Meeting Docs - Laserfiche WebLink

Home Browse Search

CenterFuse Broadband Feasibility Report For a survey to produce valid results the calling must be conducted randomly, meaning that the calling shouldn't be clustered around any one particular portion of the study universe. For example, the survey would not be considered to be valid if all of the calls were placed only to one portion of the city. We used the list of telephone numbers in a way to make sure we got random results. We called every fifth phone number on the list and cycled through the calling list until we got the desired number of completed surveys. Most business and political surveys strive to achieve an accuracy of about 95%with results that are plus or minus 5%. In layman's terms, this means that the results of such a survey are reliably accurate (the 95% number) and you would expect to get the same results (within 5%) if you could ask the same questions to everybody in the city. CCG uses an online survey tool to determine the number of surveys needed to achieve the desired accuracy. The tool is provided by Creative Research Systems and is found online at https://www.surveysystem.eomisscalc.htm. We've used this tool for many years and have manually done the mathematical calculations that demonstrate that the tool is accurate. This tool told us that we needed 367 completed surveys in Ellensburg to achieve the desired accuracy of 95%, plus or minus 5%. In the US we know that many people distrust the results of surveys, mostly due to results obtained from political surveys. This speaks to the issue of bias. When callers are asked about sensitive topics like politics, religion, or anything personal or controversial, it's well-known that many respondents don't answer questions honestly to a stranger like a survey taker. The best example of this is when surveyors ask people for their household income. Survey companies have often said that as many as half of residential homeowners do not give an accurate response to the salary question. However, experience shows that there is high reliability with surveys that look at non-emotional topics, such as this survey talking about a routine product purchased by most households. We have anecdotal evidence that broadband surveys are good market predictors because CCG has been giving these surveys for 20 years and we've had many opportunities to see the broadband penetration rates in communities that have built networks to compare achieved customers to the predictions made by our surveys. Surveys are never 100%accurate because sometimes an ISP does something to change the public perception. For example, an ISP that has problems during a network launch might underperform a prediction made by a survey. In general, we've learned to have faith in the predictions made by these broadband surveys. If there is an issue that might slant your survey, it's that it was administered during the pandemic. It's possible that more people will say they need better broadband now and that response could possibly change after the pandemic has passed. Because of this, and in general, we always recommend being conservative in using the results of a survey. Survey Results The survey produced some interesting results. A full copy of the survey questions and the responses are included in Addendum I of this report. Here are highlights of the survey results: Page 20