Methodology

The 2019 PDK survey is based on a random national sample of 2,389 adults and an oversample to 556 public school teachers and 1,000 parents of school-age children.

The 2019 PDK Poll of the Public’s Attitudes Toward the Public Schools was designed, managed, and analyzed, and this report written by Langer Research Associates of New York, N.Y. The poll was conducted using the nationally representative Ipsos KnowledgePanel®, in which participants are randomly recruited via address-based sampling to participate in survey research projects by responding to questionnaires online. Households without internet connections are provided with a web-enabled device and free internet service.

Langer Research Associates is a charter member of the Transparency Initiative of the American Association for Public Opinion Research. After six months, researchers will have access to the complete dataset through the Roper Center for Public Opinion Research.

The survey was designed to include about 1,000 adults in the general population and, through oversampling, 1,000 parents of students in grades K-12, including 100 Asian-American parents, and 500 public school K-12 teachers. The full sample was weighted to reflect the correct proportions of these oversampled populations.

The questionnaire was pretested April 5-8, 2019, and field work was conducted in English and Spanish April 12-27, 2019. After initial invitations, reminder emails were sent on the third, seventh, 11th, and 15th days of the field period. Out of 5,323 panel members invited to participate, completed, qualified surveys were provided by 2,425. Participants completed the survey in a median time of 14 minutes.

For quality control, the following were flagged for possible inattention:

  • The fastest 2% of respondents in total completion time per survey path (parent, teacher, parent and teacher, or neither). Forty-eight were flagged.
  • Those who skipped more than one-quarter of all survey items. Twenty-seven were flagged.
  • Those who entered the same response items in four or more of six grid questions (Q8_parent, Q8_teacher, Q15, Q26, Q33, and Q36). One hundred twenty-five were flagged.

Respondents received 1 point for each of these measures. No points were assigned to 2,256 respondents, 1 point to 146, 2 points to 15, and 3 points to eight. Those flagged on two or three measures were removed from the dataset (23 cases).

In addition, respondents who completed the survey in less than one-quarter of the median time (eight cases) and the fastest 2% in completing the lengthiest questions (Q13, Q24, Q31, and Q33) (five cases) were removed. In total, 36 respondents were removed in the quality-control process.

The sample composition after quality control was 1,913 general population adults, including 1,083 K-12 parents (698 White, 95 Black, 153 Latinx, 103 Asian-American, and 34 mixed-race parents), and 556 K-12 public school teachers. The total sample size of 2,389 includes oversamples of 891 K-12 parents, including 64 Asian parents, and 471 K-12 public school teachers.

Data were weighted via iterative proportional fitting to the following benchmark distributions of general population adults from the U.S. Census Bureau’s March 2018 Current Population Survey Supplement and 2017 American Community Survey:

  • Gender (male, female) by age (18-29, 30-44, 45-59, 60+)
  • Race/ethnicity (White, Black, other, Latinx, 2+ races)
  • Census region (Northeast, Midwest, South, West) by metropolitan status (metro, non-metro)
  • Education (less than high school, high school, some college, bachelor’s degree, or higher)
  • Household income ($0-$24,999, $25K-$49,999, $50K-$74,999, $75K-$99,999, $100K-$149,999, $150K+)
  • Asian-American parents with children 6-18, non-Asian-American parents with children 6-18, nonparent
  • Marital status (married, not married)
  • Language proficiency (English-proficient Hispanic, bilingual Hispanic, Spanish-proficient Hispanic, non-Hispanic)
  • Hispanic nativity (U.S.-born Hispanic, not U.S.-born Hispanic, non-Hispanic)

Within the public school K-12 teacher sample, data were weighted to the same categories, with these differences: language proficiency and Asian parents were not used; other race and 2+ races were netted; education was netted as college degree/no degree, and gender and age were weighted separately, as were region and metro status. The teacher data also were weighted to the following distributions from the 2015-2016 National Teacher and Principal Survey:

  • School type (traditional public, charter)
  • Grade kind (primary, middle, high, combined)
  • Teacher tenure (fewer than 4 years of teaching, 4-9, 10-14, 15+)
  • School size (fewer than 100 students, 100-199, 200-499, 500-749, 750-999, 1,000+).

For K-12 public school teachers, weights were trimmed at 1.8% and 98.92% of their distribution (minimum 0.054, maximum 6.419). The survey has a design effect due to weighting of 2.20 for this population, for a margin of sampling error of plus or minus 6.2 percentage points.

For the general population, weights were trimmed at 2% and 98% of their distribution (minimum 0.152, maximum 3.529). Given oversampling, the survey has a design effect due to weighting of 1.87, for a margin of sampling error of plus or minus 3.1 percentage points for the full sample. Error margins are larger for subgroups.

The survey included four open-ended questions — Q1, Q5, Q48, and Q50. Responses were independently coded by two researchers using the same code book; codes were then reconciled.