0

I have a data set for a PhD paper in which I need to look for differences between 3 groups of observations. The issue is that ~95% are in one group and the rest are in the other two. The data table is attached below. Will any research on the differences between the three groups is valid from a formal statistical point of view?

    group 1  group 2
  1 402286   383596
  2   8523   6107
  3  15112   8535
Michael R. Chernick
  • 39,640
  • 28
  • 74
  • 143
user49422
  • 173
  • 2
  • 11
  • 1
    Possible duplicate of [When is unbalanced data really a problem in Machine Learning?](https://stats.stackexchange.com/questions/283170/when-is-unbalanced-data-really-a-problem-in-machine-learning) – AdamO Jul 21 '17 at 16:53
  • Not at all, I'm asking about descriptive statistics for research and not about the use of algorithms to solve such issue. – user49422 Jul 21 '17 at 16:56
  • 1
    Perhaps more description of the problem is needed. I do not see why you believe this would raise an issue. You have over 6,000 obs in the sparsest group. It's hardly an issue. – AdamO Jul 21 '17 at 16:59
  • I agree with you @AdamO, I just want to see if someone with more academic background will think differently. – user49422 Jul 21 '17 at 17:12
  • 1
    I apologize for my academic background being insufficient to your present needs. Good luck with the PhD and beyond. – AdamO Jul 21 '17 at 17:13

1 Answers1

1

Typically, you can measure treatment differences for two or three treatments in unbalanced data. Correct me, if I am wrong but theoretically I don't see a problem. However, it is worth pondering over if comparing the treatment difference between when the data is not imbalanced, qualitatively makes sense.

Also, do you know what specific statistical method do you plan to use for your problem?

user3373954
  • 21
  • 2
  • This data is about measuring basic statistics metrics like quantiles, mean, median etc,.Later on I'm going to use CART to look for variable importance for my main research question. – user49422 Jul 21 '17 at 17:15