Laserfiche WebLink
what was found in the survey of voters for a particular question and what would have been found <br />if all of the estimated 78,658 likely November 2026 voters identified in the city had been sur- <br />veyed for the study. <br />Figure 17 provides a graphic plot of the maximum margin of error in this study. The maximum <br />margin of error for a dichotomous percentage result occurs when the answers are evenly split <br />such that SO% provide one response and SO% provide the alternative response. For this survey, <br />the maximum margin of error is ± 3.9% at the 95% confidence level. <br />FIGURE 17 MAXIMUM MARGIN OF ERROR DUE TO SAMPLING <br />8 <br />w <br />4- <br />0 <br />c <br />14% <br />------- <br />---- <br />---- --- <br />--------r--- <br />---- <br />12% <br />------- <br />----r-------- <br />----r---+------------ <br />--------*--- <br />------------- <br />10% <br />----------- <br />------------ <br />--- <br />------------ <br />-------- <br />--- <br />------------- <br />8% <br />----- <br />--1----L-------- <br />1----L---L------------J <br />---- <br />'---- L---- <br />---- '--------- ' <br />6% <br />4% <br />2% <br />0% <br />623 Respondents <br />+ 3.9% <br />F - r---, <br />0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 <br />Sample Size (Number of Respondents) <br />Within this report, figures and tables show how responses to certain questions varied by sub- <br />groups such as age, gender, and partisan affiliation. Figure 17 is thus useful for understanding <br />how the maximum margin of error for a percentage estimate will grow as the number of individ- <br />uals asked a question (or in a particular subgroup) shrinks. Because the margin of error grows <br />exponentially as the sample size decreases, the reader should use caution when generalizing <br />and interpreting the results for small subgroups. <br />The survey followed a mixed -method design that <br />employed multiple recruiting methods (telephone, text, and email) and multiple data collection <br />methods (telephone and online). Telephone interviews averaged 17 minutes in length and were <br />conducted during weekday evenings (5:30PM to 9PM) and on weekends (1 OAM to 5PM). It is stan- <br />dard practice not to call during the day on weekdays because most working adults are unavail- <br />able and thus calling during those hours would likely bias the sample. <br />Voters recruited via email and text were assigned a unique passcode to ensure that only voters <br />who received an invitation could access the online survey site, and that each voter could com- <br />6W 1�0_4140fa Ana 8 fir e North Research, Inc.-V _ W <br />