Help me be more accurate
January 4, 2017 5:01 PM   Subscribe

I need tips or suggestions to be more accurate at work

I think I make more mistakes than normal when handling data or making calculations at work. I try to pay a lot of attention to everything I do but when I audit later on I find crazy, unexplainable mistakes. I'm reaching the point I don't trust much anything I do. Please tell me what works for you.
posted by 3dd to Work & Money (11 answers total) 9 users marked this as a favorite
 
Does all your work have to be done independently? Can you run it by someone else for a second set of eyes? That's probably your most reliable option!

Without more details on the type of work, it's hard to say, but here are some ideas that work for me:

-Step away from the work and revisit it after thinking about something else
-Spot check, or have someone else spot check
-Keep track of the types of mistakes you make, check for those first when you do similar work
-Find an outside source for numbers to compare against
-Compare figures against your previous work to look for internal consistency
-Try to apply your results to test if they make sense
-Check calculations done in excel on a calculator or on paper every once in a while
-Make numbers as simple as possible (few decimal places, logical formatting) so it's easier to spot trends
-Use conditional formatting or graphs to help highlight trends; do a sanity check with these trends
-Before doing calculations, think through what your expected outcome is; compare your result to your estimate

I'm a very visual person, so I tend to lose myself in spreadsheets after a couple hours. I make liberal use of graphs and conditional formatting, which I find works well.
posted by AFittingTitle at 5:25 PM on January 4, 2017 [4 favorites]


Can you break down calculations into smaller parts? That can help you spot errors sooner, or let you better understand where they got introduced.
posted by nalyd at 5:33 PM on January 4, 2017


how disciplined are you at focus? Or do you frequently check Internet, etc? Try breaking your time into half an hour increments (or even 15) and then be really strict about only doing one task in those time increments. If you can't do this easily, distraction may be your problem.

Also, and more prosaically-- are you getting enough sleep? are you bored with your job? Both of those cases lead to errors.
posted by frumiousb at 5:57 PM on January 4, 2017


Are the mistakes in chunks or are they throughout your work? I work with data and normally when we find crazy mistakes it's limited to a day or two so it's clear that something was going on with the employee in that time. If it's throughout your work you really need to look at the suggestions above. Can you do a regular audit? Maybe not right after you do it but before it gets sent out or recheck your previous work the following day?
posted by oneear at 6:15 PM on January 4, 2017


Is there a typical step in your process where this tends to happen? Make a point of double checking if so. Also, is there a range that your data needs to fall within? If so, then excel can change the cell color if you fall outside that range (conditional formatting)
posted by alchemist at 6:28 PM on January 4, 2017


If you are manually entering in data, check every number or ask someone else to. Everyone makes mistakes (really!)

I do a lot of number crunching type things using Excel in my job and here's what helps me:
Clearly separate out the original information you're working with and the calculations you're doing. If it helps, use different color column headings so you don't confuse the the two.
Use the same formula all the way down. Avoid making fiddly manual adjustments, and, if you do, put them in a separate column and label what you did.
If the total of items on Tab A should be equal to the total on Tab B less the total on Tab C, have a formula that checks this.
If you have some numbers that you want to do calculations on, put just that information in its own tab. Then copy the tab and do the calculations in the second tab. Then make a check formula that compares some totals on the calculation tab with the original information tab. If you accidentally change something in the original information while you're doing your calculations, you'll catch it.

Self-review, self-review, self-review. What I found honestly helps the most is that my department has a required review process for a lot of our work products; just knowing that my supervisor (and possible other managers too) will be checking things makes my work way more accurate mostly because I end up checking it myself before sending it to him for review so he doesn't find mistakes. Of course, it's not reasonable for every company or type of work but you can just pretend that your boss is going to be looking through everything trying to find errors. It helps.
posted by eeek at 6:39 PM on January 4, 2017 [3 favorites]


I work on projects that require a lot of data that then go into print with huge circulations (usually in the healthcare realm, so not a place where incorrect data should exist) and I've found that no matter who it is on the team, from the most experienced to the least, mistakes happen and having a fresh set of eyes to check each other's work is invaluable. We have this built into our process. We also hire outside sources on bigger projects to help proof our work. Is there any way for you to do the same with particularly complex/large projects? I also check my work by printing it out, as changing how I look at it makes it easier for me to float past mistakes I've seen a hundred times on screen. Lastly, I give myself time to do another task before I check over work so that I have a fresher perspective.

But like I said, everyone makes mistakes. We work with other companies work often and nearly always find a whopper. I find mistakes in old projects that had sign offs from 10 different sources. Sometimes it scares me, too, the things that I miss, but it helps make me feel like there's always room to improve when my eagle eye boss doesn't notice something either or makes mistakes herself. I'm not saying that mistakes are fine, but that you might not make more than other people. Can you get some feedback from coworker who you know does good work or from your boss about what is to be expected? Sometimes these conversations can lead to learning processes from coworkers that can make a huge difference.

