How to save predicted values of a regression model in SPSS:

advertisement
How to make dummy variables in regression
Consider the example in computer lesson 2.
Dependent variable: % correct spelling
Explanatory variables: Children‟s chronological age, their reading age, their
standardized reading score and their standardized spelling score
Assume we have a variable with three categories which corresponds to their
ethnicity:
Ethnicity { 1=Black 2=Hispanic 3= White }
Since it has three categories, we need to introduce two dummy variables in the
regression model as follows:
Z2=1 if the subject is Hispanic , Z2=0 otherwise
Z1=Z2=0 if the subject is White
Transform ► Recode into different variables
Click here
“old and new values”
Z1=1 if Ethnicity=1 , Z1=0 otherwise
In other
words
Z1=1 if the subject is Black , Z1=0 otherwise
Z2=1 if Ethnicity=2 , Z2=0 otherwise
Z1=Z2=0 if Ethnicity=3
Click
here
After this procedure you have made a new variable Z1 which has two values (0 & 1).
Now, we do the same procedure to make Z2
Transform ► Recode into different variables
How to save predicted values in a regression model in SPSS:
This example is from Computer lesson 2. The data is available in on the course web.
Dependent variable: % correct spelling
Explanatory variables: Children‟s chronological age, their reading age, their
standardized reading score and their standardized spelling score
Click on Analyze ►Regression ► Linear ► Save
Predicted values are saved
as a new variable in your
data set.
Here you can save the
prediction intervals
(Lower & Upper bounds).
Two new variables
(Lower and Upper) are
added to the data set.
Choose the confidence
level in the box (95% is
the default)
Prediction intervals for
Mean. Lower and upper bounds for the prediction interval of the mean
predicted response. For example the mean predicted of “ % correct spelling”
, knowing the values of chronological age, reading age, standardized reading
score and standardized spelling score. In other words, confidence interval for
the mean “% correct spelling” of all children, given the values of explanatory
variables.
Reminder from the lecture:
Individual. Lower and upper bounds for the prediction interval of the
dependent variable for a single case. It is titled as „Prediction interval‟ in the
graph below.
Hint: If you need to predict the
response variable (Y) for a certain
individual with known x-variables,
you can add a new case to your
data set and predict the response
variable and its confidence interval
as it was explained above.
Download