How do you use statistics in your job

[donrogers]

I am soon to be teaching a statistics course to masters of accounting students. I want to be able to motivate their learning of the material in relation to how they will actually use what they are learning when they begin their career or as their career progresses.

Can anyone provide examples of when you (or someone you work with) uses statistics in their job tasks?

Particularly, things such as regression (linear, nonlinear, and logistic), hypothesis testing, t-tests, etc. Things that fall under "data analytics."

The more details you can provide, the better.

Thanks in advance.

Don

 

[Bev D]

I teach Deming’s Enumerative vs Analytic studies first. (See “On Probabiity as a Basis for Action”, Deming 1975). then I spend some time discussing the foundation of statistics which is often given very short attention - if at all: Homogeneity. (See “Why do we keep having hundred year floods” and “the secret foundation of statistics” both by Donald Wheeler available free at Quality Digest.)

THEN I teach things like SPC (Deming and Wheeler) which would be very useful for accounting students as they tend to think of changes as absolute and not just random variation. I also reintroduce non-homogeneity since many ups and downs in accounting are driven by human manipulation to ‘make the budget’.

I never teach t-tests or ANOVA at all as they simply aren’t needed. I do teach multi-vari and graphical analysis for well designed studies, but I don't think accounting people would use them very much. I think dissagregation of data sources would be very important tho. So the concept of components of variation would be very important. A neat little paper on this is: “A painless look at using statistical techniques to find the root cause of a problem”

One thing I would also caution you on is to be clear about the definition of words. Statistics uses ‘common’ words that non-statisticians use with very different definitions. For example: variance has a very specific meaning to accountants that is NOT the same as when a statistician discusses variance.

 

[Miner]

I use a very wide variety of statistical tools in my my job as it touches almost every function within the organization, so I will focus on those that I feel would be most useful to accounting majors.

As Bev stated, most accounting people view numbers as absolute, so the single most important concept they need to understand is that of variation. Otherwise, they go into life beating some poor sap over the head for what is simple random variation over time over which they have no control. Next, they need to be able to run "what if" or sensitivity analysis scenarios using Monte Carlo simulations or Excel's Scenario Manager to incorporate the expected variation in forecasts, etc. into budgets or business proposals.

Most of the tools that you mention such as regression are tools used to develop a mathematical model of an unknown system. In accounting those models are typically well known, but the impact of variation upon those models is not known. Hence, my recommendations.

Added in Edit: The most useful distribution to an accounting major is the triangular distribution (Best case, worst case, most likely).

 

[Statistical Steven]

Don

If you are teaching accountants, you need to make it relevant to them. Accountants understand auditing, which is a sampling of the financial records. So I think you need to frame it with the idea of Type I and II errors. I would start with a simple introduction to probability (flip a coin, roll dice, etc). Introduce the binomial distribution. Then show how you can calculate the probability of detecting an out of conformance in the sample, based on the true percent defective. For example, I use extensively the idea that even when you sample 30 items and none are defective, you still have a 95% probability that the true defective rate is 10%. Does this make sense to you?

I am soon to be teaching a statistics course to masters of accounting students. I want to be able to motivate their learning of the material in relation to how they will actually use what they are learning when they begin their career or as their career progresses.

Can anyone provide examples of when you (or someone you work with) uses statistics in their job tasks?

Particularly, things such as regression (linear, nonlinear, and logistic), hypothesis testing, t-tests, etc. Things that fall under "data analytics."

The more details you can provide, the better.

Thanks in advance.

Don

 

[Matt33]

The single biggest change I have seen for accounts is to get them to look at more than two numbers at a time and to put this in chart form.
Many income statements will list: this period, last period, and maybe the difference. i.e. Jan-June, 2018 xxx, Jan-June, 2017 xxx and the reader is supposed to make sense of these two points in time.
Instead, we try to get them to look at various metrics (Net Sales, Total Expenses, Net Income, etc.) over many time periods using a Run (or Individuals) chart. That way, they begin to appreciate variation in the system.

 

[donrogers]

Thank you all for the responses (and any more that might come in). This has given me some ideas.

 

[bobdoering]

It is also critical to understand the requirements of each statistical tool. If the tool requires random and independent data, make sure your process output IS random and independent, and not a function - or the results may yield incorrect analysis - in some cases the exact opposite of the true process condition!!!