5 ms·

The title is a bit odd. The article details a sloppy use of substitute data from adjacent rows or even columns to fill gaps in data, excel has little to do with

The title is a bit odd. The article details a sloppy use of substitute data from adjacent rows or even columns to fill gaps in data, excel has little to do with it. The worst example is using United Kingdom data to fill gaps in USA data because they were adjacent in the list.

smt88 3y ago
Excel has everything to do with it. From the article: > In email correspondence seen by Retraction Watch and a follow-up Zoom call, Heshmati told the student he had used Excel’s autofill function to mend the data. The most charitable interpretation is that the professor has no idea what autofill is or isn't capable of doing, so he misused it.
- grey-area 3y ago
  That’s not even the worst case though, worse by far than interpolating (whatever method is used) is simply taking data from countries with a similar name to fill gaps.
  - InSteady 3y ago
    "What was your rational for using data from the Netherlands to fill in the missing values in New Zealand?" "Um, both countries start with the letter N"
    - jakeinspace 3y ago
      Should’ve used Denmark rather than Netherlands, as it’s the Old Zealand.
- kgwgk 3y ago
  Do _you_ have an idea of what autofill is and what is or isn’t capable of doing? When he dragged down a selection he _saw_ the values being copied or a trend being extrapolated.
  - LeifCarrotson 3y ago
    Flash Fill and Series Fill are great at recognizing simple patterns like 1,2,3,4,5 or 2,4,6,8,10 and can sometimes (unreliably!) recognize more complicated multiplications, repetitions or formulaic string modifications (like house numbers and street names). I use both frequently. But neither these tools nor an LLM has real understanding, they're not going out into the real world and collecting data. The results need to be verified. Automatically filling an index column with 1,2,3... is one thing, automatically filling a data column with guesses from pattern matching is different. This problem is only going to get worse as LLMs proliferate and can more reliably but still imperfectly and still opaquely fill in more complicated data.
  - smt88 3y ago
    I don't use autofill precisely because I have no insight into the algorithm behind it. I suspect it's very "stupid," but I don't know exactly how stupid it is. It sounds like he didn't realize or notice that autofill was using adjacent cells (from other countries, in this instance). That behavior would shock me as well. I thought autofill only took columns of data into account.
    - kgwgk 3y ago
      > It sounds like he didn't realize or notice that autofill was using adjacent cells (from other countries, in this instance). No, it doesn’t sound like that: “in several instances where there were no observations to use for the autofill operation, the professor had taken the values from an adjacent country in the spreadsheet.” It was not Excel’s fault.
- secondcoming 3y ago
  Right, but the mistake wasn't an issue with Excel, it was with the professor misusing it.
  - kurthr 3y ago
    Might as well blame ChatGPT for writing your term paper. If they'd been using R/Matlab/Python written by a copilot it would have been just as much gibberish.
  - smt88 3y ago
    Yes, I understand that and agree. The headline isn't blaming Excel. "Undisclosed tinkering in Excel" is not far from the accurate headline: "Misuse of commonly-used Excel tool..." The interesting thing is that a supposedly sophisticated person was using an unsophisticated tool and then relying on the data it produced. Failing to mention "Excel" in this case is doing a disservice to the thrust of the problem.
    - grey-area 3y ago
      The accurate headline would be incomplete data falsified. The tools have nothing to do with it, and the same could have been done with paper and pencil. Autocomplete is the problem, it just repeats a single number or a series, and when using it you see exactly what the results were. This was not done by mistake or in ignorance or because the tool encouraged them, it was done to knowingly falsify incomplete data so that they could finish their paper and publish.
scotty79 3y ago
Also Netherlands to New Zealand. Hilarious.
- bouncing 3y ago
  And United States <~> United Kingdom. To compare OECD countries. What’s the point of even doing the research?
  - geysersam 3y ago
    To write papers
  - fransje26 3y ago
    Personally, I prefer using data from Uruguay to cover the holes in the United States data.
riedel 3y ago
Such things are encouraged by the excel UI because it fails to make relationships and manual corrections easily visible. The human computer interface matters beyond UX.
- martynr 3y ago
  Surely this can’t be justified as a ux driven “oversight” - I am not an economics professor but I would be feeling totally sketchy if I manipulated a data source in such an apparently arbitrary manner. In combination with the (reportedly)blatant lack of any reference to the methodology in his paper this feels like wilful gross misconduct in professor-land. IMO this sort of thing rightly undermines confidence and credibility of research in a much broader context and should be a sackable matter if shown to be true as asserted.
- csydas 3y ago
  I get what you're saying, but I would disagree that the focus should be on Excel and its UI/UX. As annoying as Excel is when you're working with _specific_ data that suffers from Excel's attempts to be helpful, I'd wager for most use cases Excel is doing exactly what most users expect it to, or the change is benign at best for most use cases. That isn't to diminish when Excel's help has disastrous consequences, but rather to say it shouldn't be unexpected. If your data requires precision, there are better tools than Excel for working with your data and there should be steps taken by more experienced members of the team to ensure the team doesn't get bit by Excel's features. Excel's behaviors are well known and I'd have to imagine at least one member of any research team has been bitten by Excel's assistance features before during research. I would be surprised if any serious educational institution that teaches about ethics and common pitfalls in research *didn't* caution about how Excel might corrupt research [0]. 0 - Okay, maybe not _that_ surprised, but I would be disappointed and think less of the institution/instructor.
croes 3y ago
Excel makes such things convenient

smt88 3y ago

Excel has everything to do with it. From the article: > In email correspondence seen by Retraction Watch and a follow-up Zoom call, Heshmati told the student he had used Excel’s autofill function to mend the data. The most charitable interpretation is that the professor has no idea what autofill is or isn't capable of doing, so he misused it.

