Hello,
I have a dataset in SPSS which contains stock index returns on a daily basis. Now I need to create dummy variables for the days Monday, Tuesday, Wednesday and Thursday.
I need to link these dummies with the stock return on its day.
So for example the stock return on Monday April 2nd is 0.01%, now I want to link the dummy 'Monday' with this. My thought was to make 4 variables (the days) and then on the date of my example, the dummy variable 'Monday' is 1 every other dummy reads 0. The same for all the other dates with the other dummy variables.
After adding all the dummies for the right day, I need to do a lineair regression on the stock returns with using the lagged return and the dummies as independent variables.
Is this the right way to do it? Or is it not as easy as I think?
Can someone please explain to me how to use those dummies for weekdays correctly?
Thanks you!
Why dummy variables? I think you need time series analysis. If I am wrong please explain more.
I need to check for day of the week effect and holiday effect. So I need to run a lineair regression with Dependent Variable=Returns and Independent Variables=Lagged Return, Dummy for Monday, Dummy for Tuesday etc. Then I need to look at the coefficients of all the variables.
The problem is that I don't understand how I can add dummies in SPSS correctly.
I gather you are using SPSS, and that you want to estimate an OLS linear regression model. In that case, you could use the UNIANOVA command rather than the REGRESSION command. The former allows you to specify day of the week as a factor (i.e., categorical) variable, eliminating the need to generate your own indicator (or dummy) variables. You can find the dialog for it by clicking Analyze > General linear model > Univariate. The basic code will look something like this:
UNIANOVA Y BY factor variables WITH quantitative variables /METHOD=SSTYPE(3) /INTERCEPT=INCLUDE /CRITERIA=ALPHA(0.05) /DESIGN= list of terms in the model.
Note that the list of terms in the model can include product terms if you like. For example, if I had two explanatory variables A and B, and wished to include their product, I would do this:
/DESIGN = A B A*B
See also this UCLA webpage:
https://stats.idre.ucla.edu/spss/seminars/introduction-to-regression-with-spss/introreg-lesson3/
HTH.
It also depends on the volume of the dataset. For disaggregated datasets such as daily data, you may attempt spectra analysis. (in my opinion) But if the effects of the days f the week are to be examined as factors, then a linear regression can be run with the 7 days of the week been converted to 6 dummy variables. The interpretations can then be with respect to the referenced day.
- Badges
- Science topic
- Similar topics
- Cognitive Science
- Thinking