Correlation Finder

1. On the Results page, click the Data Lab menu tab, and drag the Correlation Finder into place in your results.
2. Click the pencil icon to edit the name that appears in the Question Header.
3. Click the drop-down and select SigTest (t/z test) or Crosstab (chi²).
4. Click the Choose a banner drop-down to select a banner from the list of demographic traits, questions, or Virtual Questions. Click the second banner drop-down to refine your variable.
   Note: The banner should be the independent variable that you are studying; the underlying factor that can lead to statistical significance. 
5. Click the Choose an attribute drop-down to select an attribute to check for correlations.
   Note: The attribute should be the dependent variable that you’re using to explore statistical significance.
6. Click Generate data to populate the chart. Once a Correlation Finder is added to your stats report you can export the chart and data table to PowerPoint or download a static image of the chart.
   

Once the chart populates you can customize the visualization a few ways:

1. Click the charts icon in the upper left corner to toggle between Interlaced Chart and Grouped Chart.
   Interlaced Chart: organizes your data by attribute along the X-axis
   Grouped Chart: organizes data by banner along the X-axis
2. Click the % drop-down to toggle the view to actual numbers instead of percentages.
3. Click the chart icon in the upper right corner to change the chart visualization output.
4. Hover over the bars to view significance data.

Cells marked with a capital letter indicate a statistically significant difference between that cell and the cell of the group indicated by the matching letter, at the primary confidence level (default 95%).

Cells marked with a lowercase letter indicate a statistically significant difference at the secondary confidence level (default 90%).

To change either of these confidence level thresholds, click the Banner Tables menu, and adjust the sliders accordingly. 

In a Crosstabs, color variation appears in the cells for which the null hypothesis (which says that the banner and attribute are completely independent) is rejected; there is statistical significance in the colored cells, indicating that the banner and attribute are related and not independent from one another. The darker the shading, the stronger the correlation.

1. Click Legend in the upper left corner of the table to see the thresholds set for color variation.

1. Click on the Banner Tables menu tab in the left side menu.
2. Adjust the sliders to your preferred primary and secondary confidence levels. Your Correlation Finder results will automatically update both in the tables and in the Legend.

1. Click the Banner Tables menu.
2. Click the drop-down to select from Banner Groups.
3. Click the Add banner drop-down to select a banner to analyze. You may add as many banners as you like, from demographic traits to questions in your survey.
4. Adjust Confidence levels, if needed.
5. Check the box(es) next to the respective Summary Metrics you'd like to include.
6. With all your banners added, click Run and Download to export a Traditional Significance Test with p-values