THANK YOU for this tutorial...very informative and easy to follow for the SPSS newbies, like me. I muddled through many videos before coming upon this one and I'm very grateful to have found it!
Thank you for making my MBA life easier :) I have been searching a lot of video clips and even fell asleep on my laptop most of the nights :P well ,this is a very informative video. Helped me a lot. Thanks !!!
Just managed my first Chi Square after lots of playing around and Table Colapsing. Just some advice on SPSS. If I have two variables from my survey: 1. Do you believe that terrorist groups are posing as irregular migrants; if so which terrorist groups do you think are posing as irregular migrants to enter the UK? 2. What terrorist group do you feel poses the most risk to you here in the UK? They have the same value's, 1 = ISIS 2 = IRA ect. ect. What would be the best test to see a relationship between the two?
Well done for running your 1st chi-square of independence test. You have 2 variables and both are nominal (categorical), so you could run the same test if you are testing whether there is a relationship between the 2. Slightly beyond the newbie level, if you find evidence to reject the null of no relationship, you should think whether there are other variables that could be related to the 2. See my vid on CMH test for more info.
Hi. I'm having difficulty with "expected counts". What to do if there are more than 20% of the expected counts are less than 5? In my case: 15 cells (53.6%) have expected count less than 5. The minimum expected count is .07. Thank you.
+Cavid Seyidov Cavid - 1st don't worry. It's not uncommon, and there are ways round it. If you have 3 or more groups, one standard thing to do is to combine groups. Say, you have a variable broken into groups "less then 2" "2-6", "6 or more". If you have problems with low expected counts in cell "less than 2", so may try combining with the next group, thus reducing the original 3 groups to 2, which will be "up to 6" and "6 or more". I have a video on how to make a new variable here: ruclips.net/video/9PZUyU2vnb8/видео.html
Correct me if I'm wrong, but wouldn't it be a leap to include "deaths due to lung cancer" in your conclusion? Speaking from a critical stand point with regards to logical interpretation of SPSS, I didn't see deaths from lung cancer a variable being measured? Unless I've missed something in which case I apologise hope I'm not being pedantic here but wouldn't a logical conclusion simply be " there is very strong evidence between smoking and lung cancer"?... Nice tutorial btw SPSS can be a mare!
+Nutrition Nerd Yes, just like with pairwise correlation, if you find evidence of association between 2 factors, this could disappear when you control for other variables, meaning that the 2 things you are looking at are not directly related.
+Anne Go Hi Anne. If the proportion of missing data is small (say below 5%, or another acceptable figure) then you can run the analysis but choose "delete listwise/casewise". What this does is to ignore the entry from any person who has not completed the whole questionnaire. If the proportion of missing values is high, then you are no longer at the level of the newbie, and then it adds another dimension of complexity to the crosstabs. I hope you only have a small proportion missing!
+Phil Chan (statisticsmentorcom) whew! the portion is 4.2% I almost had a mini heart attack. how do I delete list wise/casewise? is that back in data view?
+Anne Go Don't quote me on the 5%. If you are doing an undergrad dissertation, just mention it's a small proportion that's missing, or you may not even mention it. As for deleting listwise, check the options box when running crosstabs. Havent got access to SPSS right now, so can't check. Indeed I haven't used SPSS for ages. Have a nice holiday.
Good question. You see, the p-value depends on the value of the test stat. Ie it's linked to it. So it's not that you don't need the test stat cuz you do, but for deciding whether to reject the null you get that from the p-value (corresponding to the test value). This is the same for any test where you have a p-value to go with it
if alpha level is 0.05 ,also even 0.01, in both cases the probability of the findings by chance is less than 5 percent or 1 percent respectively, and we shall reject the null hypothesis and will state that there is an association ( or difference).
Hi James. Yes, you are right! It's better practice to set the variables to nominal/ordinal. I know, however, that SPSS does not look at the the choice of measurement, so I was just lazy that day.
Hi. You are referring to Measures column in Variable View where the variables are labelled scale. The variables are nominal, however. BUT the test does not look at what you entered in this column, so it doesnt matter what its set to. This Measures column matters if you use Chart Builder.
