Every survey is different, however many projects share some fundamental elements.  This is a description of the methodology of many of our surveys, but of course there may be  changes to the applied methodology for each project according to its needs.

Our surveys can be fielded on any date specified by the client, and depending on the size of the survey calling usually takes 1-2 days.  Our clients may choose to have as many completed surveys in their project as they like.  Generally most clients choose 300, 350, or 400 "completes".  If 300 completed respondents are interviewed, the margin of error for the data collected is +/- 5.66 % for a 300 response survey for the data analyzed in its entirety with a 95% confidence level; meaning that we can be 95% sure that if every resident of the population sampled was actually surveyed, the results would vary by no more than 5.66% in either direction.  If 350 completed respondents are interviewed, the margin of error for the data collected is +/- 5.24 % for a 350 response survey for the data analyzed in its entirety with a 95% confidence level; meaning that we can be 95% sure that if every resident of the population sampled was actually surveyed, the results would vary by no more than 5.24% in either direction. If 400 completed respondents are interviewed, the margin of error for the data collected is +/- 4.9 % for a 400 response survey for the data analyzed in its entirety with a 95% confidence level; meaning that we can be 95% sure that if every resident of the population sampled was actually surveyed, the results would vary by no more than 4.9% in either direction.

Most of our calling lists are from a computer generated random list derived from a database of all households in the population to be surveyed with working telephone numbers.  This cuts down our incidence of bad numbers and non responses.  Essentially by using a list of good numbers, we can achieve greater accuracy than if we just used random digit dialing as other companies choose to do as a way to decrease their costs.  In political surveys, we can isolate phone numbers of households who have at least one member who has voted in at least one of the last two local elections or whatever criteria is appropriate for that particular project. If it is appropriate for the client to supply a list of names to us for calling, for example a membership survey, those names will be called in random order as well.  Each sample frame is carefully designed to accurately reflect the actual population percentages and in political surveys the information is available by voting district and precinct cross tabulations.

All respondents were screened to confirm they are over 18 and each household is attempted at least four times over a period of several days to account for non-response bias. Data is analyzed utilizing SPSS Quancept Software.  Interviews are conducted by trained staff with on-site supervision utilizing state of the art computer assisted telephone interviewing (CATI).  Random interviews were monitored to assure quality control measures are followed.  Dialing results are analyzed daily to assure statistically accurate distribution of call disposition in order to account for “non-response” variables.

If you would like more information on survey research or any of the other services we offer, please contact Hays Research Group.