Board 8 > board 8, i am in a bit of a spot, and I need a small bit of statistics help

(+) Yeah! (+)

i am working on my year-end stats project, which was an observational experiment

my hypothesis is that men are more likely to ask for help if they are by themselves than if they are with their wives or girlfriends


the method, in short, was that while i was at work, i noted the gender and grouping of people who asked for help from employees. the groups are "alone", "couples", and "2+ of the same gender / mixed group"

i have 273 pieces of data

141 m
132 f

126 alone
91 couples
56 groups

the problem is this: what test statistic(s) should I run (I'm using SPSS) to determine whether or not my hypothesis was accurate (and so on)?


i have perused my notes, but we never really did any problems quite similar enough to this to be of use.

--
http://img.imgcake.com/yesjpgad.jpg

Purple

--
XIII_rocks, the cream of XIII fanboyism.

Lot of variables. Do multiple regression.

--
Predatorian234 - Hey, guess what, brainless uneducated parents are a market group.

also

is the third grouping irrelevant, and can i exclude (and delete) them

and would that make things easier

augh

--
Cokes as The Green Lantern, by fr0q
http://img.imgcake.com/greencokespngsa.png

Hey I'm no statistician but to accurately determine if men are more likely to ask for help when they're not with their wives/girlfriends, shouldn't you also have recorded how many men DIDN'T ask for help? Who's to say men aren't more likely to come into the store alone?

--
[NO BARKLEY NO PEACE]

First, you're only interested in the men, right?

I'd say find the proportion that ask for help in each of those populations (alone, group, SO) and then you can do 2-way (P1-P2) 3 times.

--
senorhousemouse

P = proportion that asked for help, Q = proportion that didn't ask for help
Pop 1 = population without SO, pop 2 = population with SO
Mean = P1-P2
SE = something like sqrt(P1Q1/n1 + P2Q2/n2)

If the margin of error doesn't include 0, you reject the null hypothesis.

I have doubts about the design of the experiment, but that's the math for you

--
senorhousemouse

From: Dauntless Hunter | #005
Hey I'm no statistician but to accurately determine if men are more likely to ask for help when they're not with their wives/girlfriends, shouldn't you also have recorded how many men DIDN'T ask for help? Who's to say men aren't more likely to come into the store alone?

i sort of gave you guys the shorthand.

i have more stuff that clarifies these things - it's costco, so everyone who comes in is a member, i got membership demographics from one of the managers, and it's like 52% M, 48% F

--
YES! GO BOSTON! - Naye

The demographic info is kind of a red herring, I think.

--
senorhousemouse

there is more to it than that that i lost by accidentally refreshing

also i have had Mr. Mathface Wigs helping me out, so I am headed in the right direction now, I think

--
http://img.imgcake.com/pyramidjpgyg.jpg
CRYING. Acceptable at funerals and the Grand Canyon

What you're testing is whether a man is more likely to be with his SO if he asked for help. To know if a man is more likely to ask for help if he's with his SO, you'd need to know how many couples enter and how many single men enter.

You're still testing something, but it's different from what you think it is.

--
senorhousemouse