Very nice, however you got loads of saliva in your mouth, maybe take a sip of water sometime. I can hear it and I don't like it. Sorry :) Otherwise great!
![Eminem - Somebody Save Me (feat. Jelly Roll) [Official Music Video]](http://i.ytimg.com/vi/Vwa0HenQMi4/mqdefault.jpg)
THANK YOU for this tutorial...very informative and easy to follow for the SPSS newbies, like me. I muddled through many videos before coming upon this one and I'm very grateful to have found it!
Thanks, dear Phil you're great in explaining the complex statistical test in the simplest manner
I was lost with the interpretation part and this video helped me clear the air.
Thanks
This quite interesting. i consider it as the best video lecture i have ever watched. your speed is superp and accurate
Thank you for making my MBA life easier :)
I have been searching a lot of video clips and even fell asleep on my laptop most of the nights :P
well ,this is a very informative video. Helped me a lot.
Thanks !!!
Excellent, useful, user-friendly and perfect clarity. Thanks
This is the best tutorial video i have watched so far! Very informative and the pace is just perfect for students. Great job and good luck.
Ah! Lovely. At last someone who is not saying that I go too fast!
Thanks so much for a very clear explanation. I learnt something today and will not forget for a lifetime :)
This is brilliant. Very nicely explained! Thank you for uploading it!
That is a really really really fantastic and clear presentation.
wow nice presentation
you are great, explain well about chi-square test
Thank you! You are awesome! I finally finally finally understand this now (after going crazy just not getting it)
You are literally my hero!!!!!
Very informative. Thanks a lot.
thank you so much!!! before watch this video it is not clear for me now I understand very well.
thank you so much for excellent interpretation.
Thank you so much for the video...it really helpful
Appreciate everyone's nice comments. I wish you all well.
Thank you.
Thank you!
Thanks for the video. Why the gender was not used in the Chi-test?
This is good however you could have saved much time by speeding up a little bit. thanks
wonderful
Thanks a lot.
this is a good vdeo u know...u must see it..!!!
Thanks for such a nice video.
The explanation is truly helpful. How if in larger table (2 x 3 table), 1 cells (16.7%) have expected count
THIS WAS SO HELPFUL, THANK YOU
+ken hz Happy xmas!
Thank you sir
Thank you this was very clear!
Have you check the odd ratio from this data, can you make a video on it too..
thank you
This really helped me, thank you
Thanks Phill
Just managed my first Chi Square after lots of playing around and Table Colapsing.
Just some advice on SPSS. If I have two variables from my survey:
1. Do you believe that terrorist groups are posing as irregular migrants; if so which terrorist groups do you think are posing as irregular migrants to enter the UK?
2. What terrorist group do you feel poses the most risk to you here in the UK?
They have the same value's, 1 = ISIS 2 = IRA ect. ect. What would be the best test to see a relationship between the two?
Well done for running your 1st chi-square of independence test. You have 2 variables and both are nominal (categorical), so you could run the same test if you are testing whether there is a relationship between the 2. Slightly beyond the newbie level, if you find evidence to reject the null of no relationship, you should think whether there are other variables that could be related to the 2. See my vid on CMH test for more info.
Hello mate
Thanks for the video
But how do I deal with data in which I have examined around 4000 animals for a specific disease
I'm selecting two variables in for the Chi square test but it is saying its a constant and cant be measured ? #
displays warning sign
powerful
Is there a way to provide 95% CI next to the percentages in the crosstab?
Hi Phil. I would like to ask, do I need to enter the no response answer option, if no one have ticked them?
nice . Thanks
if we where to look for Cohen's W where is SPSS can we find the phi
good,thanks
Can't we use 'correlations' to check the degree of association ??
Thanks for the help! :D
Hi. I'm having difficulty with "expected counts". What to do if there are more than 20% of the expected counts are less than 5? In my case: 15 cells (53.6%) have expected count less than 5. The minimum expected count is .07. Thank you.
