Escolar Documentos
Profissional Documentos
Cultura Documentos
1
Practice: Get quality data into your evaluator’s hands
Key Action: Use technique to ensure valid and reliable data
Magnet schools are designed to produce both cognitive (improved learning in reading and
mathematics) and noncognitive outcomes (increased student and staff engagement with academic
content; decreased racial isolation). Commonly, some outcomes are measured through surveys, and
it’s important that such surveys are both valid and reliable, so that they yield credible information.
Validity: Instrument validity means that the survey or test measures what it says it’s measuring, and
not some other concept. For example, if a survey of students claims to measure their engagement in
school activities, then it should ask questions that address the extent to which they voluntarily commit
time and effort to school-related endeavors. The keys here are the words “voluntarily” and “effort.” For
example, we might ask how much time they devote to extracurricular activities (i.e., Do they spend
more time in school than is required?) or what books they read for pleasure (i.e., Is the content of the Type here
books related to the content of the curriculum?). We shouldn’t ask if they “like” school—we can like
something (“I like having the TV on when I’m home”) without being engaged with it (“the TV is a nice
background while I’m cooking dinner”).
Reliability: Instrument reliability means that the survey or test yields the same results on repeated
trials. For example, a thermometer should read 212 degrees Fahrenheit every time it is placed in boiling
water. Without research tools and procedures whose reliability is documented, it’s not possible to draw
credible conclusions from your evaluation or make data-based decisions about how to improve your
program.
1. Does the survey instrument include accompanying data on its validity and reliability?
2. If so, which of the following methods were used to establish these psychometric
properties?
a. Test-retest reliability: Administering the same test to the same groups multiple
times, with similar results
b. Factor analysis: Determining whether items that are conceptually linked show
similar linkages in the answers
c. Split-half reliability: Using two versions of questions and comparing the results