SlideShare a Scribd company logo
DFMW – Homework 1Spreadsheet ErrorsApril 5, 2010Motoki Bandaimotoki_bandai@mba.berkeley.edu
Since 1995, studies show that up to 90% of spreadsheets contain some type of error2Select spreadsheet error studiesAll sizeable, real-world spreadsheets with many cells, rows, and/or formulas
In the 2nd model of Lukasic’s study, the investment value was overstated by 16%
Lawrence & Lee’s study used the 30 most financially significant spreadsheets audited by Mercer. All had errors.Spreadsheet errors are abundant. Even the best make mistakes and some are extremely serious
We are not perfect at catching these errors, even if we had the chance to review the spreadsheets3Select code inspection experimentsIn these experiments, subjects examined spreadsheets that had several errors
The subjects reviewing the spreadsheets were significantly less than perfect in detecting errorsNot only do we make mistakes in our spreadsheets, we are only moderately good at finding them

More Related Content

PPTX
Machine Learning Vital Signs
PPTX
Spreadsheet Errors
PDF
A/B Testing myths and quagmires
PDF
A/B Testing: Common Pitfalls and How to Avoid Them
PDF
Data Curation - Data probity in a time of COVID
PPTX
Dfmw Spreadsheet Errors Presentation Jake Carney
PDF
Conversion Hotel 2018 Keynote: Lizzie Eardley
PDF
Decisions, decisions (#ovopdc 26/03/2019)
Machine Learning Vital Signs
Spreadsheet Errors
A/B Testing myths and quagmires
A/B Testing: Common Pitfalls and How to Avoid Them
Data Curation - Data probity in a time of COVID
Dfmw Spreadsheet Errors Presentation Jake Carney
Conversion Hotel 2018 Keynote: Lizzie Eardley
Decisions, decisions (#ovopdc 26/03/2019)

What's hot (7)

PDF
Rekard Edgren - Curing Our Binary Disease - EuroSTAR 2012
PDF
[CXL Live 16] How to Utilize Your Test Capacity? by Ton Wesseling
 
PPTX
State of selenium survey results
PPTX
Testing for everyone agile yorkshire
PDF
From Lag to Lead: Actionable Analytics
PDF
Transtech presentation: Advances in computer vision technology and analysis
PPTX
Exploratory testing part 1
Rekard Edgren - Curing Our Binary Disease - EuroSTAR 2012
[CXL Live 16] How to Utilize Your Test Capacity? by Ton Wesseling
 
State of selenium survey results
Testing for everyone agile yorkshire
From Lag to Lead: Actionable Analytics
Transtech presentation: Advances in computer vision technology and analysis
Exploratory testing part 1
Ad

Similar to Spreadsheet Errors Mb (20)

DOCX
Discussion Questions Chapter 15Terms in Review1Define or exp.docx
DOCX
Page 1 of 2 Introduction to Management Assignment 1 Ca.docx
PPTX
Spreadsheet Errors
PPTX
Spreadsheet Errors
PPTX
Spreadsheet Errors
DOCX
Quantitative Critique Rubric- 5.25.2020Student Name .docx
PDF
Simulation pitfalls p302023
PDF
Customer retention measurement.
PDF
Introduction to Software Requirement Engineering.pdf
DOCX
Unit VIII Course ProjectUniversity Case StudyCongratulatio.docx
PDF
Constraint Management
PPTX
Manufacturing in Northeast Ohio: Where We Stand, Where We’re Headed
PPTX
Applying lean thinking in the food supply chain presentation
PDF
Essay Verbs Discuss. Online assignment writing service.
PDF
How to Ruin your Business with Data Science & Machine Learning by Ingo Mierswa
DOCX
What does a lims really cost
PPTX
Case study 1
PDF
InfinityQS_7 Habits of Quality Obsessed Manufacturers_Print final
PPT
Martin Koojj - Testers in the Board of Directors
PDF
Modelling for decisions
Discussion Questions Chapter 15Terms in Review1Define or exp.docx
Page 1 of 2 Introduction to Management Assignment 1 Ca.docx
Spreadsheet Errors
Spreadsheet Errors
Spreadsheet Errors
Quantitative Critique Rubric- 5.25.2020Student Name .docx
Simulation pitfalls p302023
Customer retention measurement.
Introduction to Software Requirement Engineering.pdf
Unit VIII Course ProjectUniversity Case StudyCongratulatio.docx
Constraint Management
Manufacturing in Northeast Ohio: Where We Stand, Where We’re Headed
Applying lean thinking in the food supply chain presentation
Essay Verbs Discuss. Online assignment writing service.
How to Ruin your Business with Data Science & Machine Learning by Ingo Mierswa
What does a lims really cost
Case study 1
InfinityQS_7 Habits of Quality Obsessed Manufacturers_Print final
Martin Koojj - Testers in the Board of Directors
Modelling for decisions
Ad

Spreadsheet Errors Mb

  • 1. DFMW – Homework 1Spreadsheet ErrorsApril 5, 2010Motoki Bandaimotoki_bandai@mba.berkeley.edu
  • 2. Since 1995, studies show that up to 90% of spreadsheets contain some type of error2Select spreadsheet error studiesAll sizeable, real-world spreadsheets with many cells, rows, and/or formulas
  • 3. In the 2nd model of Lukasic’s study, the investment value was overstated by 16%
  • 4. Lawrence & Lee’s study used the 30 most financially significant spreadsheets audited by Mercer. All had errors.Spreadsheet errors are abundant. Even the best make mistakes and some are extremely serious
  • 5. We are not perfect at catching these errors, even if we had the chance to review the spreadsheets3Select code inspection experimentsIn these experiments, subjects examined spreadsheets that had several errors
  • 6. The subjects reviewing the spreadsheets were significantly less than perfect in detecting errorsNot only do we make mistakes in our spreadsheets, we are only moderately good at finding them
  • 7. Case Study: the simplest of mistakes can cause a big difference in the end result4CorrectInput errorActual error made in the firm I used to work forFX calculation was correct. Simple input errorCaused drastic difference in results. ~20% more products deemed overpriced compared to US, France, Germany, UKWhat made it worse: we sent the model with the error to the client for interim review and they caught itLooking back at our old spreadsheets, turns out ~90% of our spreadsheets had errors – some minor, some significant. This time was the first time client picked out a mistake
  • 8. Why do we make so many mistakes and why don’t we catch them?A study conducted by Hall (1996) concluded:Spreadsheets are important and complexBut used only limited safeguards…And used limited data testingInspection was also limited18% of those surveyed say their project was rushedCaused by corporate policy?A minority of companies surveyed have development rules and guidelines, and only a minority fully follow themCaused by overconfidence?Spreadsheet developers were found to be very confident with their spreadsheets, even while their work contained errorsOverconfidence tends to blind people to the need to take steps to reduce risks5
  • 9. What we did6Spreadsheet createdSend to clientManaged to still senderror prone models