FAQ: Monitoring the Future Restricted-Use Geographic and Other Variables

Monitoring the Future: A Continuing Study of American Youth Restricted-Use Data

MTF Restricted-Use Geographic Variables

Q: Can my project focus on one or more states?

A: MTF is not designed for nor is it appropriate to generate state-level prevalence estimates of any outcome for small states like Colorado and Washington (including the outcome of marijuana prevalence). In addition to small sample size, the schools in these states were not drawn to be state-representative and they are not state-representative.

Use of MTF data to publish state-level prevalence estimates for small states will be prohibited for all researchers, including the MTF investigators (some of whom have participated in such work with MTF data in the past). Such estimates will not get through disclosure review.

Q: How can I use the MTF state-level identifiers to answer my research question?

A: MTF is not designed for nor is it appropriate to generate state-level prevalence estimates of any outcome for small states like Colorado and Washington (including the outcome of marijuana prevalence). In addition to small sample size, the schools in these states were not drawn to be state-representative and they are not state-representative.

Use of MTF data to publish state-level prevalence estimates for small states will be prohibited for all researchers, including the MTF investigators (some of whom have participated in such work with MTF data in the past). Such estimates will not get through disclosure review.

This being said, state and zip code information is suitable for aggregate analyses that include all or the vast majority of states. Analyses that use a specific state or a collection of small states to examine associations would most likely pass disclosure review. Examples that would likely pass disclosure review include:

  • An analysis that reported estimates such as "An xx% increase in a school's level of marijuana use is associated with a one unit increase in the concentration of dispensaries in that school's zip code area".
  • A correlation or regression showing that higher levels of state taxes on tobacco use are associated with lower prevalence of adolescent tobacco use. If an application for a specific question like these were submitted, the application reviewers will tell you at the start if it would pass disclosure review or not. If asked to comment on such a published analysis, MTF staff would say that the data were not drawn to be representative of states nor are they representative of states, and that analyses using only small states are essentially samples of convenience (which is OK for some analyses and some journals).

Q: What is the time period for "Zip Code of School"? (12th Grade Data: SDE_BY_GEO_V141; 8th/10th Grade data: SDE_BX_GEO_V533)

A: The Zip codes associated with the participating MTF schools will be accurate for the year in which the data were collected. No attempt has been made to update or change the school zip code for an MTF school when a change was later made by the US Postal Service.

Other MTF Restricted-Use Variables

Q: How was "Respondent's age in months" created? (12th Grade Data: SDE_BY_V9504; 8th/10th Grade data: SDE_BX_V9504)

A: This variable indicates the respondent's (Rs) age in months. For all 12th grade Rs and for the Rs in grades 8 and 10 beginning with the 1993* surveys, two questions were asked to determine an approximate age: "In what year were you born?" and "In what month were you born?" As the response options available for year of birth change every two years in the surveys, it is difficult to determine age without a lot of coding. To facilitate calculating (approximate) age in the survey year, a variable is created that represents "Rs age in months". This variable is derived from the year of administration variable, respondent's self-reported birth year and month variables, and the survey date of administration. (Note: if respondent information for birth month is missing, April is assumed.) If date of administration and birth month variables indicate the same month, it is not included in the total sum of months. The resulting number can then be used as a close approximation to the respondent's age in months at the time of the survey.

*In 1991 & 1992, 8th and 10th grade respondents were asked the single question, "What was your age on your last birthday?", with eight discrete response options ranging from ≤ 11 years old to ≥ 18 years old. Therefore, there are only eight discreet options for age in months for these two years, and the possible range of values for the "Age in months" variable is smaller than the range of values for 1993 forward.