BBC news

[emony]

I may be misunderstanding, because me and science and statistics have never been on very close terms, but as far as I can tell, this story, about how1 in 25 fathers may be unknowingly raising another man's child, is based on data gathered from people who had paternity tests done, right? So for a fair number of them, there were doubts in the first place, or they wouldn't have been testing. And .. no one thought that perhaps, just perhaps, that might skew the results a tiny bit?

ETA: Also, for those who are interested - the pilot of Lost got over 6 million viewers last night. Rock on.

Comments

[davechicken.livejournal.com]

Lies, damned lies and statistics.

Take the right sample or draw the right graph and you can prove ANYTHING.

Story is better than truth, it seems.

[emony.livejournal.com]

I imagine the scientists' report had tonnes of disclaimers and stuff all through it about what it meant, and the press department of John Moores and then the Journos went ".. 1 in 25 - that sounds good!" and made it into a story, hoping that no one would notice. Silly journos.

[davechicken.livejournal.com]

Yep. Precisely. *eyeroll*

[kangaruth.livejournal.com]

I had a statistics class last year, and if it taught me anything, it's that statistics don't mean anything if they aren't in context, or are used by people who don't understand what they're doing. Like the 1 in 7 million chance of two siblings dying of cot death. And there are certain studies in which a cancer treatment is shown to be helpful for men with cancer, and helpful for women with cancer, but not for people with cancer if gender isn't taken into account.

They probably include some kind of factor for the skew factor, but that won't necessarily reflect the actual situation.

However, I heard recently of an interesting study into the mitrochondrial DNA of Jewish people who believe they are directly descended from some Old Testament figure. It showed what appeared to be a surprisingly high level of infidelity.

[emony.livejournal.com]

Interesting. Maybe 1 in 25 isn't such an out-there figure. People are pretty untrustworthy in general, after all. I don't know, stats - I only did about two weeks worth of stats, and that was over five years ago. I can't even remember what we did, never mind how it was done. Like you say though, context is so important.

[niennah.livejournal.com]

It doesn't directly say that. It says that they base an estimate of lower than 10% on paternal testing and come to a figure of 4%, which is 1 in 25. Since, in various studies, it ranges from 1% to 30%, so 4% seems reasinable given those figures.

Anyway. Sorry for being argumentative, I'm not really. I'm just frustrated about being at work and only vaguely using my brain, you know how it is... lol. Smingeldy pingeldy.

[niennah.livejournal.com]

Since, in various studies, it ranges from 1% to 30%, so 4% seems reasinable given those figures.

Of course, that's nothing compared to the number of typos in that sentence. I appear to have the concentration span of a goldfish. Apologies.

[emony.livejournal.com]

Hehe *g* It's okay, I was asking because I know I'm useless at interpreting stats and I wanted to see if I was right. I do think the "it's somewhere between 1% and 30% .. how do you guys feel about, say, 4%?" method probably isn't the most scientific thing ever, but I'm sure there was some actual science and stuff involved when they did it in reality.

I didn't notice the typos, you know, until you pointed them out. But then, I was up until midnight watch Lost, so my eyes aren't that reliable right now. I was doing that micro-sleep thing on the train this morning. I swear, I nodded off at least three times, for about two or three seconds a time. Very odd sensation.

[gobsmackit.livejournal.com]

6 million! Hurrah! \o/