Friday, June 26, 2009


When is a lot of variability too much? How precisely, quantitatively reproducible should biological signaling activation be? If you have different populations of the same cell line, treated all the same way, does it make sense if they still show about 20-40% variance in their response over the course of a bunch of replicates?

How do I know when it's us and our hands, or when it's just the fundamental variability of biology???

UPDATE: Thank gawd for sisters who understand ANOVA.


Comrade Physioprof said...

You never really know that there isn't some unknown source of variability that you *could* be controlling, but aren't, as opposed to intrinsic variability of the biological system. What is important from an experimental standpoint is the ratio of variability between replicates within an experimental condition to that between experimental conditions. This is exactly what is assessed by parametric statistical comparisons like t-tests and ANOVA. The greater your within-condition variability is, the larger the between-condition differences need to be to be statistically detectable.

Arlenna said...

I wish I knew how to run the t-tests and ANOVAs properly in Prizm. I figured out where the menu optino is to run the analysis, but I don't know what parameters to select or input in order to make sure I am running the tests on the relevant columns and rows...

Illyria said...

can I leave a comment?

Arlenna said...

Apparently you can!

Illyria said...

Dear Arlenna,

I didn't know in what other way to get in contact with you, so I figured this is the best way. I responding to a comment you made on this blog:

In which you responded to my e-mail and my intelligence as:

" Wow, that is pretty much the best book review I have ever read!! I am keeping a copy of that baby for the files.
I kind of want to say something nasty about the intelligence of the person writing that email compared to most others regardless of whether they are a man or a woman… but I guess it really is unfair. No, men are not inherently smarter than women, and any intelligent person taking ten minutes to read that “study” and even do a lazy thought experiment on the issue should be able to discern that an attempt at a truly unbiased assessment of the issue of demonstrated “accomplishments” and “advances” may be contaminated by the historical situation our culture has been in since time immemorial. Like others have implied above, given its methodology and the reputation of the “scientist” who performed it, the linked study is about as reliable as an opinion poll on the matter."

To be honest, I was surprised that it all was given any type of attention by Sheril, as I recieved no e-mail reply, and merely figured it would wind up in the bottom of the mail box. I know it comes off as a stupid question, and I perhaps come off as very young, yes, I do have a lot of problems with that particular study, but when I hear over and over that "men are smarter" that "men make civilization" etc. I start to feel inadequate. I guess I just contacted Sheril because I desperately wanted to hear that, hey, what you are being fed is wrong, don't think that way. In addition, I asked someone else about it, who pointed out numerous problems with it, including the G factor.

Well, I guess that's all I have to say, but I hope you don't think I'm stupid or that I was trolling.

Have a nice day.

Arlenna said...

Well I really hope from all those comments validating that that information was wrong that you can feel less inadequate! The energy and frustration I expressed by being snarky, as did many others, comes from the anger that anyone would have to feel inadequate from the outside pressure and bogus studies like that guy's crappy pseudoscience.

Like I said in my comment there, while my initial reaction was to be like "WOW who the hell would think that?", I recognized it would be unfair to make a nasty comment on the intelligence of the person writing the email, since clearly if you felt that way it was because something and somebody was majorly affecting your confidence.

I hope you were able to take the replies to your message as discussed in that thread as confidence-creating, since a lot of people stood up to say that those 'studies' from that guy are a bunch of crap, and look beyond the internet dramatic-flair-snark with which we all replied!

Comrade Physioprof said...

I am not specifically familiar with "Prizm"--whatever the fuck that is--but if you want to consult with me via e-mail, I'm happy to try to help you.

Anonymous said...

I'd suggest spending some time reading through the Prism statistics guide (PDF accessible thrrought the help menu). It's a nice overview, especially if that's the program you'll be using.