Dear list members,
I have interpretation problems. Pallkho is the % of patients in a certain
month which therapy has been initiated in certain hospital type. Zkho is the
% of patients in a certain month which have been documented by a certain
hospital type. Here is the data:
Data list free
/time (F2) pallkho (F3.2) zkho (F3.2).
Begin data
1 15,58 24,05
2 11,11 28,79
3 8,06 24,29
4 6,1 28,42
5 5,61 33,87
6 5 33,94
7 6,31 26,72
8 3,23 37,62
9 3,19 31,63
10 3,85 31,82
11 3,23 34,34
12 6,98 36,26
End data.
Save outfile='C:/trend.sav'.
Exe.
Pallkho and zkho are both significantly time depended. Right? Here is the
syntax I used:
* Curve Estimation.
TSET NEWVAR=NONE .
CURVEFIT /VARIABLES=pallkho WITH time
/CONSTANT
/MODEL=LINEAR
/PRINT ANOVA
/PLOT FIT.
* Curve Estimation.
TSET NEWVAR=NONE .
CURVEFIT /VARIABLES=zkho WITH time
/CONSTANT
/MODEL=LINEAR
/PRINT ANOVA
/PLOT FIT.
The correlation between pallkho and zkho is also significant. Right?
* Curve Estimation.
TSET NEWVAR=NONE .
CURVEFIT /VARIABLES=pallkho WITH zkho
/CONSTANT
/MODEL=LINEAR
/PRINT ANOVA
/PLOT FIT.
If I do a regression as follows, I have "something" (need interpretation
help) which is significant in my ANOVA table but the coefficients for time
and zkho are no longer significant, although there is a significant
correlation between pallkho and zkho on one side and pallkho and time on the
other. I would very much appreciate an explanation for this!
REGRESSION
/MISSING LISTWISE
/STATISTICS COEFF OUTS R ANOVA
/CRITERIA=PIN(.05) POUT(.10)
/NOORIGIN
/DEPENDENT pallkho
/METHOD=ENTER zkho time .
Thanks a lot in advance.
Answers can also be sent directly to my Email address.
Christian
Email:
[hidden email]
Locked
