On Tue, Jul 14, 2015 at 12:23 AM, John Darrington < j...@darrington.wattle.id.au> wrote:
> On Mon, Jul 13, 2015 at 04:17:17PM -0400, Ether Jones wrote: > On Mon, Jul 13, 2015 at 3:10 PM, John Darrington < > j...@darrington.wattle.id.au> wrote: > > > On Sun, Jul 12, 2015 at 10:45:14PM -0400, Ether Jones wrote: > > ???Can someone please post a small working example script using > > GLM??? (with no > > constant term) . > > > > Here is one such example: > > > > begin data. > > 1 4 332 > > 1 4 380 > > 1 4 371 > > 1 4 366 > > 1 4 354 > > 1 0 259.5 > > 1 0 302.5 > > 1 0 296 > > 1 0 349 > > 1 0 309 > > 2 4 354.67 > > 2 4 353.5 > > 2 4 304 > > 2 4 365 > > 2 4 339 > > 2 0 306 > > ?????? > > 2 0 339 > > 2 0 353 > > 2 0 351 > > 2 0 333 > > end data. > > > > ?????? > > ?????? > > ?????? > > glm points by Factor0 Factor1 > > /intercept=exclude > > . > > > > Thank you, but ???I ran that script with PSPP ???0.8.5 on Windows and > got the > following output: > > Oops. I dropped the first line when I pasted that code. Here it is: > > data list notable list /Factor0 * Factor1 * points (F10). > > Sorry about that. > Thank you. The script runs now. Here is the output I got: warning: GLM is experimental. Do not rely on these results. Tests of Between-Subjects Effects #=================#=======================#==#===========#=======#====# # Source #Type III Sum of Squares|df|Mean Square| F |Sig.# #=================#=======================#==#===========#=======#====# #Model # 2264685.69| 4| 566171.42| 988.21|.000# #Factor0 # 2256332.03| 2| 1128166.02|1969.12|.000# #Factor1 # 2261176.15| 2| 1130588.07|1973.35|.000# #Factor0 * Factor1# 2259214.79| 2| 1129607.40|1971.64|.000# #Error # 9166.87|16| 572.93| | # #Total # 2273852.56|20| | | # #=================#=======================#==#===========#=======#====# ... but I don't see the regression coefficients (least-squares model parameters) anywhere in the above report. I am trying to get PSPP to display the regression coefficients (with no intercept) and the standard error of the regression coefficients. Here is the output from an Octave script showing what I am looking for: + Y = [332, 380, 371, 366, 354, 259.5, 302.5, 296, 349, 309, 354.67, 353.5, 304, 365, 339, 306, 339, 353, 351, 333]' Y = 332 380 371 366 354 259.5 302.5 296 349 309 354.67 353.5 304 365 339 306 339 353 351 333 + X = [[1, 4]; [1, 4]; [1, 4]; [1, 4]; [1, 4]; [1, 0]; [1, 0]; [1, 0]; [1, 0]; [1, 0]; [2, 4]; [2, 4]; [2, 4]; [2, 4]; [2, 4]; [2, 0]; [2, 0]; [2, 0]; [2, 0]; [2, 0]] X = 1 4 1 4 1 4 1 4 1 4 1 0 1 0 1 0 1 0 1 0 2 4 2 4 2 4 2 4 2 4 2 0 2 0 2 0 2 0 2 0 + [cases, predictors] = size (X) cases = 20 predictors = 2 + degrees_of_freedom = cases - predictors; + N = X' * X; + d = X' * Y; + ## (aka model parameters) + coefficients = N \ d *coefficients = 175.88 22.026* + Ye = X * coefficients; + ## predicted Y values + residuals = Y - Ye; + SSres = residuals' * residuals; + MSE = SSres / degrees_of_freedom MSE = 10264 + standard_error_of_estimate = sqrt (MSE) standard_error_of_estimate = 101.31 + Ni = inv (N); + standard_error_of_coefficients = sqrt (diag (Ni) * MSE) *standard_error_of_coefficients = 19.32 