grey-area 3y ago
That’s not even the worst case though, worse by far than interpolating (whatever method is used) is simply taking data from countries with a similar name to fill gaps.
- InSteady 3y ago
  "What was your rational for using data from the Netherlands to fill in the missing values in New Zealand?" "Um, both countries start with the letter N"
  - jakeinspace 3y ago
    Should’ve used Denmark rather than Netherlands, as it’s the Old Zealand.
kgwgk 3y ago
Do _you_ have an idea of what autofill is and what is or isn’t capable of doing? When he dragged down a selection he _saw_ the values being copied or a trend being extrapolated.
- LeifCarrotson 3y ago
  Flash Fill and Series Fill are great at recognizing simple patterns like 1,2,3,4,5 or 2,4,6,8,10 and can sometimes (unreliably!) recognize more complicated multiplications, repetitions or formulaic string modifications (like house numbers and street names). I use both frequently. But neither these tools nor an LLM has real understanding, they're not going out into the real world and collecting data. The results need to be verified. Automatically filling an index column with 1,2,3... is one thing, automatically filling a data column with guesses from pattern matching is different. This problem is only going to get worse as LLMs proliferate and can more reliably but still imperfectly and still opaquely fill in more complicated data.
- smt88 3y ago
  I don't use autofill precisely because I have no insight into the algorithm behind it. I suspect it's very "stupid," but I don't know exactly how stupid it is. It sounds like he didn't realize or notice that autofill was using adjacent cells (from other countries, in this instance). That behavior would shock me as well. I thought autofill only took columns of data into account.
  - kgwgk 3y ago
    > It sounds like he didn't realize or notice that autofill was using adjacent cells (from other countries, in this instance). No, it doesn’t sound like that: “in several instances where there were no observations to use for the autofill operation, the professor had taken the values from an adjacent country in the spreadsheet.” It was not Excel’s fault.
secondcoming 3y ago
Right, but the mistake wasn't an issue with Excel, it was with the professor misusing it.
- kurthr 3y ago
  Might as well blame ChatGPT for writing your term paper. If they'd been using R/Matlab/Python written by a copilot it would have been just as much gibberish.
- smt88 3y ago
  Yes, I understand that and agree. The headline isn't blaming Excel. "Undisclosed tinkering in Excel" is not far from the accurate headline: "Misuse of commonly-used Excel tool..." The interesting thing is that a supposedly sophisticated person was using an unsophisticated tool and then relying on the data it produced. Failing to mention "Excel" in this case is doing a disservice to the thrust of the problem.
  - grey-area 3y ago
    The accurate headline would be incomplete data falsified. The tools have nothing to do with it, and the same could have been done with paper and pencil. Autocomplete is the problem, it just repeats a single number or a series, and when using it you see exactly what the results were. This was not done by mistake or in ignorance or because the tool encouraged them, it was done to knowingly falsify incomplete data so that they could finish their paper and publish.

scotty79 3y ago

Also Netherlands to New Zealand. Hilarious.

bouncing 3y ago
And United States <~> United Kingdom. To compare OECD countries. What’s the point of even doing the research?
- geysersam 3y ago
  To write papers
- fransje26 3y ago
  Personally, I prefer using data from Uruguay to cover the holes in the United States data.

riedel 3y ago

Such things are encouraged by the excel UI because it fails to make relationships and manual corrections easily visible. The human computer interface matters beyond UX.

martynr 3y ago
Surely this can’t be justified as a ux driven “oversight” - I am not an economics professor but I would be feeling totally sketchy if I manipulated a data source in such an apparently arbitrary manner. In combination with the (reportedly)blatant lack of any reference to the methodology in his paper this feels like wilful gross misconduct in professor-land. IMO this sort of thing rightly undermines confidence and credibility of research in a much broader context and should be a sackable matter if shown to be true as asserted.
csydas 3y ago
I get what you're saying, but I would disagree that the focus should be on Excel and its UI/UX. As annoying as Excel is when you're working with _specific_ data that suffers from Excel's attempts to be helpful, I'd wager for most use cases Excel is doing exactly what most users expect it to, or the change is benign at best for most use cases. That isn't to diminish when Excel's help has disastrous consequences, but rather to say it shouldn't be unexpected. If your data requires precision, there are better tools than Excel for working with your data and there should be steps taken by more experienced members of the team to ensure the team doesn't get bit by Excel's features. Excel's behaviors are well known and I'd have to imagine at least one member of any research team has been bitten by Excel's assistance features before during research. I would be surprised if any serious educational institution that teaches about ethics and common pitfalls in research *didn't* caution about how Excel might corrupt research [0]. 0 - Okay, maybe not _that_ surprised, but I would be disappointed and think less of the institution/instructor.

croes 3y ago

Excel makes such things convenient