NMatrix is a registered trademark around the world,
and it is patented in Japan and the USA.
O
ne of the unavoidable processes after research data are obtained is the statistical analysis process and the choice of how to present the results.
Placing a large number of figures and tables on a limited number of pages can be quite a challenge for researchers and authors.
NMatrix, a new statistical display method, is a tool that, instead of presenting a large number of graphs as they are, replaces the significance of a difference or the degree of change represented in a single graph with a single cell, and presents the results of a wide variety of tests derived from multiple testing methods in a single matrix.
NMatrix is a tool that shows its true value when multiple drugs are to be compared simultaneously based on multiple criteria.
I
f the four drugs were compared between any two points on the basis of multiple items, 204 cells would be needed for a graphical display, but again, only one NMatrix display is needed, as shown in Figure 3. When creating an NMatrix, it is possible to select either a parametric or a nonparametric method, and to apply a nonparametric method to the weighing data. NMatrix can also be used to apply nonparametric methods to quantitative data or parametric methods to ordinal data.
NMatrix is designed to be used in different ways for different types of research.
Parametric Methods
Comparison of initial values (measured values at period 1) between groups
Two-sample t-test (t-test without correspondence, for two groups) One-way ANOVA (for multiple groups) *These methods are used automatically depending on the data being analyzed.
Comparison within groups by time period 1 and each time period
Corresponding t-test (for two time periods) Multiple comparisons based on two-way ANOVA (for multiple time periods)
※These methods are automatically used according to the content of the data to be analyzed.
※Multiple comparisons can be specified as follows: Dunnett (optional), Bonferroni, Scheffe, Tukey, Fisher
Group Comparison of Actual Measurements by Time Period
Two-sample t-test (t-test without correspondence, for two groups and two time periods) Multiple comparison based on one-way ANOVA (for many groups and two time periods) Multiple test (for many groups and many time periods) These methods are automatically used according to the data to be analyzed. The following methods can be specified for multiple comparisons: Tukey type (when omitted), Bonferroni type, Scheffe type, Dunnett type, Fisher type
※Multiple tests are Bonferroni type.
Group Comparison of Change by Time Period
Two-sample t-test (t-test without correspondence, for two groups and two time periods)Multiple comparisons based on one-way ANOVA (for many groups and two time periods)Multiple tests (for many groups and many time periods) *These methods are automatically used according to the data to be analyzed. The following methods can be specified for multiple comparisons: Tukey type (when omitted), Bonferroni type, Scheffe type, Dunnett type, Fisher type
*Multiple tests are Bonferroni type.
Nonparametric Methods
Comparison of initial values (ordinal data at period 1) between groups
Fisher's exact test (for 2 groups and 2 ranks) ann-Whitney's U test (for 2 groups and multiple ranks) χ square test (for multiple groups and 2 ranks) Kruskal-Wallis H test (for multiple groups and multiple ranks)
*These methods are automatically used depending on the data to be analyzed.
Within-group comparison of ordinal data for time period 1 and each time period by group
Sign test (for two time periods and two ranks) Multiple comparisons based on Cochran's Q test (for multiple time periods and two ranks) Multiple comparisons based on Friedman's test (for multiple time periods and multiple ranks)
These methods are automatically used according to the contents of the data to be analyzed.
The following methods can be specified for multiple comparisons: Dunnett type (when omitted), Bonferroni type, Scheffe type, Tukey type, and Fisher type.
Group comparison of ordinal data by time
Fisher's exact test (for 2 groups, 2 time periods and 2 ranks) Mann-Whitney's U test (for 2 groups, 2 time periods and multiple ranks) Multiple comparisons based on the χ square test (for multiple groups, 2 time periods and 2 ranks)
Multiple comparisons based on the Kruskal-Wallis H test (for multiple groups, two time periods, and multiple ranks) Multiple tests (for multiple groups, multiple time periods, and multiple ranks) *These methods are automatically used according to the contents of the data to be analyzed.
The following methods can be specified for multiple comparisons: Tukey type (when omitted), Bonferroni type, Scheffe type, Dunnett type, and Fisher type.
Bonferroni type for multiple comparisons
Group comparison of changes in ordinal data by time period
Mann-Whitney's U test (for 2 groups, 2 time periods, and multiple ranks) Multiple comparisons based on Kruskal-Wallis' H test (for multiple groups, 2 time periods, and multiple ranks) Multiple tests (for multiple groups, multiple time periods, and multiple ranks)
These methods are automatically used according to the contents of the data to be analyzed.
The following methods can be specified for multiple comparisons: Tukey type (when omitted), Bonferroni type, Scheffe type, Dunnett type, and Fisher type.
Bonferroni type for multiple comparisons
NMatrix is a registered trademark around the world, and it is patented in Japan and the USA.
