Attachment 4

ATTACHMENT 4

Incentive Experiments Report

Racial and Ethnic Approaches to Community Health Across the U.S. (REACH U.S.) Evaluation

OMB CONTROL NUMBER: 0920-0805

Report on Incentive Experiments

OMB instructed CDC and NORC to conduct experiments regarding the use of respondent incentives for the REACH U.S. evaluation and to submit a report of the results of the experiments (Notice of Office of Management and Budget Action, 02/25/2009). We report here the results of two experiments conducted during Year 1 of the REACH U.S. Risk Factor survey (May 2009 through November 2009).

We had three objectives in examining the use of incentives: (i) to compare the response rates in the experimental and control groups, (ii) to provide an estimate of potential incentive bias; and (iii) to assess the cost/benefit of using incentives for the study in general.

CATI refusal conversion Incentive experiment

Design

REACH U.S. conducted an incentive experiment with telephone respondents who initially refused to complete the interview. Cases were eligible for the experiment if they completed the household screener, met the REACH U.S. interview eligibility criteria for the community, and after being selected for the interview twice refused to complete the interview (e.g., refused by saying “not interested,” “don’t have time,” etc.). These respondents are extremely important to the success of the survey because they represent the target racial or ethnic population for a given community, and time and resources have been devoted to identify these individuals.

A random half of the sample in each REACH community was flagged to be eligible for an incentive (should they refuse twice in CATI) while the remaining sample served as a control group. In the experimental group, after the second CATI refusal, a refusal conversion letter that addressed the respondent’s concerns and $5 was mailed to the CATI respondent. The letter promised an additional $10 token of appreciation upon completing the interview. Control group refusals received a conversion letter only.

Results

Table 1 shows the interview completion rates in the experimental (incentive) and control groups for all communities combined. A greater percentage of respondents offered an incentive completed the interview than did respondents in the control group, but this increase was not significantly different.

Table 1. CATI Refusal Conversion by Incentive Category

Experiment Group	N (Households)	Interviews Pending at time of experiment	Interviews Completed	% Completed
No Incentive	1,300	1,851	167	9.02%
Incentive	1,339	1,888	192	10.16%

There were large variations in the impact of an incentive Refusal conversion rates among different communities (Table 2). However, the number of respondents was small for each community.

Table 2. CATI Refusal Conversion Rate by Incentive Category and by Community

Community	No Incentive	Incentive	Community	No Incentive	Incentive
1	0.0%	10.0%	15	8.9%	11.7%
2	10.7%	0.0%	16	10.2%	15.2%
3	1.8%	12.7%	17	0.0%	6.5%
4	4.7%	7.4%	18	13.7%	8.9%
5	5.9%	5.3%	19	16.5%	7.1%
6	4.0%	15.4%	20	12.5%	7.1%
7	11.0%	8.7%	21	7.7%	13.4%
8	10.6%	10.6%	22	6.7%	4.8%
9	9.3%	2.0%	23	5.7%	6.7%
10	8.9%	7.8%	24	10.1%	10.8%
11	9.9%	6.6%	25	11.1%	11.1%
12	10.3%	15.6%	26	7.5%	5.3%
13	15.7%	20.0%	27	5.9%	19.4%
14	8.2%	19.6%	28	9.3%	9.3%

We examined several demographic variables to ascertain whether offering an incentive resulted in potential bias. Table 3 shows the characteristics of the respondents in the refusal conversion experiment (incentive) and control groups that completed after the experiment. The incentive and no incentive groups did not significantly differ in terms of age, gender, household income, education level, home ownership, language spoken at home, and foreign born status.

