1

I was wondering if it makes sense to run a linear regression when my dependent variable only has 20 observations in total (I do have it for 6 years).

This is the scenario: 20 observations consists which is one observation for every neighborhood in a certain city. Furthermore, I have 6 independent variables with just as many datapoints (which makes sense since I have data for each neighborhood). There is really also no way to increase the datapoints, unless I look at street level data (which I do not have).

I was wondering if it is still possible (since it does represent each part of the city) to draw a useful conclusion from said linear regression.

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
Frank Li
  • 11
  • 1
  • The traditional ["magic" number](https://www.researchgate.net/post/What_is_the_rationale_behind_the_magic_number_30_in_statistics) of minimum observervations is 30, however, please advise your dataset and reproducable code for a better analysis. –  Jun 29 '20 at 15:54
  • I would like to, but sadly I am not allowed to share the dataset.. However it is a fairly simple dataset: 20 areas (which are all the areas in the city), and for each area, there is additional data for the (in)dependent variables. Another option could be aggreagating the data from the last six years, but that would still leave me with 20 observations in total. –  Jun 29 '20 at 15:57
  • 4
    That magic number 30 has something to do with the Central Limit Theorem. As such, it has absolutely no relevance with regression problems. – Michael M Jun 29 '20 at 16:15

0 Answers0