Property boom

[janey]

What a sweeping and wrong generalisation

woud you like to qualify your statement?

[Bagpuss]

woud you like to qualify your statement?

The medias reporting of statistics is often wrong. Most of the maths and methods behind what is oft misreported are correct.

Reminds me of an audience member on Questiontime the other week on bird flu. "Why are we looking at statistics, surely we should be looking into the numbers". FFS!

[janey]

The medias reporting of statistics is often wrong. Most of the maths and methods behind what is oft misreported are correct.

Reminds me of an audience member on Questiontime the other week on bird flu. "Why are we looking at statistics, surely we should be looking into the numbers". FFS!

I agree, in part

I'm sure they have used the methods and calculations correctly, however my issue with statistics is that you can use various methods on a data set and each one will prove something different... what I mean is that if you had 3 methods and used them on the same data the follwoing can potentially happen:

1. method one - proves your hypothesis
2. method two - doesn't prove or disprove your hypothesis
3. method three - disproves your hypothesis

obviously method one is always going to be the one that is reported on... be it in the media or any other publication... because it says "told you we were right"

[Bagpuss]

I agree, in part

I'm sure they have used the methods and calculations correctly, however my issue with statistics is that you can use various methods on a data set and each one will prove something different... what I mean is that if you had 3 methods and used them on the same data the follwoing can potentially happen:

1. method one - proves your hypothesis
2. method two - doesn't prove or disprove your hypothesis
3. method three - disproves your hypothesis

obviously method one is always going to be the one that is reported on... be it in the media or any other publication... because it says "told you we were right"

Very unlikely the situation you describe would ever happen. In any credible research you specify what you are going to do up front and justify why you are doing it. Usually there is supporting literature for doing it the way you have chosen. What you describe sounds like data dredging and/or someone who has no idea what they are doing.

[janey]

Very unlikely the situation you describe would ever happen. In any credible research you specify what you are going to do up front and justify why you are doing it. Usually there is supporting literature for doing it the way you have chosen. What you describe sounds like data dredging and/or someone who has no idea what they are doing.

erm well I know for a fact that is exactly what happens.

the author would be more than capable of justifying why they have used the methods however what concerns me is what they don't tell you... i.e. all the calculations they used which didn't prove anything or more importantly actually completely disproved what they were claiming.

[sasguru]

I agree, in part

I'm sure they have used the methods and calculations correctly, however my issue with statistics is that you can use various methods on a data set and each one will prove something different... what I mean is that if you had 3 methods and used them on the same data the follwoing can potentially happen:

1. method one - proves your hypothesis
2. method two - doesn't prove or disprove your hypothesis
3. method three - disproves your hypothesis

obviously method one is always going to be the one that is reported on... be it in the media or any other publication... because it says "told you we were right"

If you provide a CI or p-value for each of the methods, that would enable you to decide which is more likely.

[Bagpuss]

erm well I know for a fact that is exactly what happens.

the author would be more than capable of justifying why they have used the methods however what concerns me is what they don't tell you... i.e. all the calculations they used which didn't prove anything or more importantly actually completely disproved what they were claiming.

Which industry does this happen in? My guess is you are talking about marketing or social sciences. Because any kind of regulated environment for example medical research, it would be thrown out. You cannot throw things out you have specified up front, you report the good with the bad.

[sasguru]

Which industry does this happen in? My guess is you are talking about marketing or social sciences. Because any kind of regulated environment for example medical research, it would be thrown out. You cannot throw things out you have specified up front, you report the good with the bad.

Indeed. In Clinical Trials, the requirements are stringent, for obvious reasons.

[janey]

Which industry does this happen in? My guess is you are talking about marketing or social sciences. Because any kind of regulated environment for example medical research, it would be thrown out. You cannot throw things out you have specified up front, you report the good with the bad.

I was actually talking about financial research.

[meridian]

on average 40% (or nearly half) of all "sick" days are taken on either a Monday or a Friday