Timeline for How to deal with non-binary categorical variables in logistic regression (SPSS)
Current License: CC BY-SA 2.5
11 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jan 20, 2017 at 4:54 | comment | added | Sympa | Student T, I am not sure that is the case. Your example feels like seasonality. When you introduce seasonality variables in a time series model, you do not include dummy variables for 12 months in the model, you have to leave one out. But, let's say you are dealing with 3 different US States: CA, AZ, and TX. In such a case, I don't think you would have to leave one out. I am not entirely sure of my position. You are welcome to correct me if I am wrong. And, maybe my 3-state thing is a bit different than a 3-level categorical variable. Maybe in the later, you would do as you suggest. | |
| Jan 19, 2017 at 3:04 | comment | added | SmallChess | Shouldn't a 3-level categorical be coded with only two binary variables? | |
| Oct 7, 2010 at 20:36 | comment | added | Sympa | @Skrikant, you are summarizing the situation correctly. I may have confused everyone with an earlier comment whereby I used "nominal" incorrectly. I thought this adjective related to real numbers so to speak. I realize that nominal means just a label which is what it should be when dealing with qualitative categorical variables. | |
| Oct 7, 2010 at 19:49 | comment | added | chl | @Skrikant It seems you are summarizing the situation much better than me! | |
| Oct 7, 2010 at 19:32 | comment | added | user28 | @Gaetan @chl To summarize my understanding: The features of SPSS and XLStat whereby you can specify the measurement scale (nominal, ordinal etc) decreases the data file size. However, in both instances, the software uses the correct coding scheme (e.g., expand a nominal variable with J categories into J-1 dummy variables) as part of the estimation process in the background. Would that be a fair assessment of the situation? | |
| Oct 7, 2010 at 19:21 | comment | added | chl | @Gatean Ok, in this case, the same can be done in SPSS (you have the choice between numerical/ordinal/nominal for each variable) -- then, the design matrix is constructed accordingly. | |
| Oct 7, 2010 at 19:01 | comment | added | Sympa | @Srikant. XLStat in its demo used an example a model where the dependent variable was probability of renewing a subscription. And, one categorical variable within a single column had 6 different age ranges of subscribers. Using a maximum likelihood algorithm, it can interpret each specific age range as a separate data set equivalent to a separate dummy variable in its own column. When you choose that specific column, you just have to state it is a "qualitative" variable (instead of a "quantitative" one). From everyone comments, I gather this is not something you can code in SPSS. | |
| Oct 7, 2010 at 18:42 | comment | added | chl | @Gaetan I don't follow your point unless you consider that your ordinal variable is treated as a continuous one (this might make sense sometimes, although we clearly assume that the variable can inherit the property of an interval scale as pointed by @Skrikant). Usually, a variable with $k$ levels is represented in the design matrix as $k-1$ columns, and I think this is quite independent of the software used (surely, XLStat takes care of constructing the correct design matrix as R, SPSS or Stata does). | |
| Oct 7, 2010 at 17:35 | comment | added | user28 | @Gaetan I am not sure I follow you. How exactly does XLStat transform the 'text' values of cold, mild or hot into numerical values for the purpose of estimation? If there is a method that will let you estimate the effects of categorical variables without using dummy variables surely that should be independent of the software you use as there should be some underlying conceptual/model based logic. | |
| Oct 7, 2010 at 16:07 | comment | added | user28 | @gaetan I do not understand the remark about a single column vs multiple columns. Are you suggesting that categorical variables should be coded as 1, 2, 3 etc in a single column instead of using dummy variables? I am not sure that makes sense to me as you are then imposing an implicit constraint that the difference in the effect on dv between leve1s 1 and 2 is the same as the difference in the effect on dv between levels 2 and 3. Perhaps, I am missing something. | |
| Oct 7, 2010 at 15:56 | history | answered | Sympa | CC BY-SA 2.5 |