+Cavid Seyidov Cavid - 1st don't worry. It's not uncommon, and there are ways round it. If you have 3 or more groups, one standard thing to do is to combine groups. Say, you have a variable broken into groups "less then 2" "2-6", "6 or more". If you have problems with low expected counts in cell "less than 2", so may try combining with the next group, thus reducing the original 3 groups to 2, which will be "up to 6" and "6 or more". I have a video on how to make a new variable here: ruclips.net/video/9PZUyU2vnb8/видео.html
Correct me if I'm wrong, but wouldn't it be a leap to include "deaths due to lung cancer" in your conclusion? Speaking from a critical stand point with regards to logical interpretation of SPSS, I didn't see deaths from lung cancer a variable being measured? Unless I've missed something in which case I apologise hope I'm not being pedantic here but wouldn't a logical conclusion simply be " there is very strong evidence between smoking and lung cancer"?... Nice tutorial btw SPSS can be a mare!
+Nutrition Nerd Yes, just like with pairwise correlation, if you find evidence of association between 2 factors, this could disappear when you control for other variables, meaning that the 2 things you are looking at are not directly related.
+Phil chan you are the best... I'm doing a stats project and your explanations are on point man! But what if i have missing data??? @ 5:14
+Anne Go Hi Anne. If the proportion of missing data is small (say below 5%, or another acceptable figure) then you can run the analysis but choose "delete listwise/casewise". What this does is to ignore the entry from any person who has not completed the whole questionnaire. If the proportion of missing values is high, then you are no longer at the level of the newbie, and then it adds another dimension of complexity to the crosstabs. I hope you only have a small proportion missing!
+Phil Chan (statisticsmentorcom) whew! the portion is 4.2% I almost had a mini heart attack. how do I delete list wise/casewise? is that back in data view?
+Anne Go Don't quote me on the 5%. If you are doing an undergrad dissertation, just mention it's a small proportion that's missing, or you may not even mention it. As for deleting listwise, check the options box when running crosstabs. Havent got access to SPSS right now, so can't check. Indeed I haven't used SPSS for ages. Have a nice holiday.
+Phil Chan (statisticsmentorcom) Thank you. not my dissertation, its just a project for my stat class. Same to you! enjoy your Christmas.
Thanks, and good luck.
Thank you very much :)
what is the point of khi square value if all we need is the P value ?? plz explain
Good question. You see, the p-value depends on the value of the test stat. Ie it's linked to it. So it's not that you don't need the test stat cuz you do, but for deciding whether to reject the null you get that from the p-value (corresponding to the test value). This is the same for any test where you have a p-value to go with it
it makes sense now thanks a lot for the explanation
what about spearman correlation?
any difference?
I M DOING RESEARCH KNOW DATA ENTRY PROCESS TO FIND OUT THE IMPACT OF TRAINING ON EMPLOYEE RETENTION
was it very cold there?
Very good video, just a bit slow for me...
Does anyone know how to interpret a result if you were to perform a (Chi) χ2 test and obtain a p-value of 0.005.
depends upon alpha level
if alpha level is 0.05 ,also even 0.01, in both cases the probability of the findings by chance is less than 5 percent or 1 percent respectively, and we shall reject the null hypothesis and will state that there is an association ( or difference).
You told SPSS that your categorical variables were scale variables
Hi James. Yes, you are right! It's better practice to set the variables to nominal/ordinal. I know, however, that SPSS does not look at the the choice of measurement, so I was just lazy that day.
What I dont understand is why you used scale variables instead of nominal for cancer and gender?
Hi. You are referring to Measures column in Variable View where the variables are labelled scale. The variables are nominal, however. BUT the test does not look at what you entered in this column, so it doesnt matter what its set to. This Measures column matters if you use Chart Builder.
Good effort but would be better if it wasn't so slow and bumbling
Smoking while doing statistics... maybe not the best idea... fml
Very nice, however you got loads of saliva in your mouth, maybe take a sip of water sometime. I can hear it and I don't like it. Sorry :) Otherwise great!
Thanks for your feedback. Perhaps I have the mic too close up.
Be nice Slo, your are no body. you don't like it, please get your sorry ass out of here. How clear do you want this explanation to be?
I feel like the broadcaster took my constructive critique better than you did. Grow up.