Once I've got the basic idea of what I'm looking at and how it's been populated and controlled, it's time to look a the data. The first thing that I do is a scan of the data. For smaller tables I'll zoom out and take a full view of the whole. For a large, table-shaped dataset like you find in many spreadsheets, I'll start with a Z-shaped scan of the data: across the columns in the top few rows of data, diagonally down to the bottom few rows, and then back across again. This allows your eye to catch some obvious signs, for example:
Some things that often show up are weirdly repeating numbers, dates and times that make no sense or seem to be recorded in the wrong order, or patterns that emerge based on who did the work or when - e.g. one person's numbers are consistently lower than everyone else's. Depending on the scope of the data I like to keep track of the things I see during these scans rather then investigating them right away. Also, if I'm not the "Quality Control" step (and I will post some other time on why I think QA should not be the QC step), if I see too many obvious problems off the bat I'll send it back to the owner with a message that someone needs to do some QC before it gets any further review or audit. Now that I've got my list of things to check out, I'll look a little closer, with the source data at hand:
Finally I'll look at any charts, graphs or formatted results tables and do a sanity check - number of points, order of magnitude, check that the corners and a few in the middle match, that kind of thing. Until next time, thanks for reading! – Brendan p.s. Enjoy this message? Read more at the Hyland Quality Systems website. |
I'm Brendan Hyland. I help regulated facilities transform their software, spreadsheets, workflows and documents from time-consuming, deviation-invoking, regulatory burdens, to the competitive advantage they were meant to be. Join me every week as we take a few minutes to explore, design, test and improve the critical systems we use in our facilities.
I'm excited to be presenting at the 2025 SQA Annual Meeting next week - "Is it the Right Tool for the Job? How QA and Regulatory Professionals can Guide Software Decisions in Regulated Environments" As our familiar software tools become more feature-laden and generalized, it's critical to ensure that software meets clear use cases and basic user requirements. And with generative AI being shoehorned into every platform, defining if and how software is appropriate for the intended use has never...
Last time we left off with a cliff-hanger of a question: How do you prove you're you when signing a document? There are several ways I've seen that the 3rd party providers prove that it's you who's signed the document: You clicked a link from an email. You paid for the service with a credit card. You provided some government issued photo ID. Someone, such as a notorized public or your HR department, has verified it's you in person. Obviously these are very different levels of assurance. Then...
There are several levels of 'signatures' that you can apply to an electronic document. The first and most basic is just an image of your written signature. One common option for this is to print the document, sign and scan it back in again. A more convenient version is to have an image of your signature saved that you can paste into documents. This is what many free versions of pdf software and word processors offer as a basic document signing option - a 'stamp' of your saved signature image....