What's the best method to identify if an external event has changed the distribution of my data?

10 October 2019 4 3K Report

Hi,

I'm analyzing a dataset with the yearly concentration indexes of 50 different sectors (e.g., automobiles, pharmaceuticals, etc.) of a specific country from 1996 until 2016.

However, I recently realized that the government of this country changed this classification from 2007 to 2008 and some companies that were classified in a sector were relocated to a different one. For example, companies that provide reparation services for a chemical industry used to be classified under the sector CHEMICALS, and now they are under the sector SERVICE TO INDUSTRIES. Consequently, the concentration indexes of some sectors were affected.

In my analysis, I want to get rid of all companies from any sector that presented significant changes in their concentration indexes from 2007 to 2008. Hence, my question is: how can I test if there was a significant change in the concentration indexes from 2007 to 2008? I thought about running a t-test comparing the indexes from 96 - 07 to the indexes from 08 to 16 for each sector, but I'm not sure if this is the best way to test, especially considering that it's common to have trends in the concentration.

Any help would be very welcomed!

Thank you,

Best,

Marco

David L Morgan

Given that you a series of data points with a know break-point, you can look into what is known as a regression discontinuity design.

Ma Ga

Thank you for your quick reply, David L Morgan !

I'm not much interested in understanding the effects of the external event. Right now I'm just cleaning the data, so I can use the concentration indexes as independent variables in our future analysis. Would the regression discontinuity still be applicable in that case?

If you think so, please, let me know and I will read more about it!

Thanks!

David L Morgan

You could examine the residuals from the regression discontinuity analysis for outliers, which would show you which cases changed much more (or less) than expected.

What's the best method to identify if an external event has changed the distribution of my data?

How to compare two groups with only two measurements?

Enhanced Yellow Fluorescent Protein (Aequorea victoria) DNA sequence?

What kind of fluid could I use for a pitfall trap that also does not invalidate molecular testing?

What is your perception on using visual analysis or quantitative approaches to evaluate data from single-case research designs (n of 1 trials)?

What's the best way to measure growth rates in House sparrow chicks from day 2 to day 10?

Do we need to write protocol for a second study of a systematic review that has been registered and published?

How to increase a model accuracy ?

No title

What is the production process of CaFe ferroalloys?

Do I need to do linear regression before running mediation analysis using PROCESS MACRO by Hayes ?