Median Test – Ordinal test that rejects the hypothesis of the equality of two populations (one sided) in the event that there are too few observations of a sample that is greater or smaller than the median of the samples combined. See also: Median/Hypothesis/Statistical Test

N – Statistical symbol for the size of the population from which a sample is taken. Example: a certain region has 1.9 million motorcyclists; therefore, the population is N = 1.9 million. See: Population/n

n – Statistical symbol for the size of a sample or the number of subjects. Example: “n = 100” signifies 100 subjects or persons in a sample. See also: Sample/N

Nomogram – A graphical representation of, for example, the confidence margin. Many research reports include a nomogram. Directions for use: locate the sample size on the X axis (for example 500). Proceed upwards from this point to the established sample results (for example: 80%); thereafter, locate the corresponding confidence margin on the Y axis. See also: Confidence Margin

Non-parametric Statistical Test – A test in parametric statistics. The model of this test does not specify conditions concerning the parameters of the population from which the sample has been taken. See also: Non-parametric statistics

Population – Syn: Universe A collection of all objects or persons from a specific class, as opposed to a sample that covers a (representative) segment of the population. All people on earth form a population, as do all residents of London or all Harvard students. See also: Sample/Hypothetical Population

Projection – The conversion of a sample score to one concerning the entire (research) population. A primary requirement is that the sample be representative for the population. For example: when in a sample of 1000 adult women, 100 readers of a particular magazine are found, ten the number of readers in the population of adult women is also estimated at 10%. If this population consists of 4 million women, the projection will be 400,000 (10% of 4 million).

Sign Test – A statistical test that is based on the symbols (+ or – ) of data rather than their magnitude. This test is used, among others, for trends in time series. Differences can be expressed in terms of + or -. the test is simple but it lacks strength, since not all information is utilized. See also: Statistical Test

Spearman Rank Correlation-Coefficient Test – One of the oldest of statistical tests. It is applicable for the testing of the correlation between variables that have been measured on an ordinal scale, permitting the classification of objects or persons in two series. See also: Statistical Test/ Statistics/Ordinal Scale

Statistic– Every statistical test has a magnitude, a constant that can be calculated. The calculation is performed with the aid of the formula. A number of operations are performed that are sometimes simple and at other times extremely complex and time consuming. The magnitude established in this manner is compared to relevant number in the table. This table indicates, per sample size, whether the magnitude in relation to the table number is “true” or based on coincidence. See also: Statistical Test

Statistical Significance – An “effect” (research finding) is statistically significant when the statistical measure utilized falls outside of the acceptable limits of coincidence. That is, the hypothesis that the effect is not real (is based on coincidence) is rejected. See also: Level of significance/Statistical Test

Statistical Test – A statistical operation aimed at the determination of the significance of a statistical datum. Only after having been tested can the datum be said with certainty to be either “true” or based on the coincidence. See also: Level of Significance/Statistical Significance/ Statistics

Wilcoxon Test – The “official” name for this statistical test, named for its designer, reads, “Wilcox on Matched Pairs Signed-Ranks test.” This test is used to evaluate scores expressed on at least an ordinal scale. The differences in the size of the pairs 9objects, persons) are tested. This test is used, among others

A.I.D. – Frequently used abbreviation for Automatic Interaction Detector, a statistical technique for the separation of groups. See also: Automatic Interaction Detector

Analysis of Variance – A series of statistical techniques that assigns the variability of data to the contributions of different sources. It clarifies or explains the origin of the data. See also: Variance

Asspat Method – A research technique developed and exploited by Dutch research bureau, Socmar. The Asspat Method (Associated Patterns) is based on simple proposition. The respondent is given a matrix in which the lines are made up of statements (qualities or properties) and the columns are made up of brands. The matrix thus formed is used for a whole set of respondents. The method can handle a matrix of only 50 x 20. The respondent associates the statements with the brands by indicating with a cross that which he feels associates best with which. A second analysis processes the information filled in by respondents and produces indices. An “expected” value per cell is calculated from the total frequencies found in the lines and columns. The individual results are then expressed as an index of the expected value (=100). By this reweighting of the association frequencies, a correction can be applied to the non-brand awareness. Brands that do not enjoy a high brand awareness can obtain a high rating by means of this reweighting procedure. See also: Benzecri Analysis/Matrix/Statement

Automatic Interaction Detector – Usually abbreviated to A.I.D. It is statistical technique having as a fundamental concept the sequential identification and separation of subgroups, per event, permitting the selection of the set subgroups that limit the errors in the predicting of the dependent variables, to as great a degree as possible, where the number of distinct groups is concerned. See also: Dependent Variable

Benzecri Analysis – A technique of analysis, developed by the Frenchman Benzecri, for application in the Asspat Method (Benzecri himself calls this technique “analysis of correspondence”). The analysis is realized in a projection on a two-dimensional plane. This analysis is based on association indices. A factor analysis is carried out on these indices in order to establish the two most important factors that, thereafter, serve as the axes of a two dimensional space. On the basis of factor loading, brand names, as well as consumer statements, are projected onto these axes. See also: Asspat Method

Leave a Comment