Here is a link to the correlation playlist...check out the playlists on my channel for many more: th-cam.com/play/PLVI_iGT5ZuRlew7Kp_mxKl1GoxK-v8evK.html
I hope you can answer my question. I am conducting a study about "correlation of marketing strategies on sales of water refilling stations". My study uses the 4 Ps of marketing (Product, Price, Place, Promotion) is it possible to correlate it to the sales of water refilling stations? Is it ok to have 4 independent variables and 1 dependent variables?
Yes, you can correlate five scale variables in a correlation matrix. Just drop all into the Variables box, put the DV in first for ease of interpretation. You could also use a regression equation if you want to study how the four predictors contribute to the variability in sales. Good luck with your study
Thanks for the thumbs up. Remember that this is video #8 for the week that I cover correlation for my stats class. the full explanation is in the previous videos and this is the how to for the SPSS part. Let me send you the playlist for correlation. This will be everything you need: th-cam.com/play/PLVI_iGT5ZuRlew7Kp_mxKl1GoxK-v8evK.html
Thank you very much for your videos! I was wondering what test you would suggest for a correlational study with 1 independent and 2 dependent variables (each variable will be measured on a likert-type scale)? My research question is the following: What is the relationship between Catholic religiosity and levels of anthropocentrism in teenagers? The first questionnaire will measure Catholic religiosity, the second will measure levels of anthropocentrism, and the third is an additional survey to measure the social desirability of respondents since both constructs will be susceptible to response bias. Please let me know if you can help or if there is any way I can contact you! Thanks again.
It is a lot to answer in the comments section, but you can use Likert items in correlation. If the scale is a single item, use Spearman's rho, but if you combine multiple Likert items into a scale by using their mean, you are safe to use Pearsons'. You can correlate each item with the other, partial correlation might be in order, or you can shift your research question to use regression to predict anthropocentrism from religiosity and social desirability. Or more likely, anthropocentrism predicts religiosity, moderated by social desirability (i.e. moderated regression). Look into PROCESS macro for ModReg. Lots of options for you. Good luck
If you were doing regression model building (instead of correlation) you would certainly consider the level of significance when choosing which variable to retain in your model.
@@ResearchByDesign actually I am doing image classification and for that I have 300 hog features for each image. And I want to do dimension reduction. That is why I am searching for the techniques which will give me minimum features and high accuracy..can you suggest me which statistics should I use for dimension reduction?
Hmmm...sounds like you may be asking about a biserial (or point biserial) correlation. That is a Pearson correlation in which one variable is scale (continuous) and the other is dichotomous. If so, you conduct the correlation exactly the same way in SPSS as any Pearson correlation, but the interpretation is a little different. Until I do a video, check out Laird Statistics for info on the interpretation. Good luck
If you have ranked data, you would use a Spearman's (non-parametric) correlation. In SPSS, that requires only checking one more box in the correlation settings, easy to do.
It is less about a one-tailed test and more about a one-directional correlation. In other words, if these variables are positively correlated, then that is good, but if they are uncorrelated or negatively correlated, that is no good.
Roughly speaking, you can regress 1 DV on the 5 IVs in a multiple regression. There is a lot to know about regression, so I would recommend you to a good advanced stats book, but that would be the most likely type of analysis.
The job satisfaction score comes from the JSS survey by Paul Spector. It was the total score from 36 items measuring 9 facets of job satisfaction. I did not cover that in this video, but I talk about it in a proviso video (these are part of a stats class). Thanks for asking.
Hello! really helpful video! Is it possible to give us the source of this example and data set? (is it from a book, a survey or somewhere else for instance)
Check the link in the description of this (or any of the other) video. There is a link that Google Drive. In that folder you will find the dataset. Right click to download. Good luck.
These videos are amazing many thanks! But I am slightly confused about their sequence. Which one is the previous video? Is it number 12? But it's about ANOVA.
For example I have likert scale questionnaire for the two variables, can I get the mean of the two variables then do the pearson!? **sorry, i am not an expert. I just need it for my research :(
This is a general answer to whether you COULD do it (I can't tell if you SHOULD do it without knowing your specific questionnaire). Yes, you could average multiple items (even if they are measured using a Likert scale) and then correlate them with another scale (which you also averaged). The "should" question depends on whether the items are intended to be averaged, whether they are conceptually related, etc. However, if the Likert scale has at least 5 answer options, it can be treated as scale data and averaged.