Table 3. Demographic Characteristics of Respondents in CATI Refusal Conversion: Incentive versus Control

Variable	Variable Value	Incentive Categories		P-Value
		none	yes
Age category	18-39	19.2	14.6	0.4515
	40-64	35.3	34.9
	65 or older	45.5	50.5
Sex	Male	46.1	45.8	0.9585
	Female	53.9	54.2
Income	less than $15,000	21.6	19.5	0.8313
	$15,000 to less than $25,000	30.2	33.5
	$25,000 to less than $50,000	25.2	22.0
	$50,000 or more	23.0	25.0
Education level completed	Kindergarten or Less	1.2	0.0	0.4552
	Elementary	7.8	9.4
	High School	50.3	49.0
	College	40.7	41.7
Own or rent home	Own	65.3	65.9	0.9118
	Rent	34.7	34.1
Language spoken at home	English	82.0	84.9	0.4658
	Non-English	18.0	15.1
Born in the United States	Yes	76.7	82.8	0.1456
	No	23.4	17.2

Table 4 shows selected health characteristics of respondents in CATI Refusal Conversion in incentive and control group. The only significant difference between the two groups for the selected health variables analyzed was the reported number of days with poor mental health.

Table 4. Health Characteristics of Respondents: Incentives versus Control

Variable	Variable Value	Incentive Categories		P-Value
		none	yes
Number of days of poor physical health (during the past 30 days)	0	58.1	58.2	0.9916
	1 to 10	24.4	25.0
	11 to 20	5.6	4.9
	21 to 30	11.9	12.0
Number of days of poor mental health (during the past 30 days)	0	68.8	65.1	0.0356
	1 to 10	14.0	24.7
	11 to 20	5.7	4.3
	21 to 30	11.5	5.9
Health care coverage	Yes	84.6	85.1	0.8886
	No	15.4	4.9
Time elapsed since last routine checkup	Within Past Year	76.9	82.9	0.7252
	Within the past 2 years	8.1	6.4
	Within the past 5 years	6.3	4.3
	5 or more years ago	6.9	4.8
	Never	1.9	1.6
Moderate activities for 10 minutes	Yes	71.7	75.1	0.4629
	No	28.3	24.9
Smoking frequency	EVERYDAY	24.1	25.6	0.8781
	SOME DAYS	14.5	16.7
	NOT AT ALL	61.5	57.7

Discussion

The CATI refusal conversion of those known eligible is important because these cases are very precious and we have already determined eligibility and spent significant time and effort on them. It is worthwhile for us to do what we can to increase response at this stage rather than incur the cost of releasing additional sample. Furthermore, in some communities the sampling frame is small and additional sample may not be available. In general, the Year 1 experiment suggests that sending a conversion letter was just effective as an incentive (the conversion rate ranged from 9-10%). However, the Year 1 experiment also showed that there was large variation in the impact of an incentive on completion rates within different communities.

During Year 2 data collection, NORC initially mailed refusal conversion letters to all households that were known to be eligible for the member interview and had refused twice to complete the interview via telephone. These letters were sent without any monetary incentive and served as an attempt to convert initially hesitant respondents into completed cases. After NORC mailed the letter, an interviewer called the respondent to answer any questions and attempt to complete the interview. Toward the end of Year 2 data collection, NORC and CDC offered monetary incentives to increase interview completion rates in 10 selected communities and to obtain the requisite number of completed interviews. As with the Year 1 experiment, a conversion letter and $5 was sent to the respondent with a promise of an additional $10 upon completion. In contrast to Year 1, the Year 2 incentives resulted in a 18.9% conversion rate for pending interviews. Hence, the same approach was followed for Year 3 data collection – the Year 3 refusal conversion letters (without incentive) resulted in a 6% conversion rate. When NORC mailed refusal conversion letters and offered incentives to pending interviews, 29% of pending interviews converted.

