Missing Data - Missing Indicator Method/Dummy Variable Adj.

Questions about the implementation and application of the PLS-SEM method, that are not related to the usage of the SmartPLS software.
Post Reply
SKNO
PLS Junior User
Posts: 5
Joined: Thu Jul 05, 2012 2:33 pm
Real name and title:

Missing Data - Missing Indicator Method/Dummy Variable Adj.

Post by SKNO »

Hello,
I'm considering whether it is feasible to implement the missing indicator method/dummy variable adjustment in PLS.

The background: In my modell I have one LV with a bigger amount of missing data. For my Items I used a 7-point-likert-scale, but for that LV I also have an aditional category "not relevant" for all the items in that construct. My idea was to replace the missing data by a constant (mean for example) as suggested by Cohen & Cohen 1985 and then ad one dummy variable for each manifest variable in that construct which is coded 1 if data is missing in that indicator.

My idea ist to do some form of complete case analysis only in respect to this LV (which is not concidered in that way, i know) because if i delete all those cases mith missing data, i lose alot of cases and my remaining sample will be biased. So I need a way to incorporate only those cases in that LV which are not "not relevant".

Can anyone suggest a good way to handle this? I'm not sure if the dummy variable method/missing indicator method ist really feasible in PLS. I was already searching in the forum, but i didn't really find an answer for that question.

Thanks alot in advance!
User avatar
Hengkov
PLS Super-Expert
Posts: 1599
Joined: Sun Apr 24, 2011 10:13 am
Real name and title: Hengky Latan
Location: AMQ, Indonesia
Contact:

Post by Hengkov »

Hi,

Multiple imputation may be reasonable method for handle your case. Used IBM SPSS for replace it, and import the result in SmartPLS.

Regards,
huracan
PLS User
Posts: 20
Joined: Wed May 08, 2013 1:13 pm
Real name and title:

Post by huracan »

Hello, Hengky Latan

Could i have some tips to handle multiple imputation?

Which one of imputated data sets(from original data to imputation data file 1 or _m) can we use for further analysis after performing multiple imputation operation in SPSS?

Best regards,

Ken
gordonchen
PLS Junior User
Posts: 8
Joined: Thu Aug 08, 2013 4:39 am
Real name and title:

Re: Indicators not show up when upload the CSV file

Post by gordonchen »

Hi Hengky,

I have tried to build a new data file by myself which is in CSV format. When I create a new project and upload this CSV file to the SmartPLS application, it shows a red icon on the CSV file. When I open the CSV file, it can show properly on the pane on the right hand window under 'Comma' option. But in the "Indicators" window on the left hand side, there is nothing show up!

I am not sure what is happening. I tried many times, this problem is still persist. I think that I need you help to look into my data file to identify what the problem is. It stops me to go any further to build my path model.

If you need the file to have a look, I can send you the data file. Please advise your contact email address.

Looking forward to hear from you.

Gordon.
Gorodn Chen
Post Reply