AB+-+Edgington+(1966)


 * **Reference: ** || Edgington, E. S. (1966). Statistical inference and nonrandom samples. //Psychological Bulletin, 66//(6), 485-487. doi:10.1037/h0023916 ||
 * **Author's: ** || Edgington, E. S. ||
 * **Title:** || Statistical inference and nonrandom samples. ||
 * **Year:** || 1966 ||
 * **Journal: ** || //Psychological Bulletin // ||
 * **Retrieval Information**: || http://dx.doi.org/10.1037/h0023916 ||
 * **Bibliography**: ||  ||
 * “ Statistical inferences cannot be made concerning populations that have not been randomly sampled ” ( <span style="font-family: 'Times New Roman',Times,serif; font-size: 90%;">p. 485 <span style="font-family: 'Times New Roman',Times,serif; font-size: 120%;">). The author suggests that while his procedure will not allow for the drawing of statistical inferences, there’s nothing preventing nonstatistical inferences to be made based on “ <span style="color: #0000ff; font-family: 'Times New Roman',Times,serif; font-size: 120%;">logical considerations <span style="font-family: 'Times New Roman',Times,serif; font-size: 120%;">” ( <span style="font-family: 'Times New Roman',Times,serif; font-size: 90%;">p. 485 <span style="font-family: 'Times New Roman',Times,serif; font-size: 120%;">). The author demonstrates that with appropriate sample sizes it does not matter whether groups are randomly selected or not because the probability (based on the Mann-Whitney U probability table) of selecting all of the high-level or low-level individuals into a single group is remotely small. By using a t-test and determining every possible combination of randomizing two groups – if the t test values of the nonrandomized sample is “ <span style="color: #0000ff; font-family: 'Times New Roman',Times,serif; font-size: 120%;">a close approximation to the randomization test <span style="font-family: 'Times New Roman',Times,serif; font-size: 120%;">” ( <span style="font-family: 'Times New Roman',Times,serif; font-size: 90%;">p. 487 <span style="font-family: 'Times New Roman',Times,serif; font-size: 120%;">) a parametric test can be considered an approximation of a randomized test. This constitutes a great argument for a nonrandomized sampling under certain specific conditions. ||
 * **<span style="font-family: 'Times New Roman',Times,serif; font-size: 130%;">Additional References: ** ||  ||
 * N/A ||
 * <span style="font-family: 'Times New Roman',Times,serif; font-size: 130%;">**Retrieval Information**: || <span style="font-family: 'Times New Roman',Times,serif; font-size: 120%;">http://dx.doi.org/10.1037/h0023916 ||
 * <span style="font-family: 'Times New Roman',Times,serif; font-size: 130%;">**Bibliography**: ||  ||
 * <span style="font-family: 'Times New Roman',Times,serif; font-size: 120%;">“ <span style="color: #0000ff; font-family: 'Times New Roman',Times,serif; font-size: 120%;">Statistical inferences cannot be made concerning populations that have not been randomly sampled <span style="font-family: 'Times New Roman',Times,serif; font-size: 120%;">” ( <span style="font-family: 'Times New Roman',Times,serif; font-size: 90%;">p. 485 <span style="font-family: 'Times New Roman',Times,serif; font-size: 120%;">). The author suggests that while his procedure will not allow for the drawing of statistical inferences, there’s nothing preventing nonstatistical inferences to be made based on “ <span style="color: #0000ff; font-family: 'Times New Roman',Times,serif; font-size: 120%;">logical considerations <span style="font-family: 'Times New Roman',Times,serif; font-size: 120%;">” ( <span style="font-family: 'Times New Roman',Times,serif; font-size: 90%;">p. 485 <span style="font-family: 'Times New Roman',Times,serif; font-size: 120%;">). The author demonstrates that with appropriate sample sizes it does not matter whether groups are randomly selected or not because the probability (based on the Mann-Whitney U probability table) of selecting all of the high-level or low-level individuals into a single group is remotely small. By using a t-test and determining every possible combination of randomizing two groups – if the t test values of the nonrandomized sample is “ <span style="color: #0000ff; font-family: 'Times New Roman',Times,serif; font-size: 120%;">a close approximation to the randomization test <span style="font-family: 'Times New Roman',Times,serif; font-size: 120%;">” ( <span style="font-family: 'Times New Roman',Times,serif; font-size: 90%;">p. 487 <span style="font-family: 'Times New Roman',Times,serif; font-size: 120%;">) a parametric test can be considered an approximation of a randomized test. This constitutes a great argument for a nonrandomized sampling under certain specific conditions. ||
 * **<span style="font-family: 'Times New Roman',Times,serif; font-size: 130%;">Additional References: ** ||  ||
 * N/A ||
 * **<span style="font-family: 'Times New Roman',Times,serif; font-size: 130%;">Additional References: ** ||  ||
 * N/A ||