Page 125 - Culture Technology Communication

108 Carleen F. Maitland and Johannes M. Bauer

Table 4
Full Sample Models for START
Model N Adjusted R 2 R 2
Model 1a 84 .553
NWSPAPR .392 .279 .01
INTCALL .123 .344 .01
SCHLNROL .054 .308 .01
Model 1b 80 .575
NWSPAPR .394 .235 .05
INTCALL .138 .331 .01
SCHLNROL .044 .265 .01
TOEFL .021 .170 .1
Model 2a 88 .552
SCHLNROL .387 .341 .01
INTCALL .145 .339 .01
GDP_CAP .035 .244 .05
Model 2b 83 .578
INTCALL .399 .329 .01
SCHLNROL .141 .287 .01
TOEFL .035 .185 .05
GDP_CAP .023 .201 .05
Model 3 122 .614
TELEDEN .526 .502 .01
INTCALL .068 .255 .01
TOEFL .030 .199 .01
Model 4 78 .548
GEMPWR .468 .495 .01
INTCALL .092 .358 .01
Model 5 79 .467
INTCALL .438 .509 .01
PCPTHO95 .042 .257 .05
Model 6 64 .506
INTCALL .388 .398 .01
NWSPAPR .116 .369 .01
LINKS .026 .170 .1
Model 7 76 .361
NWSPAPR .304 .492 .01
CENTRAL .074 .278 .01

again a stronger predictor than School Enrollment, and School En-
rollment is stronger than GDP per capita. The reduction in the
sample size resulted in the bivariate correlation of Newpapers with
START going from .637, the highest correlation in the first analysis,
to .565, only the fifth strongest variable. As in the previous analysis
it has less explanatory power than International Call Cost, and