How to find the number of observations used in the SMARTPLS?

 PLS Junior User
 Posts: 1
 Joined: Tue Jan 31, 2017 11:51 am
 Real name and title: Mr. Aditya Shetty
How to find the number of observations used in the SMARTPLS?
Hello everybody,
I estimated the Path Model using Casewise deletion method as the missing value are more than 5% for the indicators. But, i get biased results. Case wise deletion definition says that any observation that has a missing value is not considered at all in the analysis. If this definition is true, my observation ( around 1447 ) will reduce to just 147, because only 147 obseravtions are without any missing value. Could you please tell me if i was right in comprehension of the definition of Case wise deletetion?
The 147 observations mentions above was calculated by me in the CSV file using the function = countif(range, "99"), after each observation to find out the number of missing values per observation.
Is there an option in the SMART PLS to find out how many observations have been used while executing the PLS algorithm.? . This info will help me to see how many observations have been actually used by the SMART PLS in Case wise deletion method.
I estimated the same path model using mean value replacement method. I got good results. What is your suggestion on which method to use for my analysis?.
Regards,
Aditya
I estimated the Path Model using Casewise deletion method as the missing value are more than 5% for the indicators. But, i get biased results. Case wise deletion definition says that any observation that has a missing value is not considered at all in the analysis. If this definition is true, my observation ( around 1447 ) will reduce to just 147, because only 147 obseravtions are without any missing value. Could you please tell me if i was right in comprehension of the definition of Case wise deletetion?
The 147 observations mentions above was calculated by me in the CSV file using the function = countif(range, "99"), after each observation to find out the number of missing values per observation.
Is there an option in the SMART PLS to find out how many observations have been used while executing the PLS algorithm.? . This info will help me to see how many observations have been actually used by the SMART PLS in Case wise deletion method.
I estimated the same path model using mean value replacement method. I got good results. What is your suggestion on which method to use for my analysis?.
Regards,
Aditya

 SmartPLS Developer
 Posts: 952
 Joined: Tue Mar 28, 2006 11:09 am
 Real name and title: Dr. JanMichael Becker
Re: How to find the number of observations used in the SMART
1) There is no single number provided on the number of observations as these can be flexible if you use, for example, pairwise missing value treatment. You could copy the latent variable scores to excel and count the number of rows. This is the number of retained observations in casewise deletion.
2) I would not recommend you to use casewise deletion if this leads to deleting almost the entire dataset. I would rather use pairwise missing values, which tries to use as much information available as possible.
2) I would not recommend you to use casewise deletion if this leads to deleting almost the entire dataset. I would rather use pairwise missing values, which tries to use as much information available as possible.
Dr. JanMichael Becker, University of Cologne, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Ja ... v=hdr_xprf
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de
Researchgate: https://www.researchgate.net/profile/Ja ... v=hdr_xprf
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de
Re: How to find the number of observations used in the SMARTPLS?
Hi Dr. Becker,
I have a similar question to Aditya. In my survey, due to intentional skip logic, my data set has a sizeable number of missing values. For example, I am only asking product satisfaction among respondents who said they've used a particular product (most people don't use all products).
To see how product satisfaction of various products contribute to the customer's overall satisfaction with the company, I've set up a 'products' latent variables embodying various product satisfaction as indicators. To deal with the missing values, I've chosen Pairwise Deletion in order to preserve as many completes as possible.
I see from your comment below that the only way to check the sample size retained by SmartPLS is to copy the latent variable scores to excel and count. To do that, I selected "Indicator Data (Standardized)" under "Base Data."
As I am looking at my product satisfaction variables, I am seeing that SmartPLS filled the missing values with 999.000 (my original missing values were 1). To check the sample size, should I simply ignore the 999.000 values?
Or should I be looking elsewhere? How do I look at the values at the latent variable level?
Thank you for your help!
Chris
I have a similar question to Aditya. In my survey, due to intentional skip logic, my data set has a sizeable number of missing values. For example, I am only asking product satisfaction among respondents who said they've used a particular product (most people don't use all products).
To see how product satisfaction of various products contribute to the customer's overall satisfaction with the company, I've set up a 'products' latent variables embodying various product satisfaction as indicators. To deal with the missing values, I've chosen Pairwise Deletion in order to preserve as many completes as possible.
I see from your comment below that the only way to check the sample size retained by SmartPLS is to copy the latent variable scores to excel and count. To do that, I selected "Indicator Data (Standardized)" under "Base Data."
As I am looking at my product satisfaction variables, I am seeing that SmartPLS filled the missing values with 999.000 (my original missing values were 1). To check the sample size, should I simply ignore the 999.000 values?
Or should I be looking elsewhere? How do I look at the values at the latent variable level?
Thank you for your help!
Chris

 SmartPLS Developer
 Posts: 952
 Joined: Tue Mar 28, 2006 11:09 am
 Real name and title: Dr. JanMichael Becker
Re: How to find the number of observations used in the SMARTPLS?
Yes, you can count the number of nonmissing values from the indicator data results and also look at the latent variable scores from the PLS output to see how many missings you would have at the latent variable level.
Dr. JanMichael Becker, University of Cologne, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Ja ... v=hdr_xprf
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de
Researchgate: https://www.researchgate.net/profile/Ja ... v=hdr_xprf
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de
Re: How to find the number of observations used in the SMARTPLS?
Thank you, Dr. Becker. Sorry for the basic question, but when I click on "Latent Variables" in my report, the table that appears is not showing up at the record level. What I see are my latent variables (as rows) and my latent variables (as columns).
Is this a setting that can be changed?
Thank you!
Chris
Is this a setting that can be changed?
Thank you!
Chris

 SmartPLS Developer
 Posts: 952
 Joined: Tue Mar 28, 2006 11:09 am
 Real name and title: Dr. JanMichael Becker
Re: How to find the number of observations used in the SMARTPLS?
Your latent variable scores should have as many rows as you have observations in your data set. Each row represents an observation in the same order as in your dataset.
Dr. JanMichael Becker, University of Cologne, SmartPLS Developer
Researchgate: https://www.researchgate.net/profile/Ja ... v=hdr_xprf
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de
Researchgate: https://www.researchgate.net/profile/Ja ... v=hdr_xprf
GoogleScholar: http://scholar.google.de/citations?user ... AAAJ&hl=de
Re: How to find the number of observations used in the SMARTPLS?
Thank you, Dr. Becker! I was not able to see it at the observation level initially because I had used consistent PLS. Once I used regular PLS, I was able to see a row for each observation.