Hello! Thank you so so much for taking time out of your day to explain all of this. I’m currently working on my Undergraduate research project and it’s been awful. This has made my reporting so much simpler to show. Thank you!! ♥️
When reporting Spearman’s rank-order correlations in APA style, the following information should be included: “r(degrees of freedom) = r-value, p = xxx” Additionally, the r-value should be reported to 2 decimal places. For example: “r(284) = .57, p = xxx”
Hi David, thanks for your explanation. Quick question: I have 3 variables, all measured on a Likert Scale (I am measuring personality traits, emotional dysfunctioning and psychological distress for a subsequent mediation). They are normally distributed according to the Wilk-Shapiro test. Am I supposed to use Spearman's test in my case?
Hi Donnie, thanks for your question. Pearson correlation may also be an option. Although Pearson correlation is usually used with continuous data and Likert responses represent ordinal data, it’s quite common to treat mean scores for multiple items with Likert scales as continuous data. Therefore, if you have several items for each variable and will complete the analyses with mean scores, you might also consider Pearson correlation. However, Pearson correlation assumes that the relationship between the variables is linear and that there is homoscedasticity, both of which can be checked by creating a scatterplot(s). (I cover this in my Pearson video.) If either of these assumptions are violated, it may be best to use the Spearman test.
Hi, David. Thank a lot for the information. I learned a lot. I have a question. What if my p-value is 0.006, while the SPSS stated that ' correlation is significant at the 0.01 level' ? How would I write my interpretation? Is it in rs(129)=.239, p-value more than 0.001, or rs(129)= .239, p-value less than 0.05? Oh ya, my r coefficient is .239, which means it is weakly correlated?? Thanks.
Hi Lim, Thanks for your question. In APA style, the p value would be report as, p = .006. However, it's quite common to also report such values as, p < .05 or p < .01. You wouldn't say, p > .001. Re. the coefficient, yes, that would usually be considered weak. Hope that helps! David.
Thank you for the presentation. I have both nominal and ordinal independent variables (4) with (2) ordinal dependent variables. Please which one should i choose between Spearman correlation and Chi-square test to analyze the association or relationship between the variables? Or can we do both of them in the same study? i mean doing chi-square to see the relationship between the 2 nominal independent variables and the dependent variables and use the spearman correlation to see the relationship between the 2 ordinal independent variable and the dependent variable. Also, can we put more than 2 variables in the variables box when doing the bivariate correlation with spearman correlation as shown on the video. Thanks, your reply will be highly helpful.
Hi, Thanks for your questions. Yes, it's fine to use different types of tests within the same study and the approach you suggest seems viable. Re. the correlation, yes, you should be able to transfer more than two variables into the box. Hope that helps! David.
Hello! I used R Studio to run a Spearman's test. Spearman's rank correlation rho ## ## data: LARF$Aux and MARF$Aux ## S = 16, p-value = 0.7833 ## alternative hypothesis: true rho is not equal to 0 ## sample estimates: ## rho ## 0.2 This is what I got... However, I noticed that my p-value is high so how would you report this? Correct me if im wrong but based on my understanding the spearman coefficient states a positive cor but the p-value would state that there is no sig diff, right? Im lost. Please help
Hi Judy, thanks for your question. N refers to the number of participants (20), though you would subtract 2 from this to get the degrees of freedom, which is sometimes reported instead of N.
hi.. Thank you for the video. But I got some issue. There was much higher value for p value. (.552). There are two ordinal variable. can you explain it.
Hello Dr. David! Nice video! As I have seen that Spearman rho correlation has 3 conditions and one of them is that the two variables should have a monotonic relationship... if the function of the scatterplot btw the 2 variables is not monotonic, can i still use spearman rho? If not, is there an alternative in that case? since it is a non parametric test Please help!
Hi Katy, some argue that monotonicity isn’t a strict assumption of the test: statistics.laerd.com/spss-tutorials/spearmans-rank-order-correlation-using-spss-statistics.php
@@DavidRobinsonPhD Thank you for your answer! I tried to click on the link but it gives me a 404 error and page not found... is there another link that could be checked instead?
