Detailed Descriptions of Individual Changes

Key changes to Retirement statement survey and experiment

September 15, 2014

We used feedback from the focus groups that we conducted in July, 2014 to refine the survey instrument and sample statements. In most cases, changes were made to improve clarity of questions and/or response categories, and to improve survey flow.

Key changes by section are summarized below:

Section I. Survey eligibility

We shortened eligibility determination from two questions to one question.

Section II. DC plan activity

We expanded the question on DC account contribution from one to two questions to improve respondent clarity, as well as to improve data quality for analysis

Section III. Retirement Expectations

We adjusted the format of the retirement expectation questions from questions that ask about likelihood of working at set ages to questions that ask about most likely age of retirement.

Section IV. Retirement Goals

We added and refined response categories based on feedback from focus group discussions.

Section V. Retirement Account Statement

Questions were edited for clarity

Section VI. Experiment

Because focus group feedback on the additional information on the statements was universally positive, we adjusted the structure of the experiment to better focus on the question of whether the additional information changes hypothetical behavior and to simplify the design. We have 3 versions of the sample statement, A, B, and C. Statement A is Version 1 from the focus groups. It is the basic statement with account information, and projected account balance. Statements B and C are adapted from Versions 2 and 3 from the focus groups, taking the parts of the statements to which focus group respondents responded best. Statement B includes additional projection information—namely monthly retirement income, as well as longer description of the projections. Statement C is identical to Statement B, except for the addition of a page that shows how monthly retirement income varies with contribution level, retirement age, and rate of return.

In the previous experiment design, we had proposed partially individualizing the statements to the respondent’s personal experience. However, after reflection, we determined that the individualized statements would make empirical identification of the impact of the additional statement impossible. Instead, we allow for a low contribution condition (statements A1, B1, C1) and a high contribution condition (statements A2, B2, C2).

Section VII. Sample Statement Survey Questions

Questions and response categories were edited for clarity