Bivariate tables
Tabular analysis is an effective and intuitive way of identifying relationships between variables. Nesstar WebView makes it possible to combine many variables in one table, and to create very large tables. Because large and complex tables are more difficult to interpret, we will restrict ourselves to presenting two simple examples: the first with two categorical variables, the second with one categorical and one metric variable.
In our first example, we want to investigate the extent to which there may be gender differences with respect to turnout in an election. In Table 3 we use the "Trust" dataset, and create a cross table with "gender" as the column variable and "voted last national election" as the row variable. Because our purpose is to find out whether men and women have answered this question differently, the percentages must be computed for each of the two groups (all male answers add up to 100 %, and all female answers add up to 100 %). In this way it is possible to compare women’s answers with men's answers.
Table 3 indicates that there is only a minor gender difference: 75.1 % of the men have voted, and 74.5 % of the women have voted. So it would appear that there is a very small gender difference with respect to turnout. (Please note that both weights are applied.)
| Male | Female | Total | |
|---|---|---|---|
| Yes | 75.1 | 74.5 | 74.8 |
| No | 18.2 | 19.9 | 19.1 |
| Not eligible to vote | 6.7 | 5.6 | 6.1 |
| Total | 100.0 | 100.0 | 100.0 |
| N= | 17,394.1 | 19,118.8 | 36,512.9 |
Weight is on
Open this table in Nesstar WebView
Detailed instructions for how to create table 3.
- Open the "Trust" dataset.
- Click the icon for weighting.
- Use the weight variable "Design and Population size weight combined".
- Click "OK". Now you can proceed with the analyses.
- Select the "Table" tab, find the variable "gender" in the variable list in the left margin, left-click and select "Add to column"
- Find the variable "Voted last national election", left-click and select "Add to row".
In the second example we will address the electoral turnout issue from another angle. Many studies have indicated that turnout is lower among younger people. How is it possible to create a table that can either support or discard this hypothesis using ESS data? Age is a continuous variable with a large number of values. Because of this, a cross table including this variable will be very large and difficult to read. By using "Voted last national election" as a column variable and "age" as the measure, we can create a small table which can show us what we need - see Table 4: The mean and median (central tendency) are lower in the group that didn't go to the polls. This indicates that the level of non-voting is greater among young people.
| Yes | No | Not eligible to vote | Total | |
|---|---|---|---|---|
| Median | 48 | 37 | 17 | 44 |
| Average | 49 | 41 | 23 | 46 |
| Minimum | 13 | 4 | 14 | 4 |
| Maximum | 102 | 97 | 109 | 109 |
| Standard deviation | 17 | 19 | 12 | 18 |
| Sum | 1,325,600 | 284,841 | 50,274 | 1,660,710 |
| Count | 27,216 | 6,922 | 2,216 | 36,354 |
Weight is on
Open this table in Nesstar WebView
Detailed instructions how to create table 4.
- Open the "Trust" dataset.
- Click the icon for weighting.
- Use the weight variable "Design and Population size weight combined".
- Click "OK". Now you can proceed with the analyses.
- Select the "Table" tab, find the variable "Age" in the variable list in the left margin, left-click and select "Add as measure".
- Find the variable. "Voted last national election", left-click and select "Add to column".