Quantitative Analysis Methods – Using Simple Linear Regression

LRA1ScatterPlotwithCurvePublished in the Project Post-Gazette, February 2014

by Cheryl Wilson, PMP, RMP, CCEP

If you are not seeing your risk management program produce valuable information that you as a project manager (PM) can use to make a difference on your project then you are more likely managing a risk program that has no true value. Valuable information is information that the PM can use to make decisions on their project that will ultimately produce better fit-for-use (FFU) deliverables. Many current projects still use the industry’s supposed best practices of risk management and more than likely, the PM and the project team cannot understand why valuable data cannot be collected from the risk program in using the current risk environments taught by these frameworks.

Current risk frameworks many times stop at the collection of risks and a qualitative prioritization of these risks. These current risk frameworks provide little in the way of performance measurements even when the organization that is managing projects and programs have instilled sound evaluation techniques. I have found that because the current risk frameworks are not challenged there is a widespread inability to sometimes make the subtle differentiation between a risk framework that has methods that work and methods that do not work because these methods provided by the so-called standards setting bodies are so widespread and largely unproven. Project managers do not even think about implementing different proactive risk management techniques because they do not know what they need to evaluate to implement a more robust risk management environment.

This article will talk about one of the quantitative analysis techniques that can be directly applied to your current projects risk management activities. I ask that you keep an open mind because you will have to take a jump from how you currently do your risk assessments to a much broader view of how to more proactively manage your risks. Many times when I asked project managers if the risk program produces valuable information that can be used to manage their projects I get a very blank stare as if they were thinking about whether or not their risk program actually could produce valuable information based on how they manage the risks within the current state. Many PM will tell me that their risk programs were successful because they manage the programs the way the current bodies of knowledge have told them to manage their programs. The blank stare is usually the outbreak of the confusion between the fact that they are managing their risk program according to the current standards-setting bodies risk management solutions and hoping that their risks programs will be a success given the use of these accepted methods. However, they also know that they probably have not produce valuable information. They answer the question that their programs are successful because they base their success on following a methodology that “identifies risks,” “communicates risks” and “records risks” in a risk register.

So now that I have set the stage for you to have an open mind to risk assessment methods with some measurable evidence of success, I hope that you will be able to move from your current comfort level to a more effective set of methods that may help you produce more valuable information oriented towards the reduction of uncertainty which, after all, is the main reason for implementing a project risk management program in the first place. If risks are not properly assessed then they cannot be adequately assessed thereby producing the valuable information on which to make decisions.

From a risk perspective, performing quantitative risk analysis before qualitative risk analysis is a more proactive risk approach in your risk management process. Qualitative assessments only allows risks to be prioritized by a subjective speculation and not by objective verifiable information. Almost every PM that uses Microsoft Excel has the capability of using very easily understood yet more powerful quantitative analysis methods. The use of these more quantitative methods produce the information whereby decisions makers can understand the reduction in uncertainty and thereby prove the value of the information obtained from the analysis methods.

In last month’s issue of the PPG we provided an overview of the qualitative risk analysis tools that a PM and project team can use. This article will talk about linear regression analysis as one of those tools. This is actually just a review of this tool as the November, December of 2013 and January of 2014 issues of the PPG spoke about regression analysis in more depth.

Let me also add here that one of the important points to remember is that as a PM you really want to be able to map a dollar amount, called a Risk Equivalent Value (REV), to your specific risks. This is the only way you are going to step from the very subjective way of scoring your risks into a truly objective approach. I know you have been told by many organization’s spouting their bodies of knowledge that performing quantitative risk analysis on your risks is too hard. With today’s advanced desktop computers and readily available software this is simply not true. Also realize that if you want valuable information from your risk management program that can actually help you manage a project that by simply qualifying your risks by a subjective “guess” you are not going to produce the valuable information that will in turn make your risk program more effective.

Okay let us get started. A linear regression is a statistical method that attempts to show the relationship between a dependent and an independent variable through the use of a technique that minimizes the differences between the dependent and independent variable when attempting to solve the simple equation of their relationship. Remember from your junior high or intermediate school algebra class where you solved what your teacher called, “a straight line equation?’ This is all the more difficult using linear regression is since we do not have to become statistic academics in order to garner value of its implementation in our risk program. Linear regressions can be used in your risk management activities to evaluate trends and make estimates or forecasts.

To simplify the explanation, we can solve a value of the dependent variable by using the equation that Excel will produce for use by using pairs of numbers from historical risk programs. That is, we gather data from the risk programs of similar projects and use these pairs of values to create a simple equation that we can use to solve the unknown risk value of our current, future project. In effect, we are using the past to help us understand the future.

This means the main application in our projects of linear regression analysis is as a predictor. What this means is if we have two values or parameters such as time and costs, or time and resource utilization, or even time and material use, can we find a relationship where if we know one of the values, can we estimate the other value. After the PM knows the trend line that is automatically calculated for him/her in a simple Excel spreadsheet (see the above three back issues of the PPG, Techniques Corner for more details,) they could use the slope of the line to help determine information for a future value that has yet to be realized. Linear regression is therefore a simple method for using your historical information to produce more valuable information for your estimates, forecasts, and risk assessments.

An example of where a PM or project team could use linear regression analysis: say you have a risk that is similar to those that you have seen, mitigated, and even had to respond to in previous projects. The risk value you are seeking to estimate for your current project is the value of the risk potential’s cost of impact or RCI. You have 8 other projects that have both the values of your risk negative impacts and the resulting RCI that was used in these projects. Taking these 8 pairs of numbers, for example the number of external parties that must interface with your IT project and the resulting RCI estimated for these projects. Putting these 8 pairs of numbers into a simple Excel spreadsheet and using the built-in formulas and asking Excel to determine the relationship equation (the linear regression statistic) you are presented with a formula that you can use in future risk assessment activities to relate the number of external parties that need to interface to your IT project and the risk equivalent value (REV). Wha La! You have your much improved and more valuable estimation of the risk’s REV: more than just a simple guess or subjective, qualitative number that bears little to reality or even history.

I hope you feel a bit more confident to start using quantitative analysis methods on your projects. They are very easily understand, utilized, and deployed by most every PM that has a copy of Microsoft’s Excel spreadsheet program.

Next month and in the following months, the PPG Technique’s Corner will talk about the other quantitative risk assessments that you are a PM and your project team can do:

  1. Multivariate regression analysis
  2. Linear programming
  3. Monte Carlo simulation
  4. Variance analysis
  5. Gap analysis, and
  6. Goodness-of-fit testing

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s