Memo of Change

February 23, 2007

MEMORANDUM

To: Brian Harris-Kojetin

From: Gail Mulligan and Chris Chapman

Subject: Addendum to NHES:2007 OMB Clearance Request (1850-0768) 07-013

The purpose of this memo is to serve as an addendum to the NHES:2007 OMB Clearance Request (OMB number 1850-0768).

As we have discussed in recent weeks, the respondent cooperation rates and refusal conversion rates for NHES:2007 have been significantly lower than in past collections. The loss in response is most notable at the Screening level, where most nonresponse is observed in random digit dial (RDD) surveys. As of last week’s data collection report, the cooperation and conversion rates were as follows:

Cooperation Conversion

Instrument Rate Rate

Screener 37.6 30.8

RDD sample 37.0 30.5

Homeschooler sample 51.5 29.8

Address sample* 42.1 36.5

School Readiness 89.2 38.5

Parent and Family Involvement in Education 85.4 30.2

Adult Education 77.9 34.1

Nonparticipants 73.5 35.8

Participants 82.2 30.2

Note that the cooperation and conversion rates for the address sample are not directly comparable to those for the RDD sample. The address sample cases that are included in the telephone data collection are those for which a telephone match was found. Therefore, the address sample currently in telephone data collection is more comparable to RDD stratum containing mailable cases from low minority areas, which as a cooperation rate of 41 percent.

In order to increase yield for the extended surveys in this collection, we currently plan to release the reserve RDD sample described in the OMB clearance package that was submitted for the NHES:2007 data collection. Because we are seeing a higher initial cooperation rate among cases in the address sample, compared to all cases in the regular RDD sample, consideration was given to drawing a new reserve sample from address lists in order to improve cooperation rates. However, there are two important considerations that lead us to recommend against creating a new, address-based reserve sample. First, as noted above, the portion of the RDD sample for which addresses have been matched have cooperation rates that are equivalent to the address sample. The lower RDD rates are primarily due to the part of the sample for which addresses cannot be obtained. Second, we are unable to obtain telephone matches for the entire address sample (about one third have no phone matches) and, therefore, those we could add to the telephone collection would be only a nonrandom subset of all addresses.

For these reasons, we will release the RDD reserve sample, which has already been drawn and includes a total of 26,664 telephone numbers. Of these, 17,828 are potential residential numbers (the remainder were screened out as nonworking or business numbers). Within this callable group, 14,578 have matched mailing addresses. We propose to send the advance letter to all of the 14,578 mailable numbers, and to include a cash incentive of $5 in the mailing to a random half of the cases and a cash incentive of $10 in the remaining half. This will allow us to evaluate the impact of these alternative incentive amounts and to compare the results with the incentive of $2 that was sent to the main sample.

The reserve sample will receive the same treatment as wave 2 of the RDD sample, that is, these cases will not receive refusal conversion calls, and will receive a maximum of 14 calls. Ideally, we would test the effect of incentives in combination with full dialing protocols used for wave 1 cases. However, there would not be enough time to use the full contact protocol from wave 1 and still close collection by summer.

We will report the results of the split sample incentive experiment to OMB as soon as they are available.

Once we receive OMB approval, we will implement this approach. Please let us know if you have any questions or comments.

Page 2 of 2