How to Use the U.S. to Overseas Custom Report

General Information

All data presented in a custom report are statistical estimates, based on survey responses weighted with data from the Immigration and Naturalization Service (INS). The estimates are subject to a certain amount of error, resulting from the sampling, data collection, and estimation processes. Because of the complicated nature of the sample design, sampling variability has not been calculated for the estimates. Instead, an indication of reliability is given by the number of respondents to the relevant questionnaire item (shown as the first row in each table). The reader must exercise judgment in determining the amount of confidence to place in an estimate.

Expanding Percentage Estimates to Represent Visitors

ITA, Office of Travel and Tourism Industries has weighted the In-flight Survey results to the INS estimates of total visitors. Because this report presents weighted results, the percentage estimates found in the report can be expanded to estimate total numbers of visitors.

To do this, start with the overall volume estimate or control total for the column of interest. To illustrate an example, start with Column #1, typically the primary column in a custom report that defines the data subset for the report. The control total for Column #1 is found in the last row in Table 1 in the first column.

To further the example, how would you calculate a volume estimate for those travelers indicating that their main purpose of trip was convention or conference? Take the control total for Column #1 and multiple it by the cell percentage found under that column in Table 12, "Q10a. Main Purpose of Trip", Row #2 for convention or conference users.

Then, always round the figures to the nearest thousand; the survey estimates are not precise enough to accurately estimate the last three digits. Also, remember to use the correct control total from the corresponding column that the cell of interest is in. This process can be used with any of the percentages found within tables and is facilitate by ordering the Excel format of the report.

Sample Size and Accuracy of the Estimates

As of 2006, the Office of Travel and Tourism Industries implemented a new policy with regard to releasing visitor estimates based upon smaller sample sizes. If the number of sampled visitors to a particular destination is under 400, TI will not calculate or provide an estimate of the total number of visitors to the destination even though an estimate could be developed. Additionally, TI strongly suggests that all users of these statistics not attempt to calculate visitor estimates when it appears that the sample size is less than 400. The next paragraph contains an explanation of how to determine when a cell may have a sample that is too small to calculate an estimate.

A guideline has been developed for estimating the number of respondents in a cell. By using the first row in a table, the number of respondents, you can estimate the unweighted, number in the cell. Divide the number of respondents in the column in question by 100. For example, what is a lowest percentage that should be used for estimating visitor volumes in a custom report showing a respondent value of 5,730 for the column? Divide 400 by 5,730 with a result of .0698 or 7.0%. Looking down the column with 5,730 respondents at the top, any cell 7.0% or more probably¹ has a sample size of 400 or greater and can be used to create an estimate for that cell. Cells with less than 7.0% probably have sample sizes under 400 and should not be used to assess visitor characteristics.

If you have any questions regarding the logic of the process, or how to obtain the visitor estimates for other segments, please call the ITA, Office of Travel and Tourism Industries, at (202) 482-4029. Any staff member can answer your "how to" questions. Mr. Ron Erdmann is the program manager for the In-Flight Survey.

More detailed questions relating to the survey results should be directed to Mr. Erdmann.  For additional written information, click here.

¹ Because the data is weighted, this mathematical procedure gives estimates of sample sizes, not exact sample sizes.