Any statisticians out there?

Re: Any statisticians out there?

David Mills — Thu, 08 Mar 2012 01:24:02 GMT

[quote user="Arlo Guthrie"]

David, how very kind of you to do all that for me. Thank you so much. Can't tell you how much I appreciate it.

I am now trying to work backwards through your figures to make sure I understand them correctly. Whether or not I succeed remains to be seen!

Thanks again.

[/quote]

No problem only took couple of minutes as I'm in process of doing some stats for some papers and its all set up in excel.

The stats test used is the chi-squared test which is ideal for your data which essentially fits into a 2 x 2 table. The significance level is set at 0.05, which is the standard level.

Re: Any statisticians out there?

Hannah Wynne Richards — Wed, 07 Mar 2012 12:34:52 GMT

I knew I should have paid more attention when we had a stats course as 1st years

Wynne

Re: Any statisticians out there?

Arlo Guthrie — Wed, 07 Mar 2012 12:03:15 GMT

[quote user="David Mills"] No. It shows that 1.8% more people with children own an animal than those without, not likeliness. Sample size is adjustable by using odds thence likeliness to own an animal according to circumstances.

See below

Odds are in the 'x:1' i.e. own:non-own

Odds ratio give an indication of how strong the association is.

Likeliness figures are in the form you're likely to want to use i.e. People with children are 1.02 times more likely to have pets than not.[/quote]

David, how very kind of you to do all that for me. Thank you so much. Can't tell you how much I appreciate it.

I am now trying to work backwards through your figures to make sure I understand them correctly. Whether or not I succeed remains to be seen!

Thanks again.

Re: Any statisticians out there?

Rob Loxley — Tue, 06 Mar 2012 18:13:48 GMT

Haven't run the stats test, but a quick eyeball would make me inclined to wonder if the overall difference between the 2 groups is so slight to not be significant (i.e. could just be chance), so you'd probably need to stick with differences between the 2 groups for particular species (e.g. vets with childern are more likely to own small furries).

Re: Any statisticians out there?

David Mills — Tue, 06 Mar 2012 17:01:43 GMT

[quote user="Arlo Guthrie"]

If not, is it correct to say you are 1.8% more likely to own any animal if you have children (96.6-94.8).

[/quote]

No. It shows that 1.8% more people with children own an animal than those without, not likeliness. Sample size is adjustable by using odds thence likeliness to own an animal according to circumstances.

See below

Odds are in the 'x:1' i.e. own:non-own

Odds ratio give an indication of how strong the association is.

Likeliness figures are in the form you're likely to want to use i.e. People with children are 1.02 times more likely to have pets than not.

		Children	No children
Any animal	Own	173.00	256.00
	Non own	6.00	14.00
	% Own	96.65	94.81
	Odds	28.83	18.29

	Odds ratio	1.58
	Likeliness	1.02

	Chi	0.85
	p	0.40

Cats	Own	121.00	164.00
	Non own	58.00	106.00
	% Own	67.60	60.74
	Odds	2.09	1.55

	Odds ratio	1.35
	Likeliness	1.11

	Chi	2.18
	p	0.14

Dogs	Own	136.00	164.00
	Non own	43.00	106.00
	% Own	75.98	60.74
	Odds	3.16	1.55

	Odds ratio	2.04
	Likeliness	1.25

	Chi	11.25
	p	0.001

Fish	Own	44.00	35.00
	Non own	135.00	235.00
	% Own	24.58	12.96
	Odds	0.33	0.15

	Odds ratio	2.19
	Likeliness	1.90

	Chi	10.00
	p	0.002

Small furries	Own	49.00	23.00
	Non own	130.00	247.00
	% Own	27.37	8.52
	Odds	0.38	0.09

	Odds ratio	4.05
	Likeliness	3.21

	Chi	28.36
	p	<0.0001

Birds	Own	14.00	14.00
	Non own	165.00	256.00
	% Own	7.82	5.19
	Odds	0.08	0.05

	Odds ratio	1.55
	Likeliness	1.51

	Chi	1.28
	p	0.30

Reptiles	Own	18.00	15.00
	Non own	161.00	255.00
	% Own	10.06	5.56
	Odds	0.11	0.06

	Odds ratio	1.90
	Likeliness	1.81

	Chi	3.19
	p	0.07

Horses/ponies	Own	25.00	35.00
	Non own	154.00	235.00
	% Own	13.97	12.96
	Odds	0.16	0.15

	Odds ratio	1.09
	Likeliness	1.11

	Chi	0.09
	p	>0.5

Farm	Own	42.00	28.00
	Non own	137.00	242.00
	% Own	23.46	10.37
	Odds	0.31	0.12

	Odds ratio	2.65
	Likeliness	2.26

	Chi	13.99
	p	0.002

SO statistically significant differences (P<0.05) exist between children/non children families in dogs, fish, small furries, and farm.

Edited as some figures wrong previously, now with significance analysis.

Re: Any statisticians out there?

Lorna McHardy — Tue, 06 Mar 2012 15:52:00 GMT

I'm definitely not a statistics person. The only thing that strikes me, looking at that, is that those with children appear to own significantly more small furries than those without; whereas the other differences look insignificant to me (this is my impression, not a statistically worked out significance). And if I'm right, that that, I think, might account for the fact that people with children have more pets overall.

And if I'm right, I think it's unsurprising, but a Bad Thing. I do not believe small furries are ideal pets for children; they're prey animals and do not like being grabbed. Far better a puppy or reasonably tolerant cat to teach them how to handle animals properly... or more to the point, how not to handle animals. It's a common misconception that small furries aren't any work; but that's only true if they're neglected to some degree. If properly looked after, they take more time than a cat, and nearly as much as a dog.

Re: Any statisticians out there?

Hannah Wynne Richards — Tue, 06 Mar 2012 11:57:40 GMT

Possibly horses and children both expensive

Wynne