Introduction

The Phoenix Area Social Survey (PASS) is a longitudinal initiative funded by the National Science Foundation through the Central Arizona–Phoenix Long-Term Ecological Research (CAP LTER) program at Arizona State University. In 2006 and 2011, the PASS research team surveyed residential addresses in selected neighborhoods in the metropolitan Phoenix area to better understand perceptions, values, and behaviors regarding several key environmental issues, including water conservation, urban growth, air pollution, land conservation, biodiversity and urban climate change. To facilitate longitudinal analysis, PASS researchers have joined the 2006 and 2011 PASS datasets across questions common to both surveys. Since the PASS was originally designed to focus on neighborhood-level changes over time, the survey strategy followed residential addresses over time, not individual respondents. The PASS 2006-2011 joined dataset captures the 365 residential addresses that were surveyed during both years. A subset of this dataset is the 196 individuals who responded to the PASS in 2006 and 2011. Researchers interested in comparing responses from the same respondent over time can do so by using the Screen2006A question in the 2011 survey to create this data subset.

The Original 2006 and 2011 Sampling Designs

CAP LTER scientists have designed the PASS as a long-term socio-ecological monitoring survey to complement long-term ecological monitoring via Survey 200, which is conducted every five years at 204 points across the greater Phoenix area and surrounding undeveloped desert. When the PASS team developed the 2006 survey, only 94 of the 204 ecological monitoring points were in residential areas. The census block groups in which these points were located became the sampling frame for selecting the 40 neighborhoods (synonymous with census tracts) for the survey.

The neighborhood selection process involved classifying the 94 block groups on the basis of median household income (low, middle, and high) and urban geographic dimensions (core, fringe, and suburban). As there were too few neighborhoods that fit into two of the income/geography groups, these were combined with another group, resulting in seven groups in total: low income core, low income suburban, middle to high income core, middle income suburban, low to middle income fringe, high income suburban, and high income fringe. The PASS team added an eighth group, retirement communities, to reflect the geographic concentration of this demographic group in certain areas of the metropolitan area. They selected five neighborhoods from each group with the objective of creating a balanced sample to represent ethnic and racial composition, homeowners and renters, and the municipalities across the greater Phoenix area. In 2011 five new neighborhoods (also analogous to census block groups), were added to these 40 neighborhoods, reflecting geographic areas associated with other CAP LTER research initiatives.

The Institute for Social Science Research (ISSR) at ASU administered the surveys from April 29 to September 27, 2006 and May 26, 2011 to January 6, 2012 for PASS 2006 and PASS 2011 respectively. During both surveys, data were collected using a multi-modal survey approach. Respondents were given a choice of completing the survey online, on the telephone, or in person via an interviewer at their home in either English or Spanish.

ISSR survey staff drew an initial sample of 40 residential addresses in each neighborhood for the PASS 2006. They obtained a response rate of at least 50% in each neighborhood (total response rate of 51%) for a total of 808 survey completions. In 2011 ISSR staff included addresses for 757 of the 2006 respondents in the initial sample; they could not extract addresses for 43 of the 2006 respondents. They chose the remainder of addresses for the 2011 sample using a random selection function and used the same process for selecting replacement sample when encountering vacant properties or non-response. This was taken into account in calculating the response rate for PASS 2011, which was 43.36 percent (44.27% if partial responses are included in the calculation) and a total of 806 responses.

Prior to administering the survey in 2006, a doorknob hanger announcing that PASS was coming was placed on the doors. 2011 used a similar approach by sending a postcard in English and Spanish announcing "It's Coming." 2006 and 2011 followed the announcement with a letter explaining the project, how to participate, and their incentive options.1 In addition, a "fast facts" brochure, a $1 bill, and a PASS magnet were sent with the invitation to participate letter. This information was sent over a period of 3 weeks in 2006, while in 2011 it was sent over a period of about 6 weeks.

For both survey periods, phone numbers were then obtained using www.whitepages.com and a phone line dedicated to PASS was installed for households to leave information and set up interview appointments. All surveys could be administered in English or Spanish and were provided in multiple forms. A simple URL address was provided to complete the survey online, telephone interviewers called numbers and set up appointments. Interviewers called each number a maximum of 20 times unless respondent insisted s/he did not want to complete the survey. The sampled participants could request an in-person interview, and for those who didn't respond, interviewers went to houses that did not complete the survey—in 2006 for 2.5 months and in 2011 for 3 weeks. Letters and/or postcards were sent periodically to each address that had not completed the survey, asking them to take it and reminding them how to.

In 2006, the PASS team used 1,584 residential addresses to conduct the survey toward completion (initial sample addresses plus replacement sample addresses). The result was 808 completed surveys with a response rate of 51 percent. Additionally, sample usage was monitored by neighborhood. In every neighborhood, a minimum response rate of 50 percent was obtained. In 2011, 2,127 pieces of valid sample were used; 806 completed surveys were obtained resulting in a minimum response rate (RR1) of 43.4% percent and a maximum (RR2) of 44.3%.2

The Development of the Longitudinal Sampling

Of the 806 individuals that responded to the 2011 PASS, 365 (45%) had also responded to the 2006 PASS. 365 of the 2011 responses came from the same addresses as 2006 responses, while 441 new addresses responded in 2011. In 2006 a random "Respondent ID" was assigned to each house and was recorded in the 2011 data to connect each house with the "New 2011 Respondent ID." These Old and New ID's were used to join the 2006 and 2011 data to compare the responses over the two time periods.

For the longitudinal PASS dataset, it is important to recognize three types of respondents: the total sample for each time period, in addition to two subsets of resampled units, including parcels that were surveyed in both 2006 and 2011 (n=365) as well as individuals who responded in both time periods. See the "Working Report on PASS Data: Change across 2006 to 2011" report for additional details.