The last thing we want to do is program the Variable View in SPSS wrong. Especially after spending months or even years on writing the research proposal, defending the proposal, applying and getting human subjects research approval, and setting up the survey. The Variable View in SPSS is not intuitive. This blog post will give you easy instruction to complete this seemingly menial, but highly important task.
Once the data is imported into SPSS, the Variable View tab needs to be fixed. This tab should show each name of your dataset, the type of data, the width of data, decimals, etc. At first glance, many may assume it is correct as is; however, there is a 99.9% chance that the Variable View needs your attention.
First, make sure each row of data has been transferred into SPSS. I call this the gut check. Just make sure everything looks the way you would expect. The name of each column should be listed as a row in Variable View. Feel free to revise the names in the Name column so they are logically named and not too long as these names will show up in the SPSS output.
The type of data is important. The options include numeric, comma, dot, etc. Typically, we need the data to be numeric, although there may be string data if you asked an open-ended question. String data will not show up as an option in SPSS analysis commands because SPSS needs to compute numbers, not alpha characters.
To fix this, a numerical code will need to be created. Dummy codes, as some call this, basically assign a numerical value to the category. For example, instead of listing male and female, code the two options as 0 and 1. To clean up SPSS, I will sometimes remove the data that I will not use during the analysis. However, if in doubt of knowing if it should stay there or not, just leave it.
The width column is basically how long the data will be. It is similar to the length in Microsoft Excel. If your data is 01-99, the default width should be 2. It does not hurt to have this number bigger than needed, but you won’t want it smaller. I don’t usually need to touch this field in the Variable View.
The decimal column is for aesthetics. As a general rule, two numbers after the decimal is usually sufficient. If you want to round to the nearest whole number, place a zero in this field. As an FYI, the width of data for APA tables may follow this rule of thumb: whole numbers (no decimal), average/mean (1 decimal), p-values (3-decimals).
The label column is more of a notes field. I don’t usually use this, but for some researchers, it may be helpful. This is completely up to you. It will not show on your SPSS output, so feel free to use whatever you want in the label field or to ignore it entirely.
Values and measure are the last two columns that I want to include in this blog post. Values will need to be revised to match your survey. If you used 1 for male and 2 for female in your survey, this would need to match in SPSS. If you used a Likert scale, add the value labels as indicated on the original survey. For example, 1 = greatly inhibits, 2 = somewhat inhibits, 3 = slightly inhibits, etc. It is important to make sure this information matches how the survey was set up.
The last column that could throw a wrench in your plans is measure. The following three options are available in SPSS: Nominal, Ordinal, and Scale.
Notice how certain datasets could be scale, ordinal, or nominal depending on how it is presented. For example, age from 0 – 120 would be scale, but if I categorize the ages into groups of 0-10, 11-20, 21-30, 31-40, etc. then it would be considered nominal. One way to think of these options is in a hierarchy. Scale is the highest level of the hierarchy.
It is possible to change scale data to ordinal or nominal, but not possible to change ordinal or nominal data into scale. The second level in the hierarchy is ordinal. It is possible to change ordinal data into nominal, but it is not possible to move ordinal data into scale.
Finally, nominal data is just that. It cannot be changed to ordinal or scale. Think about the measure options as a hierarchy with one being Scale, two being Ordinal, and three being nominal. It is mathematically possible to move the scale to ordinal or nominal, but it is not possible to move nominal to ordinal or scale.
Is your data continuous such as age or income? Scale might be the best option if your data could be placed on a measurement stick such as a ruler, thermometer, or a scale to measure weight.
One might assume that scale is used for Likert-type scales, but that is not correct. Scale is used for a measurement of 0-1000, as an example. This may also include test scores for ACT, GED, GRE, or SAT. The possible option may only be 1 – 36, but it would still be treated the same as if the possible outcome was 1-100 or an infinite number beyond 1000.
When your data is categorized with some type of natural order or rank order, choose ordinal. This may include ranking data as mild, medium, or hot. It could also include Likert-type scales such as measuring levels of satisfaction from 1 = highly dissatisfied to 5 = highly satisfied or measuring cultural differences with 1 = greatly inhibits, 2 = somewhat inhibits, 3 = slightly inhibits, etc. Ordinal data could also include a grading system such as A, B, C, etc.
Categorical variables cannot be added, subtracted, divided, or multiplied. When your data is categorized but has no specific rank order, select nominal. This could include the department a person works in, gender, mode of transportation, region, or religious affiliation.
In closing, the Variable View in SPSS is something that deserves attention. This seemingly boring task, especially defining the measures, could have an unintended impact on your results. If in doubt, reach out to your methodologist for help or reach out to us and schedule a strategy call!