@@DavidRobinsonPhD thanks yes it works now! In the same page they also say: "If your two variables do not appear to have a monotonic relationship, you might consider using a different statistical test, which we show you how to do in our Statistical Test Selector" ... So, it means that we cannot use Spearman rho correlation if that assumption is not verified, are u sure that we can?
Hi Oumme, thanks for your question. The process would be the same; you would just enter the composite variable into the Variables box in the Bivariate Correlations window along with any other variables you’re interested in.
This video is awesome! A life-saving crash course! Question... Someone assisted me with my data computations. This is the result: (rs [69] = -.10, p = .42; rs2 [69] =.01) She says .01 is my p-value... That doesn't make sense to me. Why is .42 not the p-value? Thanks in advance for your feedback.
Hi Charlotte, thanks for your question. If the results are presented like that, the reader will conclude that .42 is the p value. If you want to check that this is correct, you can look at the "Sig (2-tailed)" row in the Correlations table as this contains the p value (aka sig value).
@@danielsaensanoh9363 Hi Daniel, thanks for your question. The Pearson analysis is normally used with interval and/or ratio data. It seems that, in your case, time would be measured with ratio data, so that variable would be suitable. If you’re measuring academic performance with grades (e.g., if the data consist of A, B, C, etc.), it may be that this variable would be measured with ordinal data, in which case a Spearman analysis might be more suitable. However, if you measured academic performance with percentages, it would be ratio data, so a Pearson analysis would be suitable.
Yes I'll be using their General average (e.g 98) for my research... But i was thinking, how should i ask that in my questionnaire. Should i ask it directly like how many hours did you spend in using social media in a day and what is your current general average? THANK YOU SO MUCH FOR YOUR REPLY.
Hello, thank you for the information. It really helps a lot. I have a question, what if the value is below 0 or has negative correlation, like -.101 do I also have to distinguish this value? or should I just say that this value does not have a correlation.
Hi Kim, Thanks for your question. If the value in the Spearman's rho row of the correlations table is negative, this indicates that the correlation is negative, which means that as one of the variable increases the other decreases. If the associated p value (aka Sig value ) is below .05, you could say that there is a significant negative correlation, and if the p value is above .05, you could say that there is a non-significant negative correlation. Hope that helps! Let me know if you have any other questions. David.
Thank you for another brilliant video. My correlation coefficient is 0.042 (looking at caregiver anxiety and their child's age. the p value is 0.584. How write this up as the p value is not siginificant? n = 176.
Hi, thanks for your question! If you're following APA guidelines, while p values below .001 are reported as p < .001, other values are reported as they appear in SPSS. E.g., a value of .584 would be reported as p = .584.
@@user-tl8uv4co3b Yes, something like that would work. However, since making this video, I've realised it might be better to present this type of info a little differently: rs(170) = .042, p = .584. Note that in APA style numbers that can't exceed 1, like correlation coeffecients and p values, don't require a 0 before the decimal point.
Hello! Thank you so so much for taking time out of your day to explain all of this. I’m currently working on my Undergraduate research project and it’s been awful. This has made my reporting so much simpler to show. Thank you!! ♥️
Thanks Ranun, glad it helped!
You are an excellent teacher,brief and concise!
Thanks Lyndah!
When reporting Spearman’s rank-order correlations in APA style, the following information should be included:
“r(degrees of freedom) = r-value, p = xxx”
Additionally, the r-value should be reported to 2 decimal places.
For example:
“r(284) = .57, p = xxx”
Thanks, is it the same for Kendall’s tau ? only replacing r with T?
How do we calculate degrees of freedom ? For instance i have 2 samples (21) and (26) ,is it 46-2=44
Some really handy tips covered in that, thank you
Thanks Sarah, glad it was helpful!
This is life saving! Thank you!
Hi Chantelle, glad it helped!
Thank you for this! :)
Thanks for watching!