Our experience with Year 2 and Year 3 data collection suggests that incentives are a valuable and useful tool for the purposes of completing the required number of interviews and converting initial refusals. A number of reasons may explain why incentives appear to be of greater utility in Years 2 and 3 than in Year 1. First, the types of sample released in Year 1 differed from that released in Years 2 and 3. Specifically, Year 1 was conducted with sampled addresses of unknown survey eligibility. In Years 2 and 3 the address-based sample was enhanced with demographic information to facilitate the sampling of addresses likely to meet the eligibility requirements of the survey. It may be that addresses for which additional information is available via commercial vendors are more amenable to completing a survey if an incentive is offered. Second, whereas the Year 1 experiment was conducted in all communities, there was substantial variability across communities in the effectiveness of the incentive. In Years 2 and 3, the incentives were offered to a subset of communities, many of which saw a marked increase in interview completion rates during the Year 1 incentive experiment. As a result, the overall conversion rates for Years 2 and 3 are higher because the project offered incentives in a targeted manner.

Self-Administered Questionnaire (SAQ) Incentive experiment

Design

The goal in REACH U.S. is to complete 900 interviews with eligible adults in the target community, and because of the increasing use of cell phones and decreasing ability to reach all households via landline telephone, mail surveys are necessary to supplement the phone in order to provide adequate coverage of the geography. However, mail surveys tend to suffer from relatively low response rates. Mail surveys require additional motivation and effort on the part of the respondent to comprehend, complete, and return the instrument. To achieve higher response rates from mailed SAQs, we proposed that it was important to explore the use of incentives for those asked to complete the survey via mail SAQ. The purpose of this experiment was to establish whether incentives increase response to the SAQ mailing; and whether the cost of the increase (if an increase is obtained) can be justified on cost/benefit grounds.

Half the sample was randomly flagged to receive an incentive while the remaining sample served as a control group. Those in the experimental treatment group were mailed an SAQ along with a $5 incentive; within the experimental group half of the households also received a promise of an additional $10 upon return of the completed SAQ. The token of appreciation was given to the REACH U.S. household and not individual respondents to avoid encouraging the completion of questionnaires for ineligible respondents.

Results

Table 5 provides the results of the SAQ incentive experiment for all communities combined. Overall, more than 20 percent of households that received a SAQ packet responded to the mailing by returning at least one completed SAQ booklet. Offering an incentive significantly increased response to the SAQ mailing, with nearly 30 percent of households offered an incentive responding to the mailing compared to 11 percent of households in the control condition (χ² (1) = 2,148.65, p< .0001). Moreover, among the households offered an incentive, a promise of an additional $10 upon completion significantly increased response to the mailing (χ² (1) = 38.16, p< .0001); approximately 32 percent of the households promised an additional $10 responded to the mailing compared to 28 percent of those provided with $5 only.

Table 5. SAQ Responses in Incentive Experiment: Incentive versus Group

	Control	Experimental
Categories	No Incentive Offered	$5 Initial only	$5 Initial & $10 Thank You	Total Across Experimental Conditions	Overall
Households Mailed	21,490	10,704	10,769	21,473	42,963
Undeliverable Addresses	1,706	878	836	1,714	3,420
Households with at least one completed SAQ	2,178	2,729	3,158	5,887	8,065
Percent Households Responding	11.0%	27.8%	31.8%	29.8%	20.4%

There were large variations in the response to the SAQ mailing among communities, but all communities showed a strong positive response to the incentive. Differences between the control and experimental conditions ranged from 12 percentage points to over 20 percentage points. Table 6.1 through Table 6.28 at the end of this report displays the individual community tables of SAQ incentive experiment results.

Tables 7 shows several demographic and health characteristics of the SAQ respondents in the experimental (incentive) and control groups. Respondents differ among the experimental and control groups on several demographic variables. For example, respondents given incentives are more likely to be young, low income, and renters than respondents given no incentive. No significant difference was found between the two groups in gender, language spoken at home, and whether they were born in the U.S. In comparing the $5 only incentive group to the no incentive group, the $5 incentive group respondents are more likely to be renter and are less likely to be born in the U.S., than the no incentive group respondents. Note that tests were also conducted to compare the $5+$10 incentive respondents to the $5 and no incentive respondents. In most cases, the $5+$10 respondents were very similar to the $5 respondents, and different from the no incentive respondents in the same ways that the $5 respondents are.

