research output back into business story). Using our output (fingers crossed for good results), you’d present results to organization leadership, for their decision-making.

Conclusion

It’s becoming increasingly important for organizations to include data analysis as part of the organizational decision-making process. While it may seem intimidating to start integrating data into your decision-making process, you can leverage the scientific method to analyze data related to your business problem in a straightforward and simple way.

It is important to emphasize that while it’s helpful to use data in decision-making, it is potentially harmful to solely use data in decision-making, as opposed to observation, judgement, and theory. This is because your data has the potential to be biased and/or inconsistent, inaccurate, or incomplete, which can lead to decisions that reflect the truth in the data, but not necessarily the truth in reality. Using the data analysis planning process and implementation strategies outlined in this blog mitigate this risk and allow anyone—even the non-researcher—to engage in evidenced-based decision-making!

Catherine Neale is a Human Capital Consultant on FMP’s Strategic Human Capital Management team and is an Industrial Organizational Psychologist. Catherine is from upstate New York and enjoys running, rock climbing, and baking.

James Wilcox joined FMP in 2019 and works in the Technology, Analytics, and Transformation Center of Excellence. James is from Orlando, Florida and enjoys learning new things and exploring new places.