icr supporting statement-b

2

SUPPORTING STATEMENT

B. COLLECTIONS OF INFORMATION EMPLOYING STATISTICAL METHODS

1. Description of the potential respondent universe and any sampling selection methods to be used.

The survey of readers of MSPB reports on studies of Federal merit systems will be sent to a sample of readers randomly selected from our distribution lists. These lists contain several thousand readers grouped by their background. Separate lists are maintained for Federal agencies, members of Congress, representatives of state and local governments, academics, and the general public. We will use stratified random sampling procedures to select target respondents from each of these groups. When the populations are small (i.e. less than 100) we will survey all the members of a group. When the group populations are medium sized (i.e.. between 101 and 600). we will randomly sample half of the group members. When the populations are large (i.e.. over 600), we will randomly select 300 people to receive our surveys.

The survey of appellants will also use a stratified random sampling of the universe of adverse action appeals received during a given year. Typically there are about 8000 appeals received by the MSPB each year. Our plan is to survey the appellant, the appellant's representative (if the appellant had one), and the agency representative associated with the same appeal. Thus, the unit of analysis is the appeal, not randomly chosen parties. The sample will be stratified by appeal outcome (i.e.. settle, appellant won, appellant lost) and by type of appellant representation (i.e., self-representation, private attorney, union attorney or other). Appeals will be randomly selected from each of these stratified groups.

We also plan to do random electronic surveys of those who use e‑Appeal Online, the MSPB’s web-based electronic filing system. The purpose of this sampling will be to identify problem areas and determine ways to improve the system.

Based upon our previous customer satisfaction surveys of we expect a return rate of about 60 percent.

1. Description of procedures for collection of information.

The sample sizes discussed in the response to question 1 above were determined based on our historical response, so that the overall accuracy of our results would be + or - 3 percent (95 percent confidence interval) for each survey. The results were designed to be within + or - 5 percent (95 percent confidence interval) for each of the stratified groupings.

3. Description of methods to maximize response rates.

We will send each person receiving one of our customer satisfaction surveys a follow-up letter or email reminding them to complete and return our survey and providing them with a web site they can use to receive a new copy of the survey. Our primary way of dealing with non-response issues is by looking at the response rates for each of the stratified groups and, when needed, undertaking additional follow-up efforts to encourage greater participation.

3. Description of test procedures.

Since the customer satisfaction survey has been used successfully in the past, no further tests of the procedures will be undertaken.

5. Name and telephone number of individuals.

Person consulted on statistical design:

DeeAnn Batten (202) 653-6772, ext. 1411

Persons who will conduct and/or analyze the information:

James Tsugawa (202) 653-6772, ext. 1338

Cynthia Ferentinos (202) 653-6772, ext. 1334

Anne Marrelli, (202) 653-6772, ext. 1341