Data omission is not as spooky as it sounds. When I first read the term, my response was negative, as if “data omission” suggested that someone, somewhere, had criminally sabotaged an important dataset.
In reality, data omission is almost always an unintentional error. It’s an expected part of input flaws that should be caught in a reconciliation process.
But we’re beating around the bush: what exactly is data omission, and what does it look like on a day-to-day basis for accountants and data analysts in general? This article covers just that.
Data Omission Definition
In short, data omission is an intentional or unintentional input error by which a machine or human excludes a data item or dataset from a larger database, whose accurateness and completeness are thereby thwarted.
In other words, data omission occurs when a person or machine leaves data out. Whether it’s intentional or not, excluding data is always bad. If you don’t have much experience with data analysis, it’s helpful to understand the history of data quality to truly grasp the impact of omission.
Before there was big data (circa mid-2000s), most decision-makers in business and government relied heavily on intuition and limited data samples, such as interviews and focus groups. However, digitalization has exposed truth through data. Decision-makers now have a better view of the playing field and the stakes at play when they make choices.
The importance of data as a decision-making tool and an information tool cannot be overstated, but a tool is only as strong as the materials used to make it. In the context of data, we refer to this idea as “data integrity” or “data quality,” each of which has a list of tenets such as consistency, coherence, clarity, credibility, reliability, relevance, and usefulness. Perhaps most important of them all is data completeness.
Data omission is a direct threat to data completeness, and therefore, a direct threat to the principles of data integrity and data quality.
That said, data omission is not always a serious problem. Depending on the dataset and its uses, an error of omission may not seriously impact conclusions made on outputs. For example, if an accountant wrongly excludes $1 of revenue in a company that makes $1B, the impact of this error of omission does not impact the story.
Nevertheless, it’s a slippery slope. If $1 is excluded here, others might be excluded elsewhere.
In this article, we’ll look at data omission in accounting and in auditing, as well as some examples, so you see how the impact of data omission plays out in different contexts.
What is an “error of omission” in accounting?
The meaning of omission in accounting is well established, and the impact of omission by machines is a growing concern.
Accountants have a job heavily dependent on their attention to detail. They source data from invoices and encode it in a software program as both a credit and a debit. Without getting too detailed, debiting usually means increase in an asset account, while crediting means an increase in an equity or liability account.
This is called double-entry accounting, and it leaves a lot of room for error. Every individual entry carries a risk of error, and the double-entry system multiplies that risk by 2 (roughly). An error of omission in accounting can happen in a few different ways:
- The accountant misses an invoice.
- The accountant encodes the invoice in the incorrect database (thus omitting it from the database in question).
- The accountant only encodes a debit, excluding the credit.
- The accountant only encodes a credit, excluding the debit.
Errors of omission are not always detrimental to the outcome of the accounts. Using the same example as above, if $1 of revenue is excluded in a given period for a Fortune 500 company, there is no reason to be concerned. However, if it were $50,000, you might have to rethink your staff, or move to a different workflow style.
Deciphering between significant and insignificant errors of omission in accounting data is tough. That’s why there’s auditors to help companies have a clear view on their books. We’ll talk about this later.
Error of Omission vs Commission in Accounting
There’s a subtle difference between errors of omission and commission. In short, omission is the exclusion of data, whereas commission is the inclusion of incorrect data.
In accounting, errors of commission are much more common. The easiest way to explain why is that it’s much harder to miss an invoice entirely than to mistake one digit in a large number.
For example, if you need to credit 13,9072.23 USD, you’re unlikely to loose the invoice for this amount. However, you could very easily encode 13,0972.23 USD. See how subtle that is?
In addition, you have to do this two times (credit and debit). Because it’s the act of encoding that leads to error of commission, you double your risk (roughly) of error with a double-entry system. On the flip side, it’s still unlikely that you miss the invoice entirely, so errors of omission are less common.
Errors of Omission in Auditing
Errors of omission in auditing are closely related to those in accounting. Auditing in a corporate context is the process of reviewing financial information to ensure its accuracy and integrity.
Auditors examine the accounting process, taking samples from a company’s trial balance entries and tracing the data encoding & output process from start to finish.
This is where the dynamic of a double-entry systems comes into play. Because auditors start with the trial balance (a list of all accounts with activity during the given period on them), any accounts with only one activity during the period in question are at risk.
The reason they are at risk is that, should the information for this one entry be placed in another account, or left off of the accounts entirely, it would be nearly impossible to recognize the error.
The double-entry system has a natural protection against this kind of account data omission–since the accountant has to make the same mistake twice, in two different accounts, in order for an activity to go unnoticed.
This line of defense exists in theory, but in practice, there’s still room for error of omission. For example, if an accountant records a Special Cost Item A under Special Cost Account B, and cash decreases as the counterpart entry, you would never see Special Costs Account A on the trial balance.
That said, if an auditor examined the Special Cost Account B, s/he may or may not catch the single Special Cost Item A.
In short, errors of omission in auditing are a core part of the job — sometimes they’re unidentifiable, but other times auditors bring serious value to the corporate accounts with their review.
When auditors find errors of omission, they can advise the company on the seriousness of the mistake and any consequences of not fixing it. As mentioned above, $1 in a $1B company is clearly insignificant. But $50,000? It’s hard to say.
Error of Omission Examples
Accounting and auditing are NOT the only domains touched by data omission. Here’s a list of other examples in which errors of data omission can occur:
- Manufacturing sensors
- Webpage crawling robots
- Webpage data consolidation
- Payment gateway transactions
- Demographic surveys
- Focus groups
- Governmental data collection
- Company address on Google
- Opening hours of service on Google
- Transportation metrics
- GDP calculations
- GNI calculations
- Unemployment rates
- Consumer data collection in retail
- Bank personal financial information
- Academic research
In all of these examples, errors of omission can occur. A manufacturing sensor (1) may malfunction and exclude a reading. Demographic surveys (5) may be subject to unrepresentative sampling. GDP calculations (11) may leave out important data that would otherwise change its value.
You don’t need to understand all of these examples in detail. The point is that they’re all subject to intentional and unintentional errors of data omission. It’s up to the organization collecting them to decide which errors are significant, and which are tolerable.
Analytic Errors of Omission vs. Errors of Omission in the Medical Field
AnalystAnswers.com is about data, finance, and business analysis, but let’s turn our attention now to the medical field. If you Google “errors of omission,” you’ll quickly see articles in a medical context.
I’m no doctor, but it’s important to understand omission and commission in medicine vs data. Data omission is the intentional or unintentional exclusion of data points or datasets in a database, whereas errors of omission in medicine are mistakes that surgeons and generalists make by not performing a step they should have.
An error of commission is also possible in medicine, and just as with data, it’s usually more serious. Errors of commission in medicine are mistakes a doctor makes while performing surgery, providing a diagnosis, or administering medicine.
Conclusion
Unintentional data omission is a normal part of data analysis — whether it’s in the fields of accounting and auditing, or data analysis generally. Machines and people are not infallible. They make mistakes, and one of those is data omission. That said, when omission is intentional, it’s a serious problem. Humans would only intentionally leave out data if they stand something to gain from its absence.
At the end of the day, data omission is always a bad thing. It breaks the data completeness tenet in data integrity and data quality. Over time, technology helps data analysts improve their processes to minimize unintentional data omission, but we should never get comfortable. Always be careful to include all of the relevant data in a data set!
