Explanation of Changes - 0960-0747

SUMMARY OF CHANGES TO THE AB BASELINE QUESTIONNAIRE

OMB No. 0960-0747

Below is a summary of the changes agreed on by SSA and OMB on 1/10/08. As per OMB’s instructions, we are submitting these as a change sheet. Please note that this change sheet satisfies Term of Clearance #2 (#1,3, and 4 were already addressed immediately after OMB’s original approval).

Section	Change	Rationale
A. Screener	For people who volunteer at the introduction that they have health insurance, train interviewers to collect the type of insurance held by the sample member, and allow interviewers to enter data in Section B (automatic screen out in Section A). We will attempt to convince sample members to complete Section B and would only allow interviewers to enter data for sample members when the only alternative would be a non-interview.	Ten people reported that they had insurance during the introduction. These people did not complete Section B (the insurance screener) and therefore were not counted in our estimate of insurance rates. Our suggested change will improve our estimate of insurance rates while minimizing burden on sample members.
B. Insurance Screener	Removed all wording appropriate for proxy interviews. Incorporated probe into the question (B1, B2). Deleted description of state Medicaid cards (B1). Moved information about plans that are excluded to the beginning of the question (B9). Added sample member’s name to first sentence of consent script (B19). Updated toll-free telephone number (B21). Revise consent (B20 and B24). We previously said, “Participating in this study will not in any way affect the amount of cash benefits you receive even if you return to work.” Revised to “Being in the study will not change any of the rules that determine whether you receive Disability Insurance cash benefits”. Revised B21 to incorporate language required by SSA. Revised <END1> so that sample members who screen out are not told that they are ineligible because they have insurance unless they specifically ask for a reason for ineligibility.	We are not allowing proxy respondents for the baseline interview. Improves question clarity and consistency. Cards change too frequently, and there are too many variations within a state for this to be useful. Make exclusions more salient and improve flow of question. Increases sample member’s attention. The revision improves the accuracy of the information we are providing to beneficiaries. Improves clarity about SSA’s role on the project, uses of data, and the duration in which SSA will obtain administrative data for sample members. Intended to reduce the likelihood that sample members would offer to drop their insurance coverage in order to get into the demonstration, and reduces the likelihood that interviewers will need to explain that in order to be eligible, sample members must be uninsured at the time we first speak to them.
C. Health and Functional Status	Made minor wording changes. Added valid values. Removed all wording appropriate for proxy interviews. Removed questions that ask respondents to self-report the main and secondary impairments that qualify them for SSDI (C4-C6). Previously asked about devices used for reading, hearing, and speaking in one question. Separated the question to ask separately about devices used for reading (C8, C8a). Revised to ask about limitations using public transportation and riding as a passenger in a car separately (C10f, C10g). Added SF-6D questions. Reordered questions when the SF-6D was added to improve overall flow of this section. Removed the PHQ-9.	We are not allowing proxy respondents for the baseline interview. Based on information from the NBS, we do not expect self-reported condition to match the impairment of record with SSA. Only the impairment of record will be used in our analyses. The question was double barreled. Some respondents who use devices for reading may not need a device for hearing or speaking. Respondents were confused about how to answer. The question was double barreled. Some respondents who had trouble with one transportation mode had no problem with the other and were confused about how to answer. OMB requested that these questions be added to the baseline because they will be included on the follow-up survey to inform the cost-utility analysis. There is some overlap between the PHQ-9 and SF-6D questions required by OMB. Respondents gave inconsistent answers across the two measures, complicating the analysis. Because OMB mandated the SF-6D, we retained these questions and dropped the PHQ-9, which is being administered periodically by CareGuide.
D. Use of Medical Services	Added skip logic. Added probes to improve question clarity. Added valid values and edit checks. Added reference date (D3). Revised range response options (D3a). Removed response options (D3e, D6, D6c) and added instruction for interviewer to enter verbatim response. Interviewers will code verbatim responses into code frame (original range response options) at the end of the interview after ending the call with the respondent. Revised wording of question D5 to ask about referrals separately from tests and x-rays. Revised wording of question D6b and related follow-up questions. Changed the reference period from a year to six months (D7a).	Improves sample member’s recall. Based on feedback from researchers. The list of response options was too long to code during the interview. Pretest respondents had difficulty answering a question that asked about visits (i.e., referrals) and tests in a single question. During pretesting, sample members were confused by the question wording about postponing or not getting recommended medical procedures. Respondents did not know how to answer if they hadn’t had the procedure yet because they were waiting to get an appointment. We revised the question to first ask if the respondent got the recommended procedures (we dropped the reference to postponement). Makes time frame consistent with other questions.
