Note in particular the question there about why some numbers in the various Federal Reserve Bulletin articles may not match exactly the results obtained from the public datasets.
In order to calculate Medians or single specified Percentiles in SDA, a user should click on the 'Analysis' button, then click on the 'Means' tab. In the form that appears, the user should enter the name of the variable of interest into the 'Dependent:' text box. Also enter the name of a break variable into the 'Row:' text box. (You can also enter another break variable into the 'Column:' text box, but that is optional.)
In the 'Median/Percentile Options' section, select either 'Median' or 'Percentile' from the drop-down list. If you select 'Percentile', you can select the single percentile value that you want to see in each cell (the default is the 90th percentile of the dependent variable). You can also specify that the chart be based on the median or percentile, instead of on the mean of the dependent variable.
Then click the 'Run the Table' button. In the table that is created, the median or percentile of the dependent variable will appear as the first statistic in each cell.
So, if you wanted to get results for 2001 through 2007, enter 'YEAR(2001-2007)' into the 'Selection Filter(s):' text box.
You can also filter for discrete years. For example: 'YEAR(1992,1998,2001)'.
For example: Say you want to calculate mean (and median or percentile) real net worth by age group for each of the survey years 2001 to 2007. To get those results, do the following: