>>Return to Tell Me About Statistics!
Crosstabs is an analytical method that clarifies the relationship between items by creating a summary table that crosses two categorical items (variables).
For example, things we want to clarify, such as “the percentage of patients with arrhythmia symptoms” or “the treatment satisfaction rate,” are called objective variables (or outcome variables). In contrast, people’s attributes (gender, age, region of residence, etc.) and matters (reception staff at a clinic, explanations about treatment, etc.) are called “explanatory variables” (or “causal variables”).
Crosstabs is a method for clarifying the relationship between explanatory and objective variables and can also be used to elucidate the relationship between cause and effect, that is, causal relationships.
Explanation with specific examples of 2 items
A table that simultaneously categorizes two question items based on their respective category data and enters the number of respondents and response percentages in the corresponding cells of the table is called a “cross-tabulation table.”
Look at the cells marked * in Table 1.
[Table 1] Cross-tabulation table
The top row shows that, among those who answered that they smoked, 30 people answered that they had arrhythmia symptoms, and the bottom row shows a response rate of 37.5%.
In a cross-tabulation table, the items located at the top of the table are called the “head items” (or “summary items”), and the items located on the left side of the table are called the “front items” (or “classification items”).
Additionally, when creating such a cross-tabulation table, it is called “cross tabulating the front items and the head items” or “breaking down the head items according to the front items.”
Types of crosstabs and how to read them
(1) Types of cross-tabulation tables
The cross-tabulation table calculates the percentage of the number of respondents in the leftmost column; thus, the sum of the horizontal percentages is 100%. This table is called a “horizontal percentage table.” The table that calculates the percentage of the number of respondents in the top row is called a “vertical percentage table.”
Normally, a horizontal percentage table is applied to cross-tabulation tables. To calculate the vertical %, the horizontal % is calculated by reversing the front side to indicate the presence or absence of arrhythmia symptoms and the head side to indicate the presence or absence of smoking.
Depending on the purpose, you may want to create a table with only percentages. In this case, the number of respondents as a percentage will be indicated in the margin (Table 2).
A table that shows the number of respondents and percentages together, such as the cross-tabulation table above, is called a combined table, and a table that shows only the percentages, such as Table 2, is called a “separate table.”
[Table 2] Cross-tabulation table/separated table
(2) How to decide head and front items
When creating a cross-tabulation table as a horizontal percentage table, the head item of the cross-tabulation table is the objective variable (result variable), and the front item is the explanatory variable (causal variable).
(3) How to read the cross-tabulation table
A horizontal percentage table focuses on an arbitrary category in the head item and vertically compares the percentage of that category.
Focusing on “Yes” for the arrhythmia symptoms in Table 2, the percentage of smokers was 37.5% for “smoking” and 16.4% for “not smoking,” which is interpreted as “smoking” exceeding “not smoking.”
(4) Number of values based on percentages
The number of responses based on percentages is called n number, or “n.” If n is less than 30, the response percentage will vary significantly, so the response percentage is used as a reference value.
For example, when n = 10, a change in the number of responses by 1 results in a 10% change in the proportion; however, when n = 30, the change is 3.3% (Table 3).
[Table 3]
>>Return to Tell Me About Statistics!
Comments are closed