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
