I have data of production factors and climate. I want assess correlations between them. As data shows non-stationary, shall I detrending or differencing data prior to correlation analysis?
Dear Bahri
The answer is one. You have to make them stationary before testing the correlations
Since you are going to use the result for policy implication. The series must be stationary.
If the variables are non-stationary then it is necessary to make them stationary for meaningful analysis. If you check correlation in non-stationary series then it would be misleading because of time trend it would be upward bias. Thus, it is necessary to make the series stationary.
The solution to the problem is to transform the time series data so that it becomes stationary. If the non-stationary process is a random walk with or without a drift, it is transformed to stationary process by differencing.
- Badges
- Science topic