Probably the simplest method would be double-click on your SPSS correlation matrix, select all, copy and paste into Excel. Un-merge cells. You can then edit the table in Excel and keep only the parts that you want.
What you are asking about is a point-biserial correlation. You put the scale variable in first, then the gender/nominal variable (code as 1 & 0 for easier interpretation - must be dichotomous). Then run a Pearson correlation the same as this example. Mathematically, this is the same as an independent-sample t test.
You are the reason I was able to write chapter four of my quantitative dissertation. Thank you so much!
I appreciate all your videos. A lot of effort is made to clearly explain the statistics and how to use SPSS. Thanks a lot.
You are welcome. Thanks for taking the time to comment
Thank you. Your just amazing with all the explanations. You helped me a lot. I thank you from the bottom of my heart.
Thank you so much for the comment. Very glad that the videos were helpful. Please tell your classmates about them, too. :o)
All the information I need! Thanks 🎉🎉🎉🎉
Thank you so much Sire. This video helped me a lots
You're most welcome!
I love you Dr. Todd Daniel... you are my angel
Oh, that is wonderful. Thank you for that. Glad that the videos are helpful to you.
Excellent explanation
Thank you Dr....👍..this really explains it well..
Glad it was helpful!
Thank you so much, very helpful and simple!
Glad it was helpful!
It would be really convenient to have links for source datasets in the description of the video
Good work! Clear explanations!..thanks a lot
Glad it was helpful!
neatly explained
Thank you !
this is amazing
How did you get r^2 value?
Very simple: any correlation is r (short for "regression"); simply square the correlation to get r-squared (r^2).
It could be a good idea to share a link for the whole playlist sir! Or links to each video in the description. Thank you!
Here is a link to the correlation playlist...check out the playlists on my channel for many more: th-cam.com/play/PLVI_iGT5ZuRlew7Kp_mxKl1GoxK-v8evK.html
I hope you can answer my question. I am conducting a study about "correlation of marketing strategies on sales of water refilling stations". My study uses the 4 Ps of marketing (Product, Price, Place, Promotion) is it possible to correlate it to the sales of water refilling stations? Is it ok to have 4 independent variables and 1 dependent variables?
Yes, you can correlate five scale variables in a correlation matrix. Just drop all into the Variables box, put the DV in first for ease of interpretation. You could also use a regression equation if you want to study how the four predictors contribute to the variability in sales. Good luck with your study
Gave you a thumbs up on this one, but it would be nice to receive a full explanation before other aspects are addressed.
Thanks for the thumbs up. Remember that this is video #8 for the week that I cover correlation for my stats class. the full explanation is in the previous videos and this is the how to for the SPSS part. Let me send you the playlist for correlation. This will be everything you need: th-cam.com/play/PLVI_iGT5ZuRlew7Kp_mxKl1GoxK-v8evK.html
Can you please give the links of older videos
Thank you very much for your videos! I was wondering what test you would suggest for a correlational study with 1 independent and 2 dependent variables (each variable will be measured on a likert-type scale)? My research question is the following:
What is the relationship between Catholic religiosity and levels of anthropocentrism in teenagers?
The first questionnaire will measure Catholic religiosity, the second will measure levels of anthropocentrism, and the third is an additional survey to measure the social desirability of respondents since both constructs will be susceptible to response bias. Please let me know if you can help or if there is any way I can contact you! Thanks again.
It is a lot to answer in the comments section, but you can use Likert items in correlation. If the scale is a single item, use Spearman's rho, but if you combine multiple Likert items into a scale by using their mean, you are safe to use Pearsons'. You can correlate each item with the other, partial correlation might be in order, or you can shift your research question to use regression to predict anthropocentrism from religiosity and social desirability. Or more likely, anthropocentrism predicts religiosity, moderated by social desirability (i.e. moderated regression). Look into PROCESS macro for ModReg. Lots of options for you. Good luck
where can I see the biostatistics videos that you mention in these SPSS videos?
Thank you for your great experience about SPSS software it really helpful for me..can you please help me out about DMRT.
Not sure what DMRT stands for. Is it a test or a research tool?
those variables which have two asterisks should be considered as final variables in the dataset??
