Data Quality


Effective data-informed decision making hinges on high quality data.

Data collection can be time consuming and costly. And, if it is not done with quality in mind, the data you collect will not be effective in guiding the critical decisions you need to make. You may already be collecting lots of data on your farm, but if that data is low quality, it will not be accurate, complete or consistent. Low quality data can severely reduce the return you receive from your digital agriculture technology investments.

In this section, we’ll define what high quality data is and show you how you can assess the quality of your on-farm data collection practices. 


What is High Quality Data?

The  Federal Committee on Statistical Methodology (FCSM)  in the United States created a useful framework for assessing the quality of data. This general purpose framework, defines high quality data as follows:

“The degree to which the data capture the desired information using appropriate methodology in a manner that sustains public trust.” ( FCSM, 2020 ) 

We can further break down the concept of data quality into three domains:

1. Utility: Is the data well-targeted to clearly identified or anticipated needs? Does the data provide useful information?

2. Objectivity: Is the data accurate, reliable, and unbiased? 

3. Integrity: Has the data been collected using a well understood protocol and protected from unauthorized access or manipulation? 


How do I Determine the Quality of my Data?

The three domains of data quality from the FCSM framework can be further broken down into eleven sub-categories, or dimensions, that provide a more complete picture of data quality. The complete array of eleven sub-categories of data quality is shown in the diagram below.


That said, how do you go about determining the quality of your agriculture data?
In the table below we provide a list of questions that you can use to think through how the eleven sub-categories apply to the quality of your farm data.


Data Quality in Variable Rate Fertilizer Application

Let’s consider the example of variable rate (VR) fertilizer application prior to, or during seeding. Many producers will choose to work with a trusted advisor who can analyze their agronomic data from previous growing seasons to develop a variable rate prescription. As part of this process, the service provider can also identify whether the existing data is of high enough quality (i.e. complete, well calibrated, sufficiently geospatial granularity, etc.) to create a meaningful nutrient prescription. If that’s not the case, additional data may be needed to establish a stronger baseline. This could include doing additional soil tests (e.g. electrical conductivity scans) or re-calibrating relevant machinery (eg. yield monitor) to improve the quality of available farm data. 

In this case, working with a trusted advisor plays an important role in identifying the best way to improve data collection practices. Trusted advisors can also help producers keep to a manageable scope, like supporting the development of the variable rate fertilizer prescriptions, rather than trying to improve the quality of ALL the agricultural data being collected at the same time. 


Why is High Quality Data Important?

The popular phrase “garbage in, garbage out” is an effective metaphor to describe the relationship between data quality and data-informed decisions. If you use low quality data to begin with, even high quality analyses will not be able to produce useful findings -  you may end up with low quality results that can be misleading and are unlikely to create the outcomes you are seeking.  


Although it is tempting to think that large amounts of data, say from precision agriculture equipment, can overcome this basic fact, it is important to remember that the same fundamental principles hold true if you are using large quantities of data, or even big data (see  🏟️⁠⁠Aggregate Data & Big Data⁠ ).  

Thinking about this another way, just as you would not put contaminated fuel in your farm equipment, it is important to avoid using low quality data to fuel your on-farm decision-making.  



Now that we've gone through some key elements of data quality, let's test out your understanding in a short quiz!




Next:  🎯⁠⁠Data Accuracy and Precision⁠