Table 7. Demographic Characteristics of SAQ Respondents: Incentives versus Control.

Variable	Variable Value	Incentive Categories			P value
		None	$5	$5+$10	Incentive vs. None	$5 vs. None
Age Category	18-39	25.1	27.8	34.8	<.0001	0.063
	40-64	54.5	52.6	53.7
	65 or older	20.5	19.7	11.6
Sex	Male	44.9	46.0	42.3	0.4521	0.452
	Female	55.1	54.1	57.8
Income	< $15,000	32.3	34.9	37.8	0.0333	0.116
	$15,000 to < $25,000	20.1	16.5	18.4
	$25,000 to < $50,000	24.8	26.3	22.8
	$50,000 or more	22.8	22.4	21.0
Education level completed	Kindergarten or Less	0.7	0.0	0.3	0.0051	0.483
	Elementary	3.3	4.4	4.4
	High School	43.2	42.3	47.2
	College	52.8	53.3	48.0
Own or rent home	Own	52.5	46.7	41.9	<.0001	0.007
	Rent	47.5	53.3	58.1
Language spoken at home	English	17.5	18.9	12.1	0.2984	0.114
	Non-English	82.5	81.1	87.9
Born in the United States	Yes	82.8	78.4	85.4	0.2072	0.007
	No	17.2	21.6	14.7

For the health characteristic variables, the $5 incentive group behaves similarly to the $5+$10 group in comparison to the control (Table 8). The incentive groups are less likely to have any days of poor physical health or mental health than the control group and are less likely to have seen a doctor within the last year; they are more likely to smoke every day. However, the $5 incentive group is no different from the control in regards to smoking behavior.

Table 8. Selected Health Characteristics of SAQ Respondents: Incentive versus Control

Variable	Variable Value	Incentive Categories			P value
		None	$5	$5+$10	No Incentive vs Incentive	$5 vs none
# of days of poor physical health (during the past 30 days)	0	45.2	53.3	52.3	0.033	0.022
	1 to 10	33.0	29.0	29.1
	11 to 20	9.9	7.9	7.7
	21 to 30	11.9	9.8	10.9
# of days of poor mental health (during the past 30 days)	0	43.9	45.2	47.9	0.031	0.008
	1 to 10	33.3	33.7	29.6
	11 to 20	9.2	13.0	12.4
	21 to 30	13.5	8.1	10.1
Health care coverage	Yes	79.2	76.7	73.3	0.099	0.904
	No	20.8	23.3	26.8
Time elapsed since last routine checkup	Within past year	69.3	64.1	61.2	0.034	0.043
	Within past 2 years	13.5	15.7	16.0
	Within past 5 years	6.3	10.1	9.4
	5 or more years ago	9.2	9.6	10.4
	Never	1.7	0.5	3.1
Moderate activities for 10 minutes	Yes	76.2	78.6	75.8	0.537	0.985
	No	23.8	21.4	24.2
Smoking frequency	EVERYDAY	32.3	32.9	40.6	0.043	0.704
	SOME DAYS	14.2	16.7	16.5
	NOT AT ALL	53.5	50.4	42.9

We examined the costs associated with the SAQ incentive effort, and the final cost per completed case. The costs by experimental condition are shown in Table 9.

Table 9. Costs per Completed SAQ: Incentives versus Control

	Control	Experimental
Categories	No Incentive Offered	$5 Initial only	$5 Initial & $10 Thank You
Cost	$96	$106	$131

Although the control condition SAQs were less expensive than the two experimental conditions (whose costs include incentives), it also garnered the lowest response. The $5 plus an additional $10 condition was the most expensive per case. In fact, the response in this condition would need to be nearly 45 percent in order for it to become as cost effective as the $5 only condition, and raising the response rate from nearly 32 to 45 percent is unlikely without additional follow-up mailings (which would incur additional costs). However, not including an incentive at all, while the least expensive per-unit cost up front, would incur additional costs in the form of extra sample release needed in order to obtain the target completes. Because the incentives increase response to the mailing significantly, they reduce the overall sample needed. (Releasing additional sample to compensate for fewer completes among SAQs mailed without an incentive would increase survey costs as the additional sample would be mailed letters and contacted by interviewers.)

