On Mon, 20 Jan 2003, VOLTOLINI wrote: > I have a doubt about this exercise. When doing a germination test > using two treatments a student had this results: Treament 1 = 30 > seeds germinated (52%) ... and ... Treatment 2 = 50 seeds germinated > (25%). The problem is that the sample size was not the same; in one > treatment he was using more than 60 seeds and in the other one only > 50 seeds !
If 30 seeds is 52%, Treatment 1 must have started with 58 seeds; and if 50 is 25%, Treatment 2 must have started with 200 seeds. I don't understand your "more than 60" (although 200 IS more than 60) and "only 50". > What kind of test he can use ? Depends on what question he wants to address. Since you haven't said, I'll assume the question of interest is whether the two germination rates differ: that is, one would test the null hypothesis that the germination rate for Treatment 1 (observed to be 0.52 in this sample) is equal to the germination rate for Treatment 2 (observed to be 0.25 in this sample). Standard tests for equality of proportions (which are means of an observed variable whose values are "1" (the seed germinated) or "0" (it did not), for equality of percentages (which are merely proportions multiplied by 100%), and for a 2x2 table of frequencies (the chi-square test to which I presume you allude below) are all equivalent. Take your pick. (If the question of interest is something else (but I cannot imagine what that might be), the kind(s) of test available may differ from these.) > Chi-square ? Could do. > A proportion test ? Could do. > Is there a test to compare percentages ? Yes, same as for proportions: a t-test, usually, although one could structure it equivalently as an analysis of variance with two groups. Hope this is 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/ . =================================================================
