Suchergebnisse

If the publication decisions of journals are a function of the statistical significance of research findings, the published literature may suffer from "publication bias." This paper describes a method for detecting publication bias. We point out that to achieve statistical significance, the effect size must be larger in small samples. If publications tend to be biased against statistically insignificant results, we should observe that the effect size diminishes as sample sizes increase. This proposition is tested and confirmed using the experimental literature on voter mobilization.

WE HAVE ENTERED UPON A NEW STAGE IN POLITICAL SCIENCE as well as in the study of world affairs, the stage of large-scale computer-based world models. These models became practicable because of three changes. First, thanks to the United Nations and many national governments, many more better quality statistical data have now become available. Secondly, because of the techniques of sampling and of interviewing, we now can have survey data from very many countries and groups of people. These survey data cover many different aspects of people's views and attitudes, or of their experiences, or of their reports of what they thought they did. We can then compare the number of people who say they have written a letter with the number of letters the post office says they sent or got; thus we can, as it were, cross-examine the statistical data critically. Thirdly, we now have large computers which can store, recall, tabulate and analyze large amounts of data, if somebody works out a suitable programme for them. The computers can then tirelessly and patiently do work in minutes which individuals could not have achieved in a lifetime. Thanks to these advances in the last twenty or thirty years–the greater availability of statistics, survey and sampling techniques, and computers–we now have world models.