10.8* + ## actual, predicted, residuals + [Y, Ye, Y - Ye] ans = 332 263.98 68.021 380 263.98 116.02 371 263.98 107.02 366 263.98 102.02 354 263.98 90.021 259.5 175.88 83.624 302.5 175.88 126.62 296 175.88 120.12 349 175.88 173.12 309 175.88 133.12 354.67 439.85 -85.185 353.5 439.85 -86.355 304 439.85 -135.85 365 439.85 -74.855 339 439.85 -100.85 306 351.75 -45.752 339 351.75 -12.752 353 351.75 1.2484 351 351.75 -0.75164 333 351.75 -18.752 + exit (); Here is the output from Excel: SUMMARY OUTPUT Regression Statistics Multiple R 65535 R Square -9.359947893 Adjusted R Square -9.99105611 Standard Error 101.313174 Observations 20 ANOVA df SS MS F Significance F Regression 2 -166924.5475 -83462.27374 -8.131269762 #NUM! Residual 18 184758.4659 10264.35922 Total 20 17833.91846 *Coefficients Standard Error* t Stat P-value Lower 95% Upper 95% Lower 95.0% Upper 95.0% Intercept 0 #N/A #N/A #N/A #N/A #N/A #N/A #N/A X Variable 1 *175.8758182 19.31966423* 9.103461429 3.71387E-08 135.2866784 216.464958 135.2866784 216.464958 X Variable 2 *22.02581818* *10.80002063* 2.039423713 0.056365119 -0.66420076 44.71583712 -0.66420076 44.71583712 RESIDUAL OUTPUT Observation Predicted Y Residuals Standard Residuals 1 263.9790909 68.02090909 0.707709876 2 263.9790909 116.0209091 1.207116227 3 263.9790909 107.0209091 1.113477536 4 263.9790909 102.0209091 1.061456041 5 263.9790909 90.02090909 0.936604453 6 175.8758182 83.62418182 0.87005099 7 175.8758182 126.6241818 1.317435847 8 175.8758182 120.1241818 1.249807903 9 175.8758182 173.1241818 1.80123575 10 175.8758182 133.1241818 1.38506379 11 439.8549091 -85.18490909 -0.886289264 12 439.8549091 -86.35490909 -0.898462293 13 439.8549091 -135.8549091 -1.413475093 14 439.8549091 -74.85490909 -0.778812855 15 439.8549091 -100.8549091 -1.049324629 16 351.7516364 -45.75163636 -0.476013704 17 351.7516364 -12.75163636 -0.132671837 18 351.7516364 1.248363636 0.012988349 19 351.7516364 -0.751636364 -0.007820249 20 351.7516364 -18.75163636 -0.195097631 Here is output from an R script showing what I am trying to do in PSPP (regression coefficients without intercept, and standard error of coefficients): > x0=c(1,1,1,1,1,1,1,1,1,1,2,2,2,2,2,2,2,2,2,2) > x1=c(4,4,4,4,4,0,0,0,0,0,4,4,4,4,4,0,0,0,0,0) > points=c(332,380,371,366,354,259.5,302.5,296,349,309,354.67,353.5,304,365,339,306,339,353,351,333) > > Ab <- data.frame(points,x0,x1) > > LM <- lm(points ~ 0 + .,data=Ab) > > LM Call: lm(formula = points ~ 0 + ., data = Ab) *Coefficients: x0 x1 175.88 22.03 * > > summary(LM) Call: lm(formula = points ~ 0 + ., data = Ab) Residuals: Min 1Q Median 3Q Max -135.85 -53.03 34.63 109.27 173.12 *Coefficients:* *Estimate Std. Error* t value Pr(>|t|) x0 *175.88 19.32* 9.103 3.71e-08 *** x1 *22.03 10.80* 2.039 0.0564 . --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 Residual standard error: 101.3 on 18 degrees of freedom Multiple R-squared: 0.9187, Adjusted R-squared: 0.9097 F-statistic: 101.8 on 2 and 18 DF, p-value: 1.544e-10 > > sqrt(diag(vcov(LM))) x0 x1 19.31966 10.80002
_______________________________________________ Pspp-users mailing list Pspp-users@gnu.org https://lists.gnu.org/mailman/listinfo/pspp-users