Hi, Katie.
Your last message is copied below, with my comments, additions, or
substitutions [in square brackets]. On the whole, your analysis is
correct. -- DFB.
Date: 27 Apr 2004 05:22:27 -0700
From: Katie <[EMAIL PROTECTED]>
To: [EMAIL PROTECTED]
Subject: Re: Yes/no ranked data nightmare
Yes, the difference is by gender. I've given it a go, but am a little
unsure about the conclusion.
| Male | Female | [All]
---------------------------+------
Yes | o: 94 o: 17 | [111 0.3383 + 1.3326 + ]
| e: 88.5 e: 22.5 |
| |
No | o: 36 o: 16 | [ 52 0.7221 + 2.8446 = 5.2376 ]
| e: 41.5 e: 10.5 |
----+----------------------+------
[All | 130 33 | 163]
[where e = (row sum)*(column sum)/(total).]
Using ((o-e)^2)/e I got a chi s of 5.29.
[Slight error due to rounding to 1 decimal point.]
I think the df for the table is 1 (is it (rows-1)*(columns -1)).
[Right.]
Therefore using a table, p 0.05 is 3.84 and p 0.01 is 6.63
so it is significant to p 0.05 but not quite all the way to 0.01
[That is, critical values of chi-square are 3.84 (p=0.05) and 6.63
(p=0.01), so 0.01 < p < 0.05.]
What does that mean?
There is a significant difference between the yes and no proportions
that males and females gave?
[Yes, if your level of significance is 0.05; no, if it's 0.01.]
Such that males were statistically more inclined to say yes. [Right.]
[because number of males who said "yes" is greater than expected,
(and number of females who said "no" is greater than expected).]
(ratio of male yes:no is 5.5:1 whereas the females is 2.25:1)
[Not true: those ratios are 94/36 = 2.61 and 17/16 = 1.06.
The ratios you report are Yes M:F and No M:F. Unless you've
mislabelled the rows and columns of the table.]
Am I on the right tracks? [Yes.]
Hope this was helpful. -- DFB.
------------------------------------------------------------
Donald F. Burrill [EMAIL PROTECTED]
56 Sebbins Pond Drive, Bedford, NH 03110 (603) 626-0816
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
