How to insert negative data using SPSS? since SPSS consider all negative values (e.g., -1,-2,...) as missing values
Hello Ehab. SPSS does not by default treat negative values as missing. If they are being treated as missing, they are user-defined missing values. I.e., someone must have issued a MISSING VALUES command--see the help for MISSING VALUES at the link given below. To remove the user-defined missing values, issue another MISSING VALUES command with no list of values. E.g., to remove user-defined missing values from variables x, y and z:
MISSING VALUES x y z ().
HTH.
https://www.ibm.com/support/knowledgecenter/en/SSLVMB_25.0.0/statistics_reference_project_ddita/spss/base/syn_missing_values.html
As Bruce Weaver says, you need to use an explicit command to treat any value as missing in SPSS. But is there some reason to record the original values (-1 etc.) if you are always going to exclude them from your analysis? In particular, one common technique in this case is to give them all the same value such as 99 -- assuming that score is outside the range of allowable scores. If your data are already entered, you could recode everything below zero to equal 99.
Actually, I want to examine the effect of intervention on A1c so, all positive results will be negative values (A1c value will be lowered), when I transfer this data to SPSS, all negative values are missed so I can't make T test or Man whitney test.
I did your suggestions but I can't get my original negative values, what is the solution in this case?
OK, I think I understand what you want to do. If you enter the A1c values from the pre- and post-tests, you can use COMPUTE to subtract the earlier score from the earlier one to get a change score. SPSS will have no problem including negative values in this result.
Also, this change score will interval-level so you can use a t-Test rather than worrying about a non-parametric test.
Thank you professor David, but I want to insert the negative result directly without going to compute function, is that possible?
Thank you for the answers provided i also had the same problem
Hello Ehab. To borrow a term invented by David Marso (in the SPSS discussion forums), my ESPss is not working very well this morning. It would help immensely if you described the design and the variables in more detail. It would help even more if you attached a small dataset that shows the original data. (CSV format would be good to make it accessible to everyone.)
From what you have said, I assume A1c is the dependent variable, is that right? And there is a treatment of some kind. I think it has 2 levels, right? I think David L Morgan inferred that it is a within-subjects treatment (i.e., repeated measures). But you mentioned the Mann-Whitney U test as one option, and it is for use with independent groups. So much remains unclear about the nature of your design. Thanks for clarifying.
The likely problem is the dash sign. Different software store the different dash signs in different ways, so that if you transfer across systems what you mean as the negative sign for one software is interpretted as an unknown character string in another.
At this point, the simplest approach might be to create a small input sheet in SPSS with a few negative values, and then see what you get after you save it. If you are at all familiar with inputting new datasets in SPSS, this should take less than five minutes.
My original data was in an Excel worksheet with field values in negative. Before getting it into excel, I reformatted these negative numeric values into string. Once in SPSS, I altered the type of these string variables into numeric (specifying the level of level of measurement as scale and modifying the width and decimal to suit the data).
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS