0

I have two variables A and B of interval scale type.

A is the birth year of a person and B is the number of years spent at university.

What method would you recommend to measure the correlation?

mpiktas
  • 33,140
  • 5
  • 82
  • 138
Raffael
  • 1,424
  • 3
  • 18
  • 30
  • What is the purpose of this exercise? – mpiktas Dec 20 '11 at 10:09
  • to help out a friend – Raffael Dec 20 '11 at 10:14
  • Well helping friend is a noble thing to do, but the question should adhere to FAQ. In my opinion it does not. Hence the downvotes. The objection is the same as in previous question. – mpiktas Dec 20 '11 at 10:18
  • but isn't t that the question of what method to choose for calculating the correlation is not just depending on the analysis afterwards but also (and that is what my questions are directed at) to the nature of the data itself? So an answer or educated guess would be possible ... !? – Raffael Dec 20 '11 at 10:21
  • 4
    "Nature of the data" generally provides only rough (or even misleading) guidance for data analysis, Raffael. It is far more important to know the purpose of the analysis: what decisions or actions will be taken as a consequence? Second in importance is to know something of the statistical characteristics of the data, such as their univariate and bivariate distributions. Without any of this information available in your question, we can only provide guesses and generalities. This makes it difficult to be useful to you or subsequent readers interested in this topic. – whuber Dec 20 '11 at 11:44

1 Answers1

5

If you would like to see correlation between birth cohorts and amount of schooling take the year and the number of years only. would that be a solution? Just like this:

cor(year_of_birth, years_in_university)

Edit

I conclude from your comments you want to choose between Spearman/Kendall/Pearson. There are numerous questions concerning this topic (e.g choosing between Spearman and Pearson)

Basically the difference is that Pearson measures a linear relationship while Spearman measures monotonic relationship. In most cases however they will be pretty similar. If you read the article referenced - you will see that you gave your answer in your question. For interval scaled variables the Spearman method is more appropriate.

Community
  • 1
Seb
  • 571
  • 3
  • 15