Pearson Correlation Analysis on SPSS with Assumption Checks and Write Up (APA style) 

Thanks Kavitha, glad it helped!

Thanks Mashael, glad it helped!

Thanks Elaine, glad it helped!

Thanks Luisa, glad it helped!

Thanks Biruk, glad it helped!

Thanks Joe, glad to hear it helped!

Hi Carol, glad to hear the video was useful! I'll add canonical correlation to the list!

Hi Darren, thanks for your question and glad the video helped! The APA Publication Manual doesn’t really give much guidance about this, though it does suggest that it’s OK to include multiple figures in a single appendix: “If multiple tables and/or figures (but no text) are combined into one appendix, label and title the appendix and also number and title the tables and/or figures within the appendix (e.g., Tables D1 and D2 are two tables in Appendix D).”

Hi Tayo, Thanks for your question. One option to consider would be running a Spearman correlation instead for the variable with non-normal data as this test doesn't assume normality. I have a video on this test too if you're interested: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-pcSTjVFu-xI.html&ab_channel=DavidRobinson%2CPhD Hope that helps! David.

Hi Moeez, Thanks for your question. Regarding running the analysis itself, the process is the same. For example, if wanted to explore the relationships between three variables instead of two, you would just transfer all three variables into the Variables box (in the window that opens when you go to Analyze, Correlate, Bivariate). Regarding the assumptions, the process for checking normality would be the same. For example, if you wanted to check the normality of three variables instead of two, you could just transfer all three variables into the Dependent list box (in the window that opens when you go to Analyze, Descriptive Statistics, Explore). To check linearity and homoscedasticity, you would need to repeat the process of creating scatterplots X number of times. For instance, if you had three variables, you would need to create a scatterplot for variables 1 and 2, one for variables 1 and 3, and one of variables 2 and 3. However, if you had lots of variables this might be time-consuming, in which case you could create multiple scatterplots at once by going to Graphs, Legacy Dialogs, Scatter/Dot, then Matrix Scatter (instead of Simple Scatter). Hope that helps! David.

Hi Risa, thanks for your question. You could look into doing multiple regression as this would allow you to see whether the IV significant predicts the DV while controlling for the categorical variables. This video might be helpful: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-aIqtdDTDClo.html&ab_channel=MathGuyZero

@@DavidRobinsonPhD Thanks much David!!! I appreciate your response :-)

Hi, thanks for your question. If the assumptions are violated, you could try run Spearman correlations instead.

Hi Snehal, thanks for your question. I'm not quite sure what you mean by "features". Feel free to let me know more about what you're trying to do and I'll try to assist.

@@DavidRobinsonPhD hello, actualyy I have number of images and for each image I have HOG features. For each image 324 hog features. Now I want to remove the features which are redundant with respect to category of image. That is dimension reduction. So how to use Pearson correlation to reduce features using correlation which will give me good accuracy.

@@DavidRobinsonPhD also I want to ask whether the data vector( feature of one image) should have normal distribution while applying fir correlation.?? Because different images have different hog features and I have standardized them between 0 to 1 scale..can I apply Pearson on these features for finding mostly correlated feature with the image category..??

@@snehalgaikwad6436 Yes, the data should be approximately normally distributed, as this is an assumption of Pearson correlation. I haven’t used Pearson correlation for dimension reduction before, though perhaps you could look into principal components analysis, as discussed in this video: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-qu4la8K212M.html&ab_channel=how2stats