feng1

As can be seen in Figure , a and β are inversely related. An increase in one will cause a decrease in the other. Links of London researcher will need to make a tradeoff between the two types of errors by moving the decision line horizontally as they cannot both be minimized. Also apparent in Figure is that a simple way to increase the power of the statistical tests used is to relax a or the Type I error by moving the decision line to the left. Other ways that can be used to increase the statistical power are discussed later. DETERMINANTS OF STATISTICAL POWER It is apparent from the above discussion that the statistical power plays Links of London Rings important role in statistical hypothesis testing. After the publication of the seminal book on the subject by Cohen, there have been calls to increase the statistical power in individual studies in various disciplines including information systems Baroudi and Orlikowski, Dyba, Kampenes, and Sjoberg, Rademacher. According to Cohen, the power of a statistical test depends on three parameters the significance level chosen, the effect size, and the reliability of the sample results. The significance level is the type I error and is typically set at the percent level. The effect size measures the phenomenon under study, which is usually unknown but can be estimated from theory or from prior studies. The reliability of the sample results refers to how close a sample statistic is to the relevant population parameter. Links of London Star of David Charm reliability can be affected by many factors including the reliability of measurement, construct validity, and sample size. Among these determinants, sample size is almost always available and forms the basis of the power analysis popularized by Cohen. All subsequent statistical power analysis studies similarly use sample size, along with the significance level and the effect size, to determine the power levels of individual studies. In information systems, for instance, Baroudi and Orlikowski reported that statistical power is typically low in information systems studies. A decade later, Rademacher found that statistical power has improved in information systems studies published in a top journal, the MIS Quarterly. Does this mean that the statistical power is no longer an issue in information systems research? We answer this question and the broader statistical inference issue by analyzing the top management support literature in the next section. TOP MANAGEMENT SUPPORT Links of London Z Charm SYSTEMS SUCCESS Top management support has long been recognized as a critical factor to the successful implementation of information systems e.g. Garrity. Empirical studies of various systems have generally supported this hypothesis Ifinedo, Rangananathan, WatsonManheim, and Keeler,Young and Jordan.