Discussion

Including $5 in SAQ mailing resulted in much higher response rates when compared to a control condition and was also cost effective. There were some significant demographic and health differences between the respondents from the incentive group and from the control. However, the $5 and $5+$10 were more similar to each other, and thus there would be no real advantage in terms of bias to incurring the cost of the additional $10. After the Year 1 experience, we included $5 in all SAQ mailings.

Table 6.1. Results of SAQ Incentive Experiment by Condition: Community 1

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	254	117	119	236	490
	Undeliverable Addresses	32	10	13	23	55
	Households with at least one completed SAQ	18	19	25	44	62
	Percent Households Responding	7.1%	16.2%	21.0%	18.6%	12.7%

Table 6.2. Results of SAQ Incentive Experiment by Condition: Community 2

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	262	127	137	264	526
	Undeliverable Addresses	41	13	18	31	72
	Households with at least one completed SAQ	14	21	21	42	56
	Percent Households Responding	5.3%	16.5%	15.3%	15.9%	10.6%

Table 6.3. Results of SAQ Incentive Experiment by Condition: Community 3

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	987	512	501	1,013	2,000
	Undeliverable Addresses	35	29	27	56	91
	Households with at least one completed SAQ	63	93	104	197	260
	Percent Households Responding	6.4%	18.2%	20.8%	19.4%	13.0%

Table 6.4. Results of SAQ Incentive Experiment by Condition: Community 4

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	1,719	869	847	1,716	3,435
	Undeliverable Addresses	161	74	68	142	303
	Households with at least one completed SAQ	205	235	303	538	743
	Percent Households Responding	11.9%	27.0%	35.8%	31.4%	21.6%

Table 6.5. Results of SAQ Incentive Experiment by Condition: Community 5

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	754	390	364	754	1,508
	Undeliverable Addresses	15	9	12	21	36
	Households with at least one completed SAQ	90	112	116	228	318
	Percent Households Responding	11.9%	28.7%	31.9%	30.2%	21.1%

Table 6.6. Results of SAQ Incentive Experiment by Condition: Community 6

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	218	130	110	240	458
	Undeliverable Addresses	16	13	13	26	42
	Households with at least one completed SAQ	14	26	32	58	72
	Percent Households Responding	6.4%	20.0%	29.1%	24.2%	15.7%

Table 6.7. Results of SAQ Incentive Experiment by Condition: Community 7

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	502	261	273	534	1,036
	Undeliverable Addresses	55	29	24	53	108
	Households with at least one completed SAQ	50	71	86	157	207
	Percent Households Responding	10.0%	27.2%	31.5%	29.4%	20.0%

Table 6.8. Results of SAQ Incentive Experiment by Condition: Community 8

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	523	255	249	504	1,027
	Undeliverable Addresses	45	7	9	16	61
	Households with at least one completed SAQ	45	54	67	121	166
	Percent Households Responding	8.6%	21.2%	26.9%	24.0%	16.2%

Table 6.9. Results of SAQ Incentive Experiment by Condition: Community 9

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	713	349	373	722	1,435
	Undeliverable Addresses	25	16	15	31	56
	Households with at least one completed SAQ	79	83	109	192	271
	Percent Households Responding	11.1%	23.8%	29.2%	26.6%	18.9%

Table 6.10. Results of SAQ Incentive Experiment by Condition: Community 10

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	413	198	189	387	800
	Undeliverable Addresses	75	30	46	76	151
	Households with at least one completed SAQ	24	38	56	94	118
	Percent Households Responding	5.8%	19.2%	29.6%	24.3%	14.8%

