Log Transformation (Log10) using SPSS with Conversion Back to Orignal Units
HTML-код
- Опубликовано: 13 дек 2015
- This video demonstrates how to conduct a log transformation (log10) using SPSS to create a normally distributed variable using SPSS. The process of converting the transformed values back to the original units is reviewed.
Dr Grande, once again - a life saver. Nothing sadder than a mature age student struggling to keep up with those shiny young ones, but your explanations, succinct, to the point are really helping me. Thanks!
Goodness me, this was outstanding. You are a statistical angel.
Very handy videos. I am doing my dissertation using multiple regression and these are nice to have.
This awesome! Thank you Dr. Grande. You saved me money and time :)
You're welcome!
Very helpful! Thank you! :)
Thanks for this helpful video
thanks dr. grande
Hi, this is really helpful thank you. I have a couple of questions about interpretation that I hope you may help me with
1. I have reflected and reversed one of my IVs- the correlation coefficients are positive instead of negatively correlated - this seems to make sense since i've reflected it- but is it okay to simply add a negative direction to the statistic?
2. You mention in this video that the back transformations can be used on the test statistics- so when I get into regressions do I need to be changing the stats before reporting? i.e. when and on what do I apply the back transformation?
Why do we need to reflect the negatively skewed data? Is that just what you do for negatively skewed data?
Dear Dr Todd Thank you for the wonderful video. I have a few question regarding the transformation of raw data for normality. Which is the best option to used log 10, square root method or inverse method? The second question, once we have done transformation data, let say if i would like to do the PCA analysis can we proceed with this transformed data? and how about the interpretation of result? do we need to convert back to the original data or proceed with the transformed data for the interpretation of result. Thank you for you respond and highly appreciate if you could answer my question.
can i use also the blom's method of transforming my data into a normally distributed ? is there an y discrepancies compared to log or reflection?
Thanks for your reply
HELLO ,ONCE i have already transformed my data into a normally distributed data set, how will i able to use those obtained data in data analysis? for instance is in multiple regression analysis wherein an equation model will be created. What will happen to the independent variables? can dependent variable be predicted using the original data value of the independent variables??
You mention in this video that you will perform the main analyses (e.g., a regression, MANOVA) on the LOG transformed data, and then after the main analyses are done, you will convert them back in order to report the values in context. However, what happens if, for example, you have seven variables, but only 6 of them are non-normal - do you do the main analyses with 6 log transformed variables and the original normal variable? How does that affect one's results? Or do you still need to transform the normal variable? Thank you!
If i were to apply correlation or linear regression on log transformed data, how can I do that? also do we only take this for correcting skewness and take the orignal for the analysis? or what? please reply.
please tell me that if i have a variable for which i want to use parametric statistics but the variable is positively skewed, so i did the log transformation, now should i perform the parametric statistics before or after back transformation. thanks
Thank you so much! This video is really helpful! I have some questions. What is the reflection and why do we have to do this for the negatively skewed variables? Thanks!
Hello Dr.
If I make a transform log 10 but my data still not normal distrubation.
what is the best suggestion from u?
tqs
Thank you Dr. Grande. Do I need to transform both my dependent and independent variable for model prediction or only the dependent variable.
Typically, only dependent variable (Y).
I see that others already asked this Q - but I can't see your answer
why you need to convert the log10 back to the original values?
In what cases it is required?
!TNX for amazing demonstration and very clear explanation
The point in back transforming is the value of the result should be log10 back not the data (Variable/ column).
I am trying to draw a path digram using lisrel , in the model all variable should be normal distributed. I am using SPSS to transfer Data to normal Data using Log10 And SQRT. Poth methods are not affected. All virables are sill non normal although N- 193. What Can I do? Need your help plz.
what we will do? we will use log negative for kurtosis also?
Hi, Why would you convert it back when you already have the data just two rows behind? Thanks
Actually, it was supposed the results converted back to the original units not the values of variable in the column, however, usually by using mean, upper and lower bound of 95% confidence interval of mean in addition of a simple formula the transformed result can be transformed back to the original unit.
We need natural log conversion pls. Thanks
I am confused why you need to convert the log10 back to the original values? Why not just use the original?
Because by transforming the data both value and its original unit will change so in terms of interpreting the output (like mean, SD of transformed variable) it might be nonsense or unreal. The point in back transforming is the value of the result should be log10 back not the data (Variable/ column).
Far id please that if i have a variable for which i want to use parametric statistics but the variable is positively skewed, so i did the log transformation, now should i perform the parametric statistics before or after back transformation. thanks
I am not as clear on the reasons for using this function just yet. So I just keep digging...
Actually, it was supposed the results converted back to the original units not the values of variable in the column, however, usually by using mean, upper and lower bound of 95% confidence interval of mean in addition of a simple formula the transformed result can be transformed back to the original unit.