B.2 Sample-waiting list

B.2: Procedures for the Collection of Information

The research team will facilitate community meetings at each of the 30 selected tribal areas to discuss housing issues and the role of IHBG in their community. While the community meetings are open to all community member, we will send specific written invitations to a sample of approximately 27 households who either live in units built or subsidized through the IHBG program or receive some type of housing assistance through the IHBG program and to approximately 27 households who are on the waiting list for IHBG assistance. To be consistent with the community culture and to ensure no one will stand out, all attendees to the meeting will be allowed to complete the community questionnaire. However, we will separately track the questionnaires completed by the sample households for the analysis of nationally representative data. We will have a registration for the meeting where we will have cover sheets with numbered questionnaires corresponding with the sign in number. After the meeting, we will separate the questionnaires completed by the targeted current beneficiary sample members, current waiting list sample members and other attendees. We will note which sample members from both beneficiary and waiting list groups did not attend or did not complete a questionnaire, then discard the cover sheets so that the completed questionnaires cannot be associated with a specific respondent.

The questionnaires will be self-administered, but the research staff will walk around the meeting room to answer questions. We will encourage respondents to complete the questionnaire prior to the meeting so that their perceptions are not influenced by the discussion. To this end, after some introductory remarks, the facilitator will set aside 10 minutes of the meeting for attendees to complete the questionnaires.

Once all the questionnaires for a given tribe are complete, they will be entered into a database, using a double entry system for quality control. In a double-entry system, two separate people enter the data and any inconsistencies are manually reviewed to ensure accuracy. There will be variables that indicate which tribe the response is from, whether or not the respondent was an invited sample member, and whether the questionnaire was completed at the community meeting or during follow-up efforts. The responses of the non-sample members will be reported separately in memo and will be used to see if they are consistent with the sample members or whether the people who attend the meeting without a specific invitation have different views than the random sample of invitees.

B.3: Methods to Maximize Response Rates and Deal with Nonresponse

We have designed the data collection plan to meet our goal of a response rate above 80 percent. First, we have designed a short instrument to minimize the burden of the respondent. Second, we are administering the survey at a community meeting rather than going door to door so that the data collection is not intrusive. Third, the meetings will be facilitated by research team members with cultural competency and expertise in Indian housing program. Fourth, the introduction and informed consent reassures potential respondents that we will not associate their name with any of the responses to protect their privacy.

In order to encourage sample members to attend the community meeting, we will send invitations to each sample member a few weeks before the meeting explaining the purpose of the meeting and the questionnaire and the value of receiving their input. Right before the meeting, we will send them a reminder card about the meeting. To further encourage attendance, we will hold a door-prize drawing at the meeting for a Pendleton blanket, a traditional part of Indian give-away activities.

We will also follow-up with sample members that do not attend the meeting. We will send them a copy of the questionnaire with a business reply envelope to return the completed questionnaire. For any sites where the response rate is less than 80 percent due to non-attendance, we will also make follow-up phone calls when that contact information is available.

In our analysis, non-response will be treated as missing at random within tribal areas. (We will not have any information on non-respondents to ascertain whether certain demographic groups or assistance types are not participating or to use in making the non-response adjustment.) That is, the non-response adjustment will be made at the tribal area level and will assume non-respondents are similar to respondents in the same tribal area for both IHBG beneficiary and waiting list populations. For example, if 22 of the 27 sample members complete the survey, the non-response adjustment will be to multiply the sampling weight by the adjustment factor 27/22.