How to import data for contingency tables, survival tables and parts of whole tables?

By Avishek Majumder on December 12, 2018

In the previous article, the selection cases for importing data in case of XY, Column and the Grouped table was demonstrated. The datasheet changes for every type of data table chosen. Out of the six, three were explained in the previous article, hence here contingency tables, survival tables, and parts of the whole have been demonstrated for importing the data.

Contingency tables

Contingency tables have a different usage than the previous three data tables. Their main application is to evaluate the actual number of subjects (or observations) that fall into the categories defined by the rows and columns of a table (GraphPad Prism, 2017). The rows and columns can be defined in different ways, based on experimental design. In this data table, rows are defined by the subjects while the column represents different outcomes (Motulsky, 2017). This method again changes when used in a case-control model, where columns represent a different group of subjects, identified based on the presence or absence of disease or intervention. In contrast, the row represents a different exposure the sample have had in the past. In the case of cross-sectional study, first the group of subjects is selected, and then they are categorized by exposure (different rows) and disease (different columns). In this type of data table, analyses such as Fisher’s exact test, Chi-square test and odds ratio are calculated (Motulsky, 2017).

Types of models from the data table Contingency
Types of models from the data table Contingency

In the subsequent image it can be seen there is not much change in the data sheet after choosing two different models from the data table. The different rows represent usually treatments while columns represent alternative outcomes. Each value is the actual number of subjects for every case-controlled study (Motulsky, 2017). Hence the entry of actual numbers is very important in this type of data table. Normalized values or percentages are to be ignored or avoided as the values may become negative (GraphPad Prism, 2017). In this case, the software will not allow a minus sign or a decimal point or create sub-columns. If such are the conditions of the data, it is recommended that the grouped data table be used.

Datasheets form different models of the Contingency tables
Datasheets form different models of the Contingency tables

Survival tables

These tables are used to enter information for each subject of the case study (GraphPad Prism, 2017). The data is calculated on the basis of percentage survival at each time and plots a Kaplan-Meier survival plot (Motulsky, 2017). The software may also compare survival with the log-rank and Gehan-Wilcoxon tests. Each row represents a distinct subject and column represents a treatment. In this type of data table, the analyses supported are Kaplan-Meier, Log-rank, and Wilcoxon-Gehan (Motulsky, 2017).

Types of models in the Survival tables
Types of models in the Survival tables

When Prism plots a survival curve, it can include SE or 95% CI error bars (GraphPad Prism, 2017). These are computed as part of the Kaplan-Meier method for creating the survival curve. However, unlike other data tables here it is required to enter error values directly they are computed from all the data (Motulsky, 2017).

Datasheets from different models of survival data table
Datasheets from different models of survival tables

Parts of the whole table

The last data table is Parts of the whole table, used to find fractions of information when the total is each value of the evaluated data (GraphPad Prism, 2017). The most common usage of this table is in pie chart formation. Apart from this, a fraction of the total and chi-square goodness of is analyzed using this data table (Motulsky, 2017). The data can either be entered into columns as required but only the values entered into the first column will be automatically graphed (Motulsky, 2017). The data sheet from the models in this data table is also the same is presented below.

Types of models from the Parts of whole data table
Types of models from the Parts of the whole data table
Datasheet from the models of the data table
Datasheet from the models of the data table

Now, from this article and the previous article, it’s evident that graphs are based on the analyses and which are in turn based on the type of model and data table chosen. Thus, it must be clear as which data table is to be used for the particular graph or plot.


Avishek Majumder