Understanding the Data

Student Confidentiality

To protect student confidentiality, individual reports have not been generated for schools where fewer than 20 students participated in the MYDAUS/YTS. When fewer than 20 students in any one grade participated, the data for that grade was not reported.

Confidence Intervals

In the new local summary reports, the data labeled “%” is the weighted proportion of students responding in a specific way to a question; for instance, 29.0% of the middle and high school students in Maine who answered the question said they had consumed alcohol at least once in the 30 days prior to taking the MYDAUS/YTS. However, because not all of the eligible students answered that question (or even took the survey), we have to determine statistically how well the response of the participating students might represent the response of all the eligible students in the population (in our example, the population is the State of Maine). We did this by calculating the 95% Confidence Interval (CI), meaning that if all the eligible students in the population answered the question, we are 95% confident that the results would lie between the Lower Confidence Limit (LCL) and the Upper Confidence Limit (UCL). The higher the proportion of eligible students in the population answering the survey question, the narrower the confidence interval and the more confidence we have that the results represent the experience of the student population as a whole. Therefore, the higher the participation rate at your school or district, the more confidence you can have in those results.

Comparing Local Data to Maine's: Is the Difference Statistically Significant?

In the local summary reports, if the Confidence Interval around the local students' results does not overlap the Confidence Interval around Maine's results, then the local results are considered to be Significantly Different from that of Maine. If the local results are significantly lower, you will find a minus (-) under "Sig Diff", and if the local results are higher, you will find a plus (+). Please consider that in many instances Maine's substance use has been higher than national rates, so you need to decide if "lower than Maine" is good enough.

Weighting the Data

The students who participate in the MYDAUS/YTS may notrepresent all of the eligible students in the population, relative to age and gender. Because grade and gender are known to influence results, we correct for under-representation by weighting the data. For example, say there were 200 students at your high school and 100 were male and 100 were female; all 100 female students took the survey but only 50 of the male students participated. We would then give each female's response a weight of 1 and each male's response a weight of 2. The actual weighting of the MYDAUS/YTS data is considerably more complex, and is discussed in the MYDAUS 2006 Technical Report, Appendix A - Methodology, Method of Weighting.

Screening Out Unusable Surveys

There are several questions in the MYDAUS/YTS that are used to detect surveys that were answered untruthfully (usually 3-4% of the total). These surveys were not included in the data available on the MYDAUS/YTS website.

Margin of Error

In the school and district reports for 2000, 2002 and 2004, a margin of error is given under the graphs as a measure of how well the proportion of students participating in the MYDAUS(/YTS) in the given geographic area represents all eligible students in that area. It is based on the number of participating students, the number of the total eligible students enrolled, the proportion of students who choose a response (a 50/50 split is assumed), and the desired confidence level (95%). (Note: in the Local Summary Reports for 2006, a Confidence Interval for each data item replaces the one Margin of Error estimate given in the reports for earlier years.)

The margin of error is reported as a plus or minus percentage. As an example, if a school's data has a margin of error of +5%, and 50% of the participating students picked a particular response, you can be 95% sure that between 45% (50% -5%) and 55% (50% + 5%) of the entire eligible student body would have picked that response. Therefore the smaller the margin of error the more confidence we have that the results of the survey represent the entire eligible student body. The margins of error for the data for Maine, the OSA regions and the counties in 2000, 2002, and 2004 are listed in the tables below.

The margin of error calculations assume that each student in the population under study had an equal chance of becoming a participant, and that all participating students in that population answered any given question. If these assumptions are not met, the margin of error estimates are not correct.

For further information regarding margin of error, you may wish to visit the American Statistical Association's website at http://www.amstat.org.