Chi Square Test in SPSS (Part 1)
HTML-код
- Опубликовано: 23 авг 2024
- How to run a chi square test of independence in SPSS and interpret the output is illustrated in this video.
Chi Square Test
Cross Tabs
Contingency Tables
chi-square test
Pearson Chi-square
For more SPSS videos: tinyurl.com/nyw...
SPSS videos statisticsinstructor/Quantitative Specialists:
RUclips channel: / statisticsinstructor
Udemy inferential course: www.udemy.com/...
Udemy descriptives course: www.udemy.com/...
Video Transcript: In this video we'll examine how to run the chi-square test of independence in SPSS. The chi-square test of Independence measures whether there is a relationship between two categorical variables. And in our example we're going to look at whether there is a relationship between gender and aggressive driving behavior. And what happened in this hypothetical study is we have 200 people, where we have 100 males and 100 females, and then we asked each of these persons through a series of questions on a survey, whether or not they engaged in aggressive driving behavior. So, for example, do they run stoplights, did they tailgate anyone, this sort of thing. And based on the survey responses, they were coded either yes, they engaged in aggressive driving behavior, or no, they did not. So notice here we have our two categorical variables, gender, that's categorical, as it is either male or female, and then aggressive driving behavior is also categorical in this example, as it's either characterized as a yes or no. And once again the chi-square test of Independence measures whether there is a relationship between two categorical variables. So the chi-square test of Independence would be appropriate here, since we have our two categorical variables, gender and aggressive driving behavior. So let's go ahead and take a look at our data here in the Data View window. Notice there's three variables, we have gender, where we have 1s and 2s, and then we have aggressive driving, where we also have 1s and 2s, and then we have our frequency variable, and you saw this in the other chi-square we talked about, the chi-square goodness of fit, where the frequency variable indicates the number of people in a given category. So this 25 means there's 25 people who have a 1 on gender & a 1 on aggressive driving, there are 75 people who have a 1 on gender and a 2 on aggressive driving, and so forth. And we went ahead and we already coded these values into SPSS and I'll show you their values here by clicking on the Variable View tab. And if we go to gender, we can see that 1 is male and 2 is female, and then for aggressive driving, we have 1 is coded as yes, they engaged in aggressive driving, and 2 is no, they did not. So if we go back to our Data View window, and I can click on the Value Labels button up here and you'll see these codes. So we had 25 people who were male and did engage in aggressive driving behavior, whereas we had 75 people who are male who did not engage in aggressive driving behavior. For females, 10 engaged in aggressive driving behavior, and 90 females did not engage in aggressive driving behavior. It's a good idea to take a few moments to make sure that it's clear how to set up the data file for the chi-square test of independence. Notice that we have our 1s and 2s for gender and then we have our 1s and 2s alternated for aggressive driving behavior, and then we have our frequencies. Notice that there's no repeat here of a combination of 1s and 2s. Here we have 1,1 and then we have 1,2 and then 2,1 and then 2,2. And then we have the frequencies for each of these different combinations of 1s and 2s. So the 1s and 2s never repeat the same sequence again. So 1,1 1,2 2,1 2,2, and looking at their actual labels that gives us male aggressive, male not aggressive, female aggressive, female not aggressive. So we want to have all different combinations expressed here in our Data View window. And then the only other thing you need to make sure is that you get the correct frequencies for the category. So we had 25 males who were aggressive, and 75 males who are not aggressive, for example. So in setting up the chi-square test of independence, we have all of our different categories expressed here, and then you just want to make sure you get the correct frequencies in each category. Next we'll go ahead and run our chi-square test of independence and let's use an alpha of .05 here, and the chi square test is naturally a one-tailed test. So to run the chi-square test, because we have our frequency variable here, first we need to weight the cases
Hey, can you do the chi squared test without the frequency?
thank u for repeating one thing again and again
Amazing video! It solves most of the problems! Keep sharing such videos!!
Hello! Thank you so much for posting such a clear explanation of the Chi Square! I have a question and hoping you can help me. I am currently working on my research study in which I am trying to see if a relationship exists between learner age and preferred learning style. In my study, I have 2 groups of learners, Traditional ages 18-24 and Non-traditional ages 25 and up. The learning styles are Auditory, Visual, and Tactile. Can I still use a Chi Square?
+Kristen Wagner
If people fall in only one style (rather than have scores on each style), then chi-square should do the trick. If they have scores on each variable, then a MANOVA or, for a simpler analysis, 3 independent t tests would work.
If I have the same variable in the same population (qualitative, like smoking habit: 1:yes 2:no 3: ex) and I want to see how it evolves on time (baseline-12 months-24 months) can I use Chi-Square?
I am a research scholar, my guide has suggested an hypothesis - there is no significant difference in the opinion of manager of spinning Mills and workers of spinning Mills regarding quality of working conditions. I have collected data on 5- point likert scale. He suggested me to use chi- square test here. Please please help me. Guide me whether chi- square test can be applied in the way u have explained if not then what am I supposed to do instead?
Awaiting your reply
Sir, kindly upload video about goodness of fit test.🙏thnx
please is it right to use percentages instead of frequencies? Thank you very much
Sir, here first category was gender and second category was a question based on yes or no type of reply. Can yes no type of category can be replaced by a question with likert scale of five points. Please reply. Awaiting your reply
Did u get ans?
i have the same question
So brilliant, thank you!
Thanks for saving my life :)
Register for 2 hour SPSS Online Workshop scheduled for Sun, 21 June : forms.gle/8o86WPs4fwf5fb379
Thank you so much
very heplful video, thanks again
My pleasure!
wonderful, thank u
thanx
nice class
gender and aggressive driving are not scale
Amrit,
You are correct in that gender and agressive driving behavior are nominal variables. However, I typically leave the deafult values in place for the measure column, as it makes no difference in the analysis. For some of the newer graphing features in SPSS, coding the measure column becomes important, but for the chi-square test it doesn't matter.
'Measure' was a feature SPSS added a handful of years ago, but typically it makes no difference whether these values are modified or not. What is important is that the user understand what the variables are and that the use the appropriate procedures.
For the chi square test, typically only nominal variables (and sometimes ordinal) should be used anyway. If interval or ratio variables are available, there are typically more powerful procedures that can (and should) be used in most cases.
QS
but the information must be always clear. it was only that. there must be no any dilemma
If one takes a deeper look into scales of measurement, actually clarity is not something that is necessarily easily obtained. For example, very few variables strictly fall into the category of interval, yet they are commonly treated that way in many analyses...Are they subinterval, supraordinal, or just ordinal?
There is even increasing debate about whether the scales of measurement, orginally proposed by Stevens in the 40s, actually even are the relevant question to an analysis at hand. Some people argue that the question is whether the assumptions of a test are met, not if the supposed scale of measurement is met. (See p.7 of Warner, Applied Statistics, for an example of this.) And then if the assumption isn't met, if the test is robust, then often it is still generally acceptable to use it.