In the previous article, the basic interface of the software was explained along with the importance of each component. However, in this article, the methods and techniques in choosing the appropriate models such as grouped tables and the corresponding graphs and plots are discussed.
Importance of choosing the models or data tables
There are 6 main tables and graphs to be chosen from the pop-up box after opening the software. These are;
- Grouped tables,
- Survival and,
- Parts of the whole table (GraphPad Prism, 2017).
Choosing the right kind of table for the particular data is very important. However, note that the graphical representation is dependent on the type of data and the analyses. The importance of choosing a data table is its prerequisite for the analyses and placing the data. Any analyses and data representation for a new project can only be done with the help of choosing a data table. Choosing a data table helps create a data table that will fit the data and allow the choice of graphical representations (Motulsky, 2017). However, it is also possible to change the format of a data table at any moment of analysis. Thus, choosing a data table or a model is of utmost importance in operating in GraphPad Prism.
XY tables or graph
XY tables or graph is defined by both an X and a Y fit with linear or nonlinear regression (Motulsky, 2017). There are multiple analyzing options when XY option is chosen. These have been represented in the figure below. Part 1 on the image shows the location of the XY type of table which is to be chosen when the aim of the analysis is regression, correlation and curve related graphs. Part 2 of the image indicates the manual selection of the dataset where one can choose from the variables in X-axis and Y-axis. It is also possible to choose the type of data to be entered or required for the datasheet from the dropdown menu with “Mean, SD and N”. The part three of the image shows the tutorial or pre-defined data tables that can be used for quick selection on the basis of the type of analysis and
Differences between various data table options have been shown in the below image. From the image, it is evident that the datasheet changes while selecting different models from the manual data table selection. Similarly, the data table also varies with the selection from the Tutorial data tables. Prism can compute and graph error bars from replicate values placed inside by side sub-columns as shown in the datasheet in the image below.
Column tables are usually used if data consists of groups defined in the manner of control vs. treated, placebo vs. low-dose vs. high-dose and other similar types (GraphPad Prism, 2017). Each column defines one group. Column-based model is used when the analyses chosen for the respective dataset are frequency distribution, t-tests, one-way ANOVA and in some cases forest plot too. For better understanding look into the image below. Other important tests that are possible are:
- Mann-Whitney test,
- ROC curves,
- Bland Altman and,
- Friedman (Motulsky, 2017).
However, in biological tests mostly before-after experiment values are used in this type of data table
There is a difference between the choice of model and each datasheet that is chosen, which is evident from the following image. As in the image, it can be seen that two different models were chosen under the columns data table and both gave different data sheets. However, for one-way tables, Prism can calculate error values and create error bars automatically. All the values for a data set are to be placed in a single column. However, if one chooses the tutorial data table all the tables will be found to be similar to the variation between the two data sets are to be evaluated.
Grouped tables gives the ideology of two-way variables and the analyses constituted in them. One grouping variable (for example age of male vs. female) is defined by rows while, the other grouping variable (control vs. treated) is defined by columns (GraphPad Prism, 2017). A two-grouping variable table and graph are used to tabulate outcomes that are measurements such as weight or blood pressure, often with error bars. The analyses usually done in this data table are:
- two way ANOVA,
- mixed model ANOVA,
- Fisher’s multiple correlations,
- Holm Sidak multiple comparisons and,
- three-way ANOVA (Motulsky, 2017).
The different data tables available are represented in the image below.
The rows categorize the data by one grouping variable while the columns (datasets) categorize the data by the other grouping variable. Blank cells represent missing values, and on repeated-measures ANOVA, there is no chance of missing data values unless an entire sub-column is empty. Even for this table or graph, graph error bars will be automatically computed from replicate values from the sub-columns. Specifying the number of sub-columns is recommended to get appropriate results.
From this article, it is evident that every data table has its own model of the datasheet. The datasheet changes on the basis of the selection models in the pop-up data table box. In the next article, the next three data tables; Contingency, Survival, and Parts of the whole will be discussed.
- GraphPad Prism. (2017). GraphPad Prism7-Highlights of what’s new. Retrieved September 1, 2017, from http://www.graphpad.com/guides/prism/7/user-guide/index.htm?whats_new_in_prism_6_.htm.
- Motulsky, H. (2017). GraphPad Prism. Retrieved September 1, 2017, from https://www.graphpad.com/guides/prism/7/statistics/.