Attached is data from 2004 to 2010. The work for 2010 is complete, but since I did not do all the work myself in SPSS I do not know the steps. Please outline the steps I need to click in SPSS 19 to complete the data for the other years just as it was for the year 2010 (what do I click in SPSS 19 to run this data?) Please check the work for 2010 and make note of any changes or improvements if necessary.
Can you run the data some of the other years (2004-2009) and see if the results hold true. Please attach the reports and explain them. Include the steps of how to get the WLS regression and the interaction effect. Also with the interpretation; specifically how to recognize a residual plot is hereroscedastic. .
The output report and data is in SPSS 19. The dependent variable is ACT scores (representing student achievement) and the two independent variables are percentage of economically disadvantaged and per-pupil district level instructional expenditures.
In the output file named "outpuweightedlinear2010withresidualplot" you will see that the residual plot is hereroscedastic. The variation increases on the right hand side of the plot. I am not sure what this means.
The file named output 1.spv displays the WLS regression output. The file named wls_interaction displays the WLS regression output where the interaction effect is included. The residual plot shows that the best model is obtained with WLS regression when the interaction was included. Please examine the outputs. I am unsure of the steps of how to get the WLS regression and the interaction effect. What do I need to click in SPSS? I also need help with the interpretation.
From my viewpoint the WLS model seems to work for 2010. The data is sorted in ascending order using the Percentage of Economically Disadvantaged as the first and the Per Pupil Expenditure as the second variable (the data file is attached named "sorted"). The relationship between the Per Pupil Expenditure and ACT was linear for the first 21 cases. For the rest of the data (22 through 116) there is virtually no relationship between Per Pupil Expenditure and ACT. You can see that in the file named "2 correlations." In short, it seems like for the districts where the percentage is economically disadvantaged students is 52% or less, there is a linear relationship between Per Pupil Expenditure and ACT. In the districts where the percentage is economically disadvantaged students is more, there is virtually no linear relationship between Per Pupil Expenditure and ACT. Is this also true for the other years (2004-2009)? Please explain.
The solution answers the questions regarding part one of WLS in SPSS 19.