Skip Maine state header navigation
Skip First Level Navigation | Skip All Navigation
The following is an overview of the 2009 MIYHS methodology. The full "2009 MIYHS Methodological Summary" can be found on the Reports and Fact Sheets Page.
The MIYHS sample was selected using a two-stage cluster design to produce a representative sample of public school and quasi-public school students. (Quasi-public schools are private schools with at least 60% publicly funded students.) The sampling frames for the 2009 MIYHS consisted of all public and quasi-public schools in Maine with students in the following grades:
In the first sampling stage, schools with any grades in the K/3 and 5/6 populations were randomly selected with equal probability. For the middle school and high school surveys, all schools in Maine were invited to participate. After sampling, a few schools were deemed to be ineligible due to their very small size, or because the school was considered a "special purpose school" which does not serve the general public school population.
Thirty-six (36) out of 408 schools with students in either kindergarten, 3rd grade or both were selected for a state-level sample. All schools had an equal chance of being selected. Three of these schools were deemed ineligible. Twenty-three (23) schools chose to participate. In the second sampling stage, parents of all 1,823 students in these grades were invited to take part in the survey.
In the 5/6 sample, 172 out of 402 schools with fifth graders, sixth graders or both were selected in a sample that was stratified by grade and by public health district. All schools within each public health district had an equal chance of being selected. This provided for statistically representative data at both the state and public health district levels. After sampling, 13 of the selected schools were determined to be ineligible due to their very small enrollment size. In addition to the 125 sampled schools that chose to participate, 37 volunteer schools were included in the 5/6 survey, for a total of 162 participating schools. All 12,138 students in the participating schools were invited to take part in the survey.
For the middle school survey, all 225 schools with grades 7 and/or 8 were invited to participate. Twelve (12) schools were determined to be ineligible. The remaining schools were randomly split into two groups, with Group 1 containing 108 schools and Group 2 containing 105 schools. For Group 1, classes within each school were randomly assigned to either Module (survey version) A or Module C. Similarly, within a school in Group 2, classes were randomly assigned to either Module B or Module D. Students in the same class completed the same module of the survey.
For the high school survey, all 134 public and quasi-public high schools with any grades 9 through 12 were invited to participate. Three (3) schools were determined to be ineligible. The remaining schools were randomly split into two groups, with Group 1 containing 65 schools and Group 2 containing 66 schools. Schools and then classes within schools were randomly assigned modules in the same manner as the middle school survey, as described previously. Students in the same class all completed the same module of the survey.
For the K/3 sample, surveys and consent forms were mailed to the home address of 1,823 parents. Parents who did not respond received a second mailing requesting their participation. Multiple attempts to reach non-respondents after the second mailing were made by telephone. As an incentive, parents who responded either by mail or by telephone were entered into a drawing for one of three $100 Hannaford gift certificates. In all, 833 parents participated in the K/3 survey for a participation rate of 45.7%.
For all other survey levels, parents were sent a notification of the survey and asked to let the school know if they did NOT consent to their child's participation. Schools were asked to select a day and a class period in which to administer the survey. Participating schools were asked to administer the survey during a day and a class period that all students were required to attend (e.g., English). Teachers were given a script to read to the students and written consent information was included on the front of the questionnaire. Students could opt out of the survey and were also told that they could skip any questions they did not want to answer. Schools were asked to hold make up sessions for students absent at the time of the initial administration, although the extent to which this happened varied by school.
Response rates by survey level are listed in the table below. Detailed response rates, broken down by school and student response rates and by county, public health district, and module, can be found in the full reports.
| Survey | # of Participating Schools | # of Participating Students | Overall Response Rates |
|---|---|---|---|
| K/3 | 23 | 8,333* | 31.7% |
| 5/6 | 162** | 11,011** | 71.1%*** |
| MS | 164 | 21,957 | 68.5% |
| HS | 108 | 40,392 | 65.0% |
*Parents
** Includes the 37 schools and 2,520 students in the volunteer schools.
***This response rate does NOT reflect the participation of the volunteer schools.
The MIYHS data were weighted (1) to reflect selection probabilities; (2) to reduce any bias in the estimates due to non-response on the part of the schools and/or the students; and (3) to produce estimates that can be generalized to the full population.
It was determined that data would be weighted if the overall response rate of a survey was at least 50%. The overall response rate was calculated as the school response rate * times the student response rate. Low response rates can result in a sample that is not a representative subset of the population. Thus, unweighted results from these samples may not accurately reflect student behaviors and therefore may be misleading. The K/3 survey data were not weighted due to the low overall response rate. In addition, county level data for the Franklin County middle school survey and Knox County for the high school survey had low response rates and are not weighted.
Student analysis weights were computed to facilitate analyses at the state-, county-, public health district-, and school-levels. Two sets of student analysis weights were computed for each of the 5/6, MS and HS surveys. One set of weights was computed for analyzing school-level data; another set for analyzing county-, public health district, and state-level data. For school-level data, the weights were not post-stratified, and therefore, unweighted and weighted rates are equal in all cases. The set of weights for county, public health district and state-level analyses was first constructed by county. The sum of the weights for a county represents the target student population in the county, taking student non-response, grade and gender into consideration. To obtain the target student population in the public health district, the weights were aggregated in the counties that comprised the public health district. Similarly, to obtain the target student population in Maine, the weights were aggregated for all the public health districts.
Further details, including each of the weighting steps for a particular survey, are described in more detail in the "2009 MIYHS Methodological Summary" found on the Reports and Fact Sheets Page. The handling of volunteer schools in the weighting of the grades 5/6 survey and school-level weighting is also described in the MIYHS Methodological Summary.
Weighted estimates and standard errors were calculated using SUDAAN (software). This special purpose computer application calculates estimates and standard errors for data from complex surveys. These values are based on the procedures that were used to select the sample. Technical details on the variance estimates can be found in the full reports. Sampling errors and variance estimates are not relevant within a school since all students in within each participating school were invited to participate for all four distinct survey modules (K/3, 5/6, MS, and HS).
The standard error is used in turn to calculate Confidence Intervals. For weighted data at the state-, public health district- or county-level, the percentage reported for each answer of interest is an estimate and, subsequently, some statistical error is involved. This is reflected in the Confidence Interval, which is bounded by the Lower Confidence Limit (LCL) and the Upper Confidence Limit (UCL). If all eligible students in the population answered a question, one can be 95% confident that the response would lie between the LCL and the UCL. If the confidence interval around a public health district or county's response does not overlap the confidence interval around the State of Maine's response, then the public health district or county response is considered to be significantly different from that of the state.
While many questions on the MIYHS offered a range of answers, all data reported on this website are based on a predetermined "answer of interest." One or more of the possible answers were selected for the reporting on this website and in the full report. In some cases, certain answers (including missing answers) were eliminated from the calculation. For example, students who indicated that they had not ridden a bicycle in the last year were excluded from the calculation of bike helmet use. For more detail, please see the 2009 MIYHS Methodological Summary (located on the Report and Fact Sheet Page) for these calculations. The percentages listed in the data are the percentage of students that answered in the way specified in the question title; for instance, 19.8% of Maine's high school students said they had consumed alcohol at least once in the 30 days prior to taking the MIYHS.
To protect student confidentiality, responses based on fewer than six (6) individuals will not be shown in any report. Individual reports were not generated for SAUs, schools or grades when fewer than 20 students participated in the MIYHS.
Copyright © 2009 All rights reserved.
