4

Preface

I work at an ecommerce saas company. I have been asked to perform an analysis on the relationship between behaviors of potential customers during free trials and their conversion to paying customers.

In analyzing the data, I have learned that our data collection is systematically NOT collecting certain behaviors of some prospects/customers.

I have communicated this bias in the data to my stakeholders, but I have been unable to convince them of the risks of making decisions based on analyses performed on bad data. I think part of the issue is that they believe that examples that are not in the business domain are inapplicable.

Question

What is a recent (i.e. last 10 years) example of negative consequences of stats/data science as a result of using bad data?

I am seeking a reference (article, white paper, journal publication, etc.) describing a recent example in the business/commercial domain in which the consequence of using flawed data was a significantly negative business outcome. Here I am defining "bad data" as data that misrepresents reality due to collection failures.

Things I have already referenced, but which do not answer my question:

xibalba1
  • 101
  • 3
  • I believe this question was incorrectly put on hold by @whuber. Per the "How to Ask" link, this question **is** 1: researched (see hyperlinks at bottom of question), 2: On topic (it's a stats question), 3: **It IS specific**, if you disagree, please explain how it violates this criterion. I provide both context and specifics on the type of answer I'm seeking, and 4: massively relevant to analysts/statsticians, and DSes working in industry. – xibalba1 Jun 03 '19 at 16:52
  • There's no problem with research, specificity, or appropriateness. I'm afraid that questions soliciting lists of answers tend not to be on topic here (although long in the past such questions did appear). The principal exceptions are focused questions requesting reference material. – whuber Jun 03 '19 at 17:24
  • @whuber, the question does not solicit a list/lists of answers. I'm not sure where you read that in my question as written. Perhaps you could cite the confusing excerpt of my question so I can edit for clarity? I am, in fact, seeking references, (eg an article, white paper, academic publication, etc.) which provides example(s) of where bad data led to negative consequences in a commercial setting. Separately, I would suggest that if a question is put on hold for criteria not specified in the "How To Ask" link provided in the "put on hold" text, that the rationale be explained via a comment. – xibalba1 Jun 03 '19 at 18:12
  • "What are some recent (i.e. last 10 years) examples of negative consequences" is in the plural; that is what I took to be an explicit request for multiple answers. Calling such examples "references" doesn't make the question comparable to those that seek reference material. – whuber Jun 03 '19 at 18:31
  • 1
    Understood. I have re-worded the question per your feedback. Is the question now acceptable, or are there other problems? I am not trying to be a nag here, but I really do believe that this question does (or can with edits) meet cited criteria, is of benefit to others, and I don't know where else to turn for an answer. – xibalba1 Jun 03 '19 at 18:42
  • 1
    I appreciate your efforts and am sympathetic with the motivation and with the difficulties involved in framing this kind of inquiry in a way suitable for an SE site. After searching for clear guidance in our help and Meta pages, I haven't found anything that is unequivocal concerning whether your question is or is not on topic, so I am voting to reopen it (+1). – whuber Jun 03 '19 at 18:44
  • 1
    suggested other sites : quora reddit/statistics/datascience – seanv507 Jun 03 '19 at 18:47
  • whuber, thank you for your time, consideration, and re-opening the question. @seanv507, those are good suggestions, I will "cross-post" to reddit and quora, and look into cross posting to other SE networks (was under the impression that sort of thing was frowned upon). – xibalba1 Jun 04 '19 at 13:43

0 Answers0