Also, we have strict version control with extensive notes about what was instructed to be updated (which person it came from, when it arrived, on what version). It has helped us track down when a mistake was made so that it doesn't derail a whole project to fix. I can't stress enough how organizing files this way saves us over and over. I don't know if this would apply to your work, but if you can save another version of your file everytime you work on it, then you can compare files, catch i a number changed because of a stray keystroke and then dominoed across the document, etc. I'm having a hard time describing what I mean so I hope that makes sense!
posted by the thorn bushes have roses at 6:44 PM on January 4, 2017 [3 favorites]


For data, checklists. Like an actual, written checklist with reminders to check all the common spots where errors pop up. Once the thing is done, run through the checklist even if you feel confident. It's an old proofreading trick, but I often do my checklist in reverse order so I start at the bottom of the page or the end of the document and work backwards. You catch more things that way.

Also, fake deadlines. Whenever I have a deadline to deliver a project, my real deadline is X days before that, depending on the complexity of the project. Then I have time to take a break from it and review it with fresh eyes before I turn it in.

And yes, a second pair of eyes is always helpful!
posted by peanut_mcgillicuty at 6:54 PM on January 4, 2017 [1 favorite]


I suppose it sort of matters what kind of calculations and data you're working with. I tend to gloss over important details when it comes to math like leaving out negative signs or moving decimal points. I do the same with writing and often leave out important words in sentences. With the calculations I work with I can usually test whether or not I've made a mistake by graphing the results. I have a pretty good idea of what the results should look like beforehand. I'll usually do this every few steps of the process rather than at the end because by then there are so many intermediate calculations that it's impossible to track down the culprit. Fortunately my results have to obey the laws of physics, so it's pretty obvious that if my algorithm says that 100 electrons are produced for every incoming photon that something has gone very wrong.

I'm still improving the way I work but I've stolen quite a few good ideas from my colleagues including the following (some of these are repeats of what others have already said):

- Break out each step into a separate part. If you're working in Excel then have each major calculation step be in it's own column or cell. Label each one so you know what it's supposed to be doing.

- If you have an idea of what your result should look like continually check that your result makes sense along the way rather than at the end.

- If you're using data processing algorithms that you've written yourself or are repurposing someone else's code make sure it lets you know what it's doing (via print statements or something similar) so it's not a black box. (Don't do what I do and change the code without changing those print statements. Much confusion.)

- Be careful about copy/pasting and autofill. I've found major problems were because I copied the wrong cell or because Excel didn't autofill in the way I expected it to.

- If you're doing calculations by hand or with a calculator I think the above still applies. Do each step separately. Try to keep things neat. Having just completed a whirlwind review of math topics for physics space engineering this past semester I found that cluttered pages and calculator screens were my number one worst enemy.

- And yes, if you have a friend, colleague, or mentor who doesn't mind checking your work that's always a good idea.

- Also, if you're like me and you might have undiagnosed adult ADHD try not to listen to or watch content with words or lyrics in them while you're doing work that requires attention or accuracy. Took me 30+ years to figure out that watching TV while doing homework was messing me up.
posted by runcibleshaw at 8:41 PM on January 4, 2017


Honestly, learn how to automate your work.

Either get good in Excel or have someone else help you make them, but have spreadsheets created with formulas for things you calculate often. Have one variable cell where you enter new data, leave the static bits alone, and voila, correct calculations spit out.

If you have repetitive enough things going on (i.e. payroll, profit reporting, etc) you can export reports and write macros for them, or make giant spreadsheets that do all the work for you.

Also, document best practices. While you are doing a thing you tend to mess up on, have a document open where you write out each step in easy-to-understand terms (doesn't have to be layman's terms, but close). Then the next time you do that task, open the document and follow along. Is there something that could have been described better or explained more clearly? Use screenshots and pictures if you need to, and include formulas or calculations. Once it's good, use it on the daily. It'll keep you as accurate as the standards you set out.
posted by rachaelfaith at 5:23 AM on January 5, 2017


I do a lot of data work and occasionally discover that I've made nutty mistakes. The way I've addressed it is by doing any important work twice. I do it once the most obvious way, set it aside for a day or so, and then do it again using a different method. So, for example, I might write R code to do a bunch of data analysis. Then, a day later, I'll stick the raw data in excel, clean it up, and at least run the most basic stats in excel to make sure the results match. For really important projects, I'll ask a colleague to try to replicate a few of my results.

I also often ask my colleagues to let me walk them through my spreadsheets/calculations. Having to explain each step to someone really helps show you where you're doing something weird or sub-optimal.
posted by snaw at 8:35 AM on January 5, 2017


« Older Emily Dickinson filter: Remember content but not...   |   Help me pick a cornet mouthpiece Newer »
This thread is closed to new comments.