Advanced Version


The advanced version of the Ballot Box displays a wider range of electoral behaviors and political preferences of various groups in federal elections.

To locate a table, choices must be made from among each of the available options.

1) Select the information source (“Current Population Citizens”) from the first pull-down menu entitled Source.

2) Select a time frame from the pull-down menu entitled Year. The user can select all years in which information is available (the default) or just presidential or midterm years.

3) Select a characteristic or electoral behavior from the pull-down menu entitled Row Variable. The results will be differentiated by responses to a single question. If “All” is selected (the default), each of the response categories for the question will be independently displayed. The table will illustrate only one of the response categories when just one of the response options in the pull-down menu is selected from the choices provided beneath the “All” option

To view tables with combinations of different Citizen Characteristics, press the Add Characteristic button and make further selections. This will retrieve a table including only the subset of respondents exhibiting the first characteristic chosen who exhibit the second characteristic. If “All” respondents to the first characteristic are chosen, it will display a table of the subset of respondents exhibiting this second characteristic for each of the first characteristic’s categories.

The above selections will appear in the horizontal rows of the table.

4) Select an option from the pull-down menu entitled Column Variable. This will display tables that are differentiated by responses to a single question. The Column Variable is preset to display tables that include all available response options.

The Column Variable will appear in the vertical columns of the table.

5) If desired, select an option from the pull-down menu entitled Embedded Variable. The Embedded Variable is preset to display all available response options.

The Embedded Variable option is similar to the Add Row Variable Option choice under the Row Variable option. Both further narrow the Row Variable results. With the Embedded Variable option, “All” available response options will automatically be displayed.  With the Add Row Variable Option, the user must select just one of the available response options.

6) The Percentage Type radial buttons fundamentally change the choice of tables.

If the % by Row button is chosen, the tables will present percentages horizontally by the row variable(s), just like they are in the Basic Tools results. That is, each row displays that subgroups’ portion for the column variable’s response options, summing to 100%.

The opposite is the case if the % by Column option is chosen. In this case, each column displays the portion for each row variable’s subgroups’ response options. The column, thus, will sum to 100%.

If the % of All is selected, then the results display each cells’ percentage among all respondents.

Margin of Error

All individuals are not given a chance to provide responses to surveys. Instead, the U.S. Census current population surveys ask questions to a probability sample of individuals. Because Census surveys use a sample, the survey results will not perfectly represent the overall population of interest. Sampling error is the error in the distribution of responses to a survey question attributed to the use of a sample rather than the entire target population. Although these deviations cannot be prevented, the likely margin of sampling error can be calculated.

The margin of sampling error indicates the interval around a sample estimate (i.e. percentage in a sample) in which we can say with a certain degree of confidence that the true population parameter (i.e. percentage in the overall population) is located. Sampling error varies by statistical estimate, meaning each finding in and across polls yield a distinct margin of sampling error. The margin of sampling error hinges most importantly on three separate factors.

  1. Sampling error depends on the type of sampling used in the respondent selection process. Multi-stage probability sam­pling, the type Census surveys employ, results in higher sampling error than polls using simple random sampling.
  2. Sampling error is conditional on the relative distribution of responses to a question. The margin of sampling error decreases as characteristics approach the extremes (100 or 0 percent of voters) and increases as it tends toward the middle (50 percent of voters).
  3. Sampling error varies by the number of survey respondents. More respondents decrease the size of the sampling error.

Users may find the margin of error, preceded by a +/- symbol, at the bottom of each cell in the tables.  As with most surveys, the margin of error reported uses the conventional 95% confidence level.

The margin of error associated with each estimate is over fifty percent larger than a margin of error calculated based on random sample assumptions.  In most cases the Dimpled Chad margin of error is a conservative estimate.  However, in certain subgroups, such as highly clustered racial groups, the actual margin of error is likely higher than the Dimpled Chad margin of error.

Note: The margin of error only accounts for the error due to sampling.  Other forms of non-sampling error exist in all surveys.

Back to the top