Problem with SPSS data import

schroer · Post by **schroer** » Wed Nov 16, 2005 3:06 pm

Hi,

I have a problem importing data from SPSS. From SPSS I saved the data set to a tab-delimted text file and the imported the file into SmartPLS.

However, the "validate" function comes up with a number of error messages. The first ones are a number of "invalid DOUBLE" entries, and then it goes on to report that a number of columns "have more or less lines than column 0".

The first message sounds to me that SmartPLS has problems importing floating point numbers, right? Is there anything I can do about it?

The dataset has quite a few missings and two text variables, are there hints how to deal with these?

Thanks in advance and best wishes from Wuerzburg,

Joachim Schroer

cringle · Post by **cringle** » Wed Nov 16, 2005 9:43 pm

Hi,

please make sure that you use text only for the cells in the first row (variable names).

All other cells must be filled with numbers - even if you have missing values (just choose a specific number to indicate missing values, e.g. -999).

However, using correctly prepared data as decribed above, something similar occured to me. Here is what I did

1.) Save your Spss-file as *.por (long variable names and other SPSS-specific things that are not compatible to other software applications are "cleaned up".

2.) Open that file and save it as *.xls

3.) Open the *.xls and use an apropriate data format (see my first comments in this reply; besides, make sure the there are no more "." (e.g. 100.000,00) that are use to make large better readable in the data)

4.) Save the file as CSV and import it into SmartPLS

This step-by-step "clean up" of SPSS data (with lage numbers and long variable names) worked fine in my case.

Best
Christian

schroer · Post by **schroer** » Thu Nov 17, 2005 3:19 pm

Hi,

first of all, thanks a lot for your fast reply! I tried the steps you suggested (but keeping long variable names for convenience), deleted some non-essential variables, and things work fine now.

And now that I can try it, I really like the program! :-)

Best wishes,

Joachim Schroer

u-ing · Post by **u-ing** » Tue Dec 06, 2005 3:04 pm

Hi,

I have the same problem that Joachim had with importing data to smart PLS, but the proposed solution does not work in my case.

Here is what I did:

First I extracted data from an online-survey to an excel file. Then I made sure that all cells were filled with numbers and no "." Missing values are now "999".

I did not follow step 1 of Christians solution, since I have not worked with SPSS till now, so I guess there's no need to "clean up" with SPSS-specific things that are not compatible to other software applications. (Besides I have never heard of a *.pro file, just *.prn???)

Instead I saved the data set to a tab-delimited text file AND a csv file. Afterwards I tried to import data to smartpls. The validate function then comes up with following message: "Column XX (in my case 21, 25 and 57) has less or more lines than column 0"

I checked the three colums in the data file, but found no irregularities.

What can I do? Please help me!

Best wishes

Sebastian

cringle · Post by **cringle** » Tue Dec 06, 2005 4:42 pm

Hi,

obviously, your data are in the correct format:

first row --> variable names (text is allowed)

other rows --> the cells are completely filled with numbers, even for missing values (a certain number like -999 is used to indicate missing values).

Sometime it happens to me that I cannot import such files as *.csv into SmartPLS as well, especially when I copied and pasted data into Excel. In those cases, I just marked the data, made a copy and pasted them into a newly created Excel-Spreadsheet. I also suggest to get familiar with the data import and the SmartPLS CSV Editor by experimenting with the sample set of data in the download area.

Good Luck ;)
Christian Ringle

u-ing · Post by **u-ing** » Tue Dec 06, 2005 5:23 pm

Hi Christian,

thank you for your quick reply.

I tried it again with a new Excel Sheet, but smartPLS still remains really stubborn...:-/

Fortunately the data I used was just a test sample, that I used to get familiar with smartPLS, since I will not have lot of time to experiment once the "real" data is generated. I hope the real data won't cause as much problems as the test sample does...

Greetings from Frankfurt

Sebastian Hewing

d.emsley · Post by **d.emsley** » Wed May 10, 2006 7:09 am

I had the same problem as the guys below when validating the data (ie invalid DOUBLE" and "have more or less lines than column 0) and wondered whether I was going crazy. The problem was that I had to transpose the data so that the variables were in row1 and the data in rows 2 to 100 (or whatever the sample size is).

Rotterdam · Post by **Rotterdam** » Sat May 13, 2006 3:17 pm

I have a problem with transporting excel data into Pls. I get the following: Column 30 has or less more lines than column 0. Whatever I do and i have tried many times (erasing this column) , I have this problem and if I check column 30, I can not find the fault.

I hope you can help me.

cyberian · Post by **cyberian** » Sat May 13, 2006 5:16 pm

edit:
the colomn problem is directly related to the double problem ... see next post.

cyberian · Post by **cyberian** » Sat May 13, 2006 5:26 pm

d.emsley wrote:I had the same problem as the guys below when validating the data (ie invalid DOUBLE" and "have more or less lines than column 0) and wondered whether I was going crazy. The problem was that I had to transpose the data so that the variables were in row1 and the data in rows 2 to 100 (or whatever the sample size is).

the double error is this (as described above):
if you use numbers like 1,000.00 the "," and "." - i guess - are both interpreted as decimalseperator. so no correct number
the solution: convert from 1,000.00 to 1000.00

Rotterdam · Post by **Rotterdam** » Sat May 13, 2006 10:09 pm

It is true that the column problem is related with the invalid double value:

new excel data.csv: Invalid DOUBLE value in row 1, column 0
new excel data.csv: Invalid DOUBLE value in row 2, column 0
new excel data.csv: Invalid DOUBLE value in row 3, column 0
new excel data.csv: Invalid DOUBLE value in row 4, column 0
new excel data.csv: Invalid DOUBLE value in row 5, column 0

I have checked many times and read the forum, but i can not find the problem.

Help.

thanks

cyberian · Post by **cyberian** » Thu May 18, 2006 8:42 am

Rotterdam wrote:It is true that the column problem is related with the invalid double value:

new excel data.csv: Invalid DOUBLE value in row 1, column 0
new excel data.csv: Invalid DOUBLE value in row 2, column 0
new excel data.csv: Invalid DOUBLE value in row 3, column 0
new excel data.csv: Invalid DOUBLE value in row 4, column 0
new excel data.csv: Invalid DOUBLE value in row 5, column 0

I have checked many times and read the forum, but i can not find the problem.

Help.

thanks

i dont know your data, but i guess, that u have text with 5 letters in ur first column. text is only in first row allowed. get rid of the text in the rest of the data!

Rotterdam · Post by **Rotterdam** » Thu May 18, 2006 11:00 am

Hi,

As mentioned before, i still can't find the problem. I only have text in the first row (variables name). The rest of the data constists of numbers.

Are there any other things that I need to consider?

Thanks.

cyberian · Post by **cyberian** » Thu May 18, 2006 3:06 pm

hmmm .... i think it would be easier if you post some of your data or pm it to me. maybe i can help you then.

Rotterdam · Post by **Rotterdam** » Thu May 18, 2006 6:41 pm

Hi Torben,

I really appreciate your support. I will post my data.

Thanks again.

Tanna[/url][/list]

forum.smartpls.com