E. Employment History and Supports	Made minor wording changes. Added valid values. Added explanation of jobs to the question (E2). Revised time reference from months to year (E4a, E9a). Added alternate wording to accommodate multiple jobs (E4). Changed reference to last time worked for pay (E7, E7b, E8). Added question about usual number of hours worked per week (E10 or Ell). Deleted questions E11, E11a, E13 and E13a, which asked about employment over the past three years, in favor of the last time worked questions. The question was changed to determine the easiest way for respondents to provide earnings information (E14). The response options for earnings were changed so that wages are reported annually, rather than hourly (E16). Deleted the question about working in the next year, to ask only about the next two years. Switched the order of questions to ask knowledge questions first, then to ask about the information booklet. Incorporated the probe into the question (E21). Deleted questions (E25 and E26). The introduction to this series of questions at E21 was softened to emphasize that it is acceptable for the respondent to guess if he or she does not know the exact answer.	Improve question clarity and consistency. We believe that “years” is easier for respondents to focus on. Accommodates coding of multiple jobs. Pretest respondents found using “before you applied for SSDI” a confusing reference period, so these questions were reworded to ask respondents about the last time they worked for pay. Omission revealed during pretesting. The revision to the E7-E8 series described above made these questions unnecessary. This accommodates respondents by allowing them to answer using the basis that is easiest for them. The interviewers reported that this method is generally easier for respondents.  By increasing the number of ranges it allows respondents to give a more precise response. Respondents found the questions about one and two years redundant. Respondents will be asked only if they see themselves working in the next two years, which corresponds to the duration of the demonstration. Did not want to make respondents feel that if they read the booklet they should know the answers. Improves question clarity and consistency. Information available from administrative records. Interviewers reported that some respondents had trouble with the questions regarding rules for receiving SSDI. Many did not know the answers and were reluctant to guess.
F. Household Composition and Income	Added skip logic. Added valid values and edit checks. Added answer category for sample members who reside in group homes. Added reference period (F6). Revised annual salary ranges to above $100,000 (F7). Added assisted living as a housing option.	Improves coding accuracy. Narrows the reference period for the respondent. By increasing the ranges it allows respondents to give a more precise response. Improves coding accuracy.
G. Background	Added race and ethnicity questions (G3, G4).	Initially we were going to rely on SSA records for these data. However, these data are sometimes missing from administrative records. To gather complete information, we added race and ethnicity questions to the baseline survey.
H. Contact Information and Study Group Assignment	Added programmer checks. Added skip logic. Made minor wording changes. Deleted references to proxies. Added questions to collect cell phone numbers for respondent and contact persons (H2a, H2b). Revised assignment scripts to give sample members some information about the health benefit, services, and co-pays that apply to the AB Health Plan and AB Health Plan Plus groups (H17, H18). Added a question to determine if sample members understood their assignment status. Revised the control group assignment script to tell the control group that they will be notified of their assignment by mail (H19aa). Added coding choices at H20 from questions D3e, D4b, D5b, D6b and related follow-up questions to the end after the interviewer has terminated the call. Added code 20 (“was afraid”) as an answer choice.	Proxy respondents are not allowed for the baseline interview. Improves our ability to contact sample members for follow up surveys. Improves clarity of health plan benefits. Allows sample members to confirm that they understand the benefits and limitations of participation. We believed that the news of being assigned to the control group would be too disappointing and that some control group members would ask to withdraw from the study at this point. Mailing the assignment letter along with the incentive check is meant to limit attrition. These questions would take too much time to code during the interview. Interviewers will review their verbatim responses for these items and enter codes at the end after terminating the call. Feedback from pretest.

79