Comments on: Insurance industry reports abortion-breast cancer link

By: Marginalized Action Dinosaur » Abortion, ager at first birth, childlessness, fertility, breast feeding, the pill, hormones

Tue, 20 Nov 2007 03:52:57 +0000

[...] http://magicstatistics.com/2007/11/13/insurance-industry-reports-abortion-breast-cancer-link/ [...]

By: StatGuy

StatGuy — Mon, 19 Nov 2007 05:27:22 +0000

dinosaur, I am very sorry to hear about your wife.

By: dinosaur

dinosaur — Mon, 19 Nov 2007 04:29:53 +0000

My wife has breast cancer and they are among the first questions they ask. Beat you by 3 days, unfortunately,

By: Thinking Christian

Thinking Christian — Thu, 15 Nov 2007 17:30:08 +0000

Abortion Strongly Linked to Breast Cancer…

Sure, it’s correlational data?but with a .98 correlation coefficient! … Meanwhile, the American Psychological Association is still updating its fact sheet on abortion and mental health….

By: Paul

Paul — Wed, 14 Nov 2007 16:56:59 +0000

This kind of problem is very common when working with time series data. I think the safest procedure is to begin with the null that each series contains a unit root. This can be tested against the alternative that they are stationary around a linear trend.

By: StatGuy

StatGuy — Wed, 14 Nov 2007 14:09:04 +0000

Why do you suspect that both series have a unit root?

By: Paul

Paul — Wed, 14 Nov 2007 08:59:43 +0000

The graph suggests a possible link but the evidence isn’t conclusive. Correlation coefficients greater than 0.9 are not unusual with time series data that contain trends.

I suspect that both these series have a unit root (even after allowing for the trend). Given this, it is necessary to test if they cointegrate before any conclusion can be drawn as to whether there is a long term relationship between them.