13.07.2015 Views

Consciousness-Based Education - Maharishi University of ...

Consciousness-Based Education - Maharishi University of ...

Consciousness-Based Education - Maharishi University of ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

U.S.-Soviet Relations and the <strong>Maharishi</strong> Effectnormal disturbances (Harvey, 1990; Brockwell and Davis, 1987; Liand McCleod, 1988). 23Sensitivity Analysis: The Seasonal Pattern Hypothesis.One potential alternative explanation for the impact <strong>of</strong> the globalWorld Peace Assemblies reported in Table 2 is that these interventioneffects may be coincidental, merely reflecting a recurrent seasonal patternin Soviet behavior toward the US. The potential plausibility <strong>of</strong> thisexplanation is enhanced by the fact that the global Assemblies wereheld every December, January, and July for the period December 1983through July 1985.To empirically examine the “seasonal pattern hypothesis,” a TFmodel was estimated in which the World Peace Assembly interventionvariable I twas replaced with an alternative “seasonal” interventionvariable. The latter took the value 1.0 for December, January, andJuly over the years <strong>of</strong> the sample prior to the first global World PeaceAssembly (December 1983), and the binary variable was otherwiseequal to zero.For the estimated TF equation, the only significant parameter estimatefor the seasonal variable was found at lag one, but the estimatewas negative, rather than positive, in sign (–20.183, t = –2.28, p =.023). This result is inconsistent with the “seasonal pattern” hypothesis:Soviet behavior toward the US over the period 1979–1983 priorto the first global Assembly significantly worsened, on average, ratherthan improved, during the specific calendar months in which the globalAssemblies were subsequently held. For this seasonal TF, none <strong>of</strong> thelags for the variable US twere significant, and the variable was thereforedropped from the model.The noise model with minimum AIC was autoregressive, with significantcoefficients at lags 1 and 5, and all roots were outside the unitcircle. 24 There were no significant autocorrelations at lags 1–24, and theLjung-Box statistic confirmed the lack <strong>of</strong> significant residual correlation.25 A large residual 3.94 standard deviations above the mean for theresidual series indicated the desirability <strong>of</strong> adjustment for outliers (Wei,1990; Chang, Tiao, Chen, 1988), as did the studentized range test (SR= 6.692, p < .01). The null hypothesis that the residuals came from a473

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!