Groundwater study reveals ‘bad data is good data’

Thursday, September 5, 2019

Researchers at Waterloo Engineering have uncovered a problem that could skew the results everywhere groundwater levels are monitored and used to set government policies.

Their study revealed that a discrepancy between scientific data and anecdotal reports on groundwater levels in southern India was caused by a statistical phenomenon known as ‘survival bias.’

Nandita Basu

Nandita Basu is a professor of civil and environmental engineering.

In effect, data on water levels in bad wells tended to be excluded from analysis as unreliable because it was incomplete, the result of periods when they had actually run dry.

When the excluded wells were added back into the mix, the results showed groundwater levels were declining in south India, not rising as analysis initially suggested.

“Our main point is that bad data is good data,” said Nandita Basu, a civil and environmental engineering professor. “When you have wells with a lot of missing data points, that is telling you something important. Take notice of it.”

Click here for the full story.

  1. 2019 (103)
    1. September (9)
    2. August (11)
    3. July (12)
    4. June (10)
    5. May (14)
    6. April (12)
    7. March (19)
    8. February (10)
    9. January (6)
  2. 2018 (100)
    1. December (6)
    2. November (13)
    3. October (10)
    4. September (6)
    5. August (8)
    6. July (7)
    7. June (11)
    8. May (9)
    9. April (7)
    10. March (14)
    11. February (5)
    12. January (4)
  3. 2017 (124)
    1. December (3)
    2. November (8)
    3. October (14)
    4. September (15)
    5. August (9)
    6. July (9)
    7. June (15)
    8. May (12)
    9. April (7)
    10. March (14)
    11. February (7)
    12. January (11)
  4. 2016 (97)
  5. 2015 (95)
  6. 2014 (133)
  7. 2013 (124)
  8. 2012 (117)
  9. 2011 (95)

Media Contact

Carol Truemner, Communications Officer (email | x33470)