@@DavidRobinsonPhD your channel is a huge help. If you're ever in Iceland I'll buy you a coffee. :)
@@sirrysif Sounds good!
Thank you!
Thanks So much.
Glad it helped!
Hi David, thanks for your explanation. Quick question: I have 3 variables, all measured on a Likert Scale (I am measuring personality traits, emotional dysfunctioning and psychological distress for a subsequent mediation). They are normally distributed according to the Wilk-Shapiro test. Am I supposed to use Spearman's test in my case?
Hi Donnie, thanks for your question. Pearson correlation may also be an option. Although Pearson correlation is usually used with continuous data and Likert responses represent ordinal data, it’s quite common to treat mean scores for multiple items with Likert scales as continuous data. Therefore, if you have several items for each variable and will complete the analyses with mean scores, you might also consider Pearson correlation. However, Pearson correlation assumes that the relationship between the variables is linear and that there is homoscedasticity, both of which can be checked by creating a scatterplot(s). (I cover this in my Pearson video.) If either of these assumptions are violated, it may be best to use the Spearman test.
@@DavidRobinsonPhD Thank you a thousand!
Hi, David. Thank a lot for the information. I learned a lot. I have a question. What if my p-value is 0.006, while the SPSS stated that ' correlation is significant at the 0.01 level' ? How would I write my interpretation? Is it in rs(129)=.239, p-value more than 0.001, or rs(129)= .239, p-value less than 0.05? Oh ya, my r coefficient is .239, which means it is weakly correlated?? Thanks.
Hi Lim,
Thanks for your question.
In APA style, the p value would be report as, p = .006. However, it's quite common to also report such values as, p < .05 or p < .01. You wouldn't say, p > .001.
Re. the coefficient, yes, that would usually be considered weak.
Hope that helps!
David.
@@DavidRobinsonPhD
Dear David,
Hi. Thanks a lot for that reply. That's help a lot.
Thank you for the presentation. I have both nominal and ordinal independent variables (4) with (2) ordinal dependent variables. Please which one should i choose between Spearman correlation and Chi-square test to analyze the association or relationship between the variables? Or can we do both of them in the same study? i mean doing chi-square to see the relationship between the 2 nominal independent variables and the dependent variables and use the spearman correlation to see the relationship between the 2 ordinal independent variable and the dependent variable. Also, can we put more than 2 variables in the variables box when doing the bivariate correlation with spearman correlation as shown on the video. Thanks, your reply will be highly helpful.
Hi,
Thanks for your questions.
Yes, it's fine to use different types of tests within the same study and the approach you suggest seems viable. Re. the correlation, yes, you should be able to transfer more than two variables into the box.
Hope that helps!
David.
@@DavidRobinsonPhD Thanks a lot
Hello! I used R Studio to run a Spearman's test.
Spearman's rank correlation rho
##
## data: LARF$Aux and MARF$Aux
## S = 16, p-value = 0.7833
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
## rho
## 0.2
This is what I got... However, I noticed that my p-value is high so how would you report this? Correct me if im wrong but based on my understanding the spearman coefficient states a positive cor but the p-value would state that there is no sig diff, right? Im lost. Please help
wouldn't N=18? Wouldn't you have to subtract 2 from 20 to get 18?
Hi Judy, thanks for your question. N refers to the number of participants (20), though you would subtract 2 from this to get the degrees of freedom, which is sometimes reported instead of N.
hi.. Thank you for the video. But I got some issue. There was much higher value for p value. (.552). There are two ordinal variable. can you explain it.
Hi Madu, a p value above .05 just indicates that the relationship is not significant.
Thank you..I think I have issue in data set. When I got the output phrase shown under the table didn't show up..what can I do for that?
Thank youu! Could you tell me how we work out variances of spearmans too please? :)
Hello Dr. David! Nice video!
As I have seen that Spearman rho correlation has 3 conditions and one of them is that the two variables should have a monotonic relationship... if the function of the scatterplot btw the 2 variables is not monotonic, can i still use spearman rho?