Table 6.11. Results of SAQ Incentive Experiment by Condition: Community 11

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	1,157	614	638	1,252	2,409
	Undeliverable Addresses	98	59	52	111	209
	Households with at least one completed SAQ	138	186	190	376	514
	Percent Households Responding	11.9%	30.3%	29.8%	30.0%	21.3%

Table 6.12. Results of SAQ Incentive Experiment by Condition: Community 12

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	1,600	797	811	1,608	3,208
	Undeliverable Addresses	263	135	144	279	542
	Households with at least one completed SAQ	209	209	272	481	690
	Percent Households Responding	13.1%	26.2%	33.5%	29.9%	21.5%

Table 6.13. Results of SAQ Incentive Experiment by Condition: Community 13

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	719	333	373	706	1,425
	Undeliverable Addresses	50	23	17	40	90
	Households with at least one completed SAQ	97	86	109	195	292
	Percent Households Responding	13.5%	25.8%	29.2%	27.6%	20.5%

Table 6.14. Results of SAQ Incentive Experiment by Condition: Community 14

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	593	286	287	573	1,166
	Undeliverable Addresses	72	34	29	63	135
	Households with at least one completed SAQ	64	67	92	159	223
	Percent Households Responding	10.8%	23.4%	32.1%	27.7%	19.1%

Table 6.15. Results of SAQ Incentive Experiment by Condition: Community 15

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	255	126	132	258	513
	Undeliverable Addresses	25	11	9	20	45
	Households with at least one completed SAQ	20	22	32	54	74
	Percent Households Responding	7.8%	17.5%	24.2%	20.9%	14.4%

Table 6.16. Results of SAQ Incentive Experiment by Condition: Community 16

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	2,651	1,272	1,311	2,583	5,234
	Undeliverable Addresses	286	124	157	281	567
	Households with at least one completed SAQ	284	370	439	809	1,093
	Percent Households Responding	10.7%	29.1%	33.5%	31.3%	20.9%

Table 6.17. Results of SAQ Incentive Experiment by Condition: Community 17

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	1,794	915	914	1,829	3,623
	Undeliverable Addresses	82	71	48	119	201
	Households with at least one completed SAQ	147	199	252	451	598
	Percent Households Responding	8.2%	21.7%	27.6%	24.7%	16.5%

Table 6.18. Results of SAQ Incentive Experiment by Condition: Community 18

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	504	273	249	522	1,026
	Undeliverable Addresses	17	8	21	29	46
	Households with at least one completed SAQ	45	46	65	111	156
	Percent Households Responding	8.9%	16.8%	26.1%	21.3%	15.2%

Table 6.19. Results of SAQ Incentive Experiment by Condition: Community 19

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	744	338	337	675	1,419
	Undeliverable Addresses	73	36	35	71	144
	Households with at least one completed SAQ	68	91	120	211	279
	Percent Households Responding	9.1%	26.9%	35.6%	31.3%	19.7%

Table 6.20. Results of SAQ Incentive Experiment by Condition: Community 20

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	292	141	133	274	566
	Undeliverable Addresses	6	7	2	9	15
	Households with at least one completed SAQ	31	29	37	66	97
	Percent Households Responding	10.6%	20.6%	27.8%	24.1%	17.1%

Table 6.21. Results of SAQ Incentive Experiment by Condition: Community 21

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	646	333	322	655	1,301
	Undeliverable Addresses	144	64	48	112	256
	Households with at least one completed SAQ	55	89	98	187	242
	Percent Households Responding	8.5%	26.7%	30.4%	28.5%	18.6%

Table 6.22. Results of SAQ Incentive Experiment by Condition: Community 22

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	1,146	559	562	1,121	2,267
	Undeliverable Addresses	38	16	11	27	65
	Households with at least one completed SAQ	122	158	175	333	455
	Percent Households Responding	10.6%	28