PDA

View Full Version : finding a relation in spss

savard
02-22-2009, 09:39 AM
hi
i'm stistics i've just graduated 4 moths ago
i have just got a job in a company and they gave me my first job so they are asking me to do a statistical analysis (describtive statistics and inferential statistics) i appreciate it if you can help me because my future in this jop is depending on doing it.

i have a quiustion about the "SPSS" in statisic and i hope i can find the answer with you:

1) ther is a test in statistics but im not sure of the name called "chi-square test" (to know if there is a relation between 2 catogoricals variables)
they asked me to eveluate the electronic companies by there machines that are used in the company im working in like hp, sony, toshipa, phones company, etc....
well, the problem is that i couldn't figure if it is right "statistecally" to use this test to know if there is a relation between 2 catigorical vriables ,for example; i have 3 variables:

kind of the company (hp, dell, mental) =variable 1

causes of the damage of the machine (humanly, technical failure, electrical failure, other causes) = variable 2

is the machine private (used by one person) or general (used by alot of people) = variable 3

ok, now i want to find if there is a relation between variable 2 and 3 by using the test i mentioned(chi-sequare) BUT only for HP company alone(i mean that i will use a select case) is it right or not using the test in this situation.

I hope u get what i want. it really depressed me :(

savard
02-22-2009, 04:02 PM
Help is all i'm asking i know it may be selly quistion but common HELP PLEASe
:(

Masteras
02-23-2009, 05:57 AM
i assume you have entered the data. two variables, two columns, on for privacy or not and one for damages. go to analyze>descriptive statistics>crosstabs. Enter the two variables one in the Row(s) and one in the Column(s). Press Statistics and then select chi square. A message will appear under the table of the chisquare test. if the percentage mentioned is more that 20-25&#37; repeat the same but then press Exact and choose Monte Carlo.

savard
02-23-2009, 12:11 PM
i assume you have entered the data. two variables, two columns, on for privacy or not and one for damages. go to analyze>descriptive statistics>crosstabs. Enter the two variables one in the Row(s) and one in the Column(s). Press Statistics and then select chi square. A message will appear under the table of the chisquare test. if the percentage mentioned is more that 20-25% repeat the same but then press Exact and choose Monte Carlo.

first THANX alot for the reply

is the tests result is "STATISCALLY" right when i use the same steps u mentioned with making a select case first before the test.

that is what i want to know

SUNSHINE09
02-23-2009, 12:17 PM
This is sunshine09. Could someone help me... I have recently made a post.. I need help still.

Masteras
02-23-2009, 03:08 PM
Savard, yes it is

savard
02-24-2009, 04:43 AM
Savard, yes it is

ok Masteras

THANX alot :):):D

Masteras
02-24-2009, 04:44 AM
e savard re malaka vale kana thanks. press thanks also.

savard
02-28-2009, 03:49 PM
e savard re malaka vale kana thanks. press thanks also.
Hi
i'm so emparresed i need your help again

I wonder if you could help me in writing the code of the file for the data in SPSS. Some of My variables (4,5,6) that causes the problem are

Variable (1) = the name of the company {hp , Toshiba , mental (phone company) , dell , etc…..}

Variable (2) = the guaranty of the company { 1year , 2year , 3 year , no guaranty }

Variable (3) = type of the machine { telephone , screen , hardisk , keyboard , etc…..}

Variables (4) = kind of the damage of the machine { humanly, technical failure, electrical failure, other kind }

Variable (5) = cause of the damage of the machine { wrong in using the machine , system problem , etc…..}

Variable (6) = is the damages can be repaired { yes , no }

Now , for example you have 1000 damaged machines and every machine damaged for at least 4 times in different kind of damaged and for different causes , how will you write a code for these kind of data without having wrong or repetition answer in the result ????????

Masteras
02-28-2009, 05:57 PM
weel, if each machine was damaged only once you could do a multinomial logistic regression. The only simple I can think of, is many chi-square tests. Yes do this, many X2 tests. If you have more problems, I am here.

savard
03-04-2009, 02:53 AM
weel, if each machine was damaged only once you could do a multinomial logistic regression. The only simple I can think of, is many chi-square tests. Yes do this, many X2 tests. If you have more problems, I am here.

Thanx Alot Masteras for ur help :)