Data cleaning

[Mubanga Mulenga]

Hi members,

i am new around here and a novice with research. i am trying to conduct an Interrupted time series analysis for my paper using monthly proportionate data of skilled births for a period of four (4) years. However, my data set has some proportions greater than 1. How do i best deal with this?

thanks in advance.

[Clyde Schechter]

Well, on its face, a proportion greater than 1, assuming you mean the proportion of all births that are skilled, is a data error. It is an obvious data error, and it raises the question of whether there may be other errors in the data that are not obvious. No attempts at analyzing bad data can give you useful results. So this needs careful investigation.

You need to confer with whoever provided you with this data set and review the process by which it was created. The source of the data should be given the examples of obvious errors, and any other highly suspicious errors you may find. Then they can review what happened and attempt to fix it. Or perhaps you (and I) are misunderstanding the meaning of the variable and there is a good explanation of why these values are correct--the variable means something different. In that case, there may be no problem at all and you will just have to think about an appropriate analysis of this kind of variable.

[Mubanga Mulenga]

Thankyou Clyde,

And you are right, i meant births that are skilled. i guess i have to follow it up with the source where i obtained the data from.

Thanks so much.