If you were doing regression model building (instead of correlation) you would certainly consider the level of significance when choosing which variable to retain in your model.
@@ResearchByDesign actually I am doing image classification and for that I have 300 hog features for each image. And I want to do dimension reduction. That is why I am searching for the techniques which will give me minimum features and high accuracy..can you suggest me which statistics should I use for dimension reduction?
Dr. Todd please make a video correlation analysis on yes/no questionaire.
Hmmm...sounds like you may be asking about a biserial (or point biserial) correlation. That is a Pearson correlation in which one variable is scale (continuous) and the other is dichotomous. If so, you conduct the correlation exactly the same way in SPSS as any Pearson correlation, but the interpretation is a little different. Until I do a video, check out Laird Statistics for info on the interpretation. Good luck
can i use this method to test whether ranking type of data has positive effect on likert scale data?
If you have ranked data, you would use a Spearman's (non-parametric) correlation. In SPSS, that requires only checking one more box in the correlation settings, easy to do.
Research By Design thank you so much ❤️❤️
anyone knows in what case one-tailed should be implied when doing a correlation test? any specific example?
It is less about a one-tailed test and more about a one-directional correlation. In other words, if these variables are positively correlated, then that is good, but if they are uncorrelated or negatively correlated, that is no good.
I have One dependent and five independent variables under each variable I have 10 questions how can I analyze it?
Roughly speaking, you can regress 1 DV on the 5 IVs in a multiple regression. There is a lot to know about regression, so I would recommend you to a good advanced stats book, but that would be the most likely type of analysis.
Question: How did you calculate the job satisfaction? Is there a formula for it? I'm sorry I'm still trying to grasp the basics...
The job satisfaction score comes from the JSS survey by Paul Spector. It was the total score from 36 items measuring 9 facets of job satisfaction. I did not cover that in this video, but I talk about it in a proviso video (these are part of a stats class). Thanks for asking.
Research By Design Do you have specific link to it?
Hello! really helpful video! Is it possible to give us the source of this example and data set? (is it from a book, a survey or somewhere else for instance)
Check the link in the description of this (or any of the other) video. There is a link that Google Drive. In that folder you will find the dataset. Right click to download. Good luck.
These videos are amazing many thanks! But I am slightly confused about their sequence. Which one is the previous video? Is it number 12? But it's about ANOVA.
The videos go in order by week and then the number of the video. This is video 13-8, so the previous video is 13-7.
For example I have likert scale questionnaire for the two variables, can I get the mean of the two variables then do the pearson!?
**sorry, i am not an expert. I just need it for my research :(
This is a general answer to whether you COULD do it (I can't tell if you SHOULD do it without knowing your specific questionnaire). Yes, you could average multiple items (even if they are measured using a Likert scale) and then correlate them with another scale (which you also averaged). The "should" question depends on whether the items are intended to be averaged, whether they are conceptually related, etc. However, if the Likert scale has at least 5 answer options, it can be treated as scale data and averaged.
What should I do if my r-square is 0?
If r-square is 0, then the variables are unrelated. No point in doing any more in the analysis if you are looking for relationship. Hope that helps.
thanks a lot.
Most welcome!
How to put the values to SPSS from the table?
Probably the simplest method would be double-click on your SPSS correlation matrix, select all, copy and paste into Excel. Un-merge cells. You can then edit the table in Excel and keep only the parts that you want.
Need a video on Cox regression analysis univariate and multivariate
Good idea. I actually got to use one of those (Cox regression) last winter for the first time. I could probably use that data as an example. Thanks!
How do I do a similar Pearson correlation with gender statistics?
What you are asking about is a point-biserial correlation. You put the scale variable in first, then the gender/nominal variable (code as 1 & 0 for easier interpretation - must be dichotomous). Then run a Pearson correlation the same as this example. Mathematically, this is the same as an independent-sample t test.
Everything is amazing except one thing. You should have provided us with the data files used in these videos
You are in luck! Check out the Description for this video to find a link to a Google Drive with all of the datasets and other tools for the course.
@@ResearchByDesign got it. thank you
anyone know how the r = 198?
Yes...the df = 198 (not "r", BTW) is because the degrees of freedom are n - 2, where n = number of pairs in the correlation.
spearman correlation please!!!
hi