I'm trying to work out how to test if there is a significant difference between years when looking at crime data. The data file has no numerical values, only states the neighborhood, month and year of each crime, so I have created a table to count the number of crimes in each year using t() and data.frame in R.

2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  2013  2014  2015  2016 
55146 54459 49662 46728 41991 39162 35579 33343 31957 33118 33512 36631 37509 7573

Is there a way I can test if any of these are significantly higher/lower than the rest?

  • 1
    $\begingroup$ I don't know what your actual research question is, but I suggest you look into intervention or change point analysis for count time series. $\endgroup$
    – Roland
    Commented Oct 7, 2016 at 11:33
  • $\begingroup$ No matter what you do in the end, make sure to remove 2016 as it doesn't have complete data (maybe also 2003 if it was not measured from the beginning of the year). $\endgroup$ Commented Oct 9, 2016 at 10:53

1 Answer 1


You can do regression with the count as the dependent variable and year as the independent variable. You will still need to check the assumptions of regression models, but the fact that the DV is a count is not, in itself, a reason not to do linear regression, unless the counts were quite low.

  • 2
    $\begingroup$ If doing regression model, than perhaps it might be prudent to control for population growth. $\endgroup$ Commented Oct 9, 2016 at 10:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.