If not, is there an alternative in that case? since it is a non parametric test
Please help!
Hi Katy, some argue that monotonicity isn’t a strict assumption of the test: statistics.laerd.com/spss-tutorials/spearmans-rank-order-correlation-using-spss-statistics.php
@@DavidRobinsonPhD Thank you for your answer! I tried to click on the link but it gives me a 404 error and page not found... is there another link that could be checked instead?
@@katy8791 The link should be working now!
@@DavidRobinsonPhD thanks yes it works now! In the same page they also say: "If your two variables do not appear to have a monotonic relationship, you might consider using a different statistical test, which we show you how to do in our Statistical Test Selector" ... So, it means that we cannot use Spearman rho correlation if that assumption is not verified, are u sure that we can?
hello, how can we correlate composite variables (ordinal likert scales)?
Hi Oumme, thanks for your question. The process would be the same; you would just enter the composite variable into the Variables box in the Bivariate Correlations window along with any other variables you’re interested in.
This video is awesome! A life-saving crash course! Question... Someone assisted me with my data computations. This is the result: (rs [69] = -.10, p = .42; rs2 [69] =.01) She says .01 is my p-value... That doesn't make sense to me. Why is .42 not the p-value? Thanks in advance for your feedback.
Hi Charlotte, thanks for your question. If the results are presented like that, the reader will conclude that .42 is the p value. If you want to check that this is correct, you can look at the "Sig (2-tailed)" row in the Correlations table as this contains the p value (aka sig value).
@@DavidRobinsonPhD thanks!
I really need help, can i use pearson r for determining the relationship of time spent in social media(hours ) and academic performance (grades )
Also, what should i ask in the questionnaire, should it be specific? Like how many hours did you spend on social media?
@@danielsaensanoh9363 Hi Daniel, thanks for your question. The Pearson analysis is normally used with interval and/or ratio data. It seems that, in your case, time would be measured with ratio data, so that variable would be suitable. If you’re measuring academic performance with grades (e.g., if the data consist of A, B, C, etc.), it may be that this variable would be measured with ordinal data, in which case a Spearman analysis might be more suitable. However, if you measured academic performance with percentages, it would be ratio data, so a Pearson analysis would be suitable.
OMG, I didn't expect that you will reply, your so kind.
Yes I'll be using their General average (e.g 98) for my research... But i was thinking, how should i ask that in my questionnaire. Should i ask it directly like how many hours did you spend in using social media in a day and what is your current general average? THANK YOU SO MUCH FOR YOUR REPLY.
Because it seems like in order for this to work, it needs an specific value
Hello, thank you for the information. It really helps a lot. I have a question, what if the value is below 0 or has negative correlation, like -.101 do I also have to distinguish this value? or should I just say that this value does not have a correlation.
Hi Kim,
Thanks for your question.
If the value in the Spearman's rho row of the correlations table is negative, this indicates that the correlation is negative, which means that as one of the variable increases the other decreases. If the associated p value (aka Sig value ) is below .05, you could say that there is a significant negative correlation, and if the p value is above .05, you could say that there is a non-significant negative correlation.
Hope that helps!
Let me know if you have any other questions.
David.
@@DavidRobinsonPhD Thanks a lot. I admire your work.
Thank you for another brilliant video. My correlation coefficient is 0.042 (looking at caregiver anxiety and their child's age. the p value is 0.584. How write this up as the p value is not siginificant? n = 176.
Hi, thanks for your question! If you're following APA guidelines, while p values below .001 are reported as p < .001, other values are reported as they appear in SPSS. E.g., a value of .584 would be reported as p = .584.
@@DavidRobinsonPhD Thank you. So can I say there was no significant correlation found rs = 0.042, n =170, p = 0.584.
@@user-tl8uv4co3b Yes, something like that would work. However, since making this video, I've realised it might be better to present this type of info a little differently: rs(170) = .042, p = .584. Note that in APA style numbers that can't exceed 1, like correlation coeffecients and p values, don't require a 0 before the decimal point.