From owner-drosophila@net.bio.net Sun May 05 23:00:00 1996
Path: biosci!news.Stanford.EDU!not-for-mail
From: adhar@leland.Stanford.EDU (Adam Noel Harris)
Newsgroups: bionet.drosophila
Subject: Re: Fly Crossing
Date: 6 May 1996 12:12:00 -0700
Organization: Stanford University
Lines: 38
Message-ID: <4mliu0$dqu@amy6.Stanford.EDU>
References: <318A7422.37B1@odyssee.net>
NNTP-Posting-Host: amy6.stanford.edu
X-Newsreader: TIN [version 1.2 PL2]

ObJustification: Posted so that mistakes/merits in my analysis can be
critiqued.

John Chan (johnchan@odyssee.net) said this stuff:
: I'm I student from Montreal and our biology class is experiencing fly 
: crossing. My team crossed Wild males and dumpy sepia females(PxP). Then 
: we crossed the offsprings and we obtained these results: 268 Wild Wild, 
: 64 Wild Sepia, 76 Dumpy Wild and 17 Dumpy Sepia. The Chi-Square Test 
: rejected our results (Our hypothesis for the ratio is 9:3:3:1). 
: Considering the fact that there was no experiment or identifying errors, 
: how can I explain the rejection of the results by the Chi-Square Test.

I didn't double check my own calculations, but for your data I find a
chi-square value of 10.25.  At three degrees of freedom, this places the
probability at <.05 but >.01.  Recall what this probability means.  There
is a greater than 1 percent chance that you will see data this far away
from what is expected even if the biology follows the rules you predict.
In other words, even if the genes above follow Mendelian genetics, some of
the time (>1/100) you will get aberrant data like that which you have
observed. It is possible, however unlikely, that this has happened in
your experiment.

Secondly, although you might be quite confident that you correctly
identified all of the offspring that you examined, isn't it possible that
the set you examined had pre-existing bias?  Since you are scoring
progeny phenotypes by looking at features in the adult, you must also
consider the health of embryos, larvae, and pupae with mutant alleles.  If
more sepia or dumpy flies die at early stages, you would expect bias
towards wild type (as you have observed).

I don't know if your teacher suggested this newsgroup for help with
analyzing your results, but if he/she didn't, do you think it is
appropriate to ask others for help in your write-up?  Did you try asking
your teacher for help, and if he would not give it to you, why do you
think not?  If discussion with outsiders was permitted, I hope this helps.

-Adam from Stanford
 

.