Polling & Survey Sampling: Margin of Error, Weighting & Poll Averages
Margins of error with NORM.S.INV, sample sizes for a target margin, the finite population correction, proportional vs equal stratified allocation, post-stratification weights by age and area, Kish's design effect, a weighted poll average with a trend chart and the margin of a lead: 8 graded labs on one fictional city referendum.
The first lesson is free — no account needed.
About this course
A polling and survey sampling course that analyses the polls of one fictional city referendum in a spreadsheet, for students, journalists, market researchers and anyone who reads polls. You put a 95% margin of error with NORM.S.INV on every share and find the single margin printed under the headline, work out the interviews and budget for margins from ±5 to ±2 points, shrink the sample for a small ward with the finite population correction, split interviews over districts by proportional and equal allocation, weight age × area cells to the electorate with post-stratification and see the Yes share move, measure the design effect and effective sample size with Kish's formula, combine eight polls into an average weighted by size and recency with a trend chart, and test whether a lead is larger than its own margin of error. The first lab is free.
Lessons
- Margin of Error for a Poll Share (95 %, NORM.S.INV)Free
- Sample Size for a Target Margin of Error Pro
- Small Electorates: the Finite Population Correction Pro
- Stratified Sampling: Proportional vs Equal Allocation Pro
- Post-Stratification Weighting by Age and Region Pro
- Design Effect and Effective Sample Size (Kish) Pro
- Poll Averages: Weighting by Sample Size and Recency Pro
- Lead vs Margin of Error: the Difference of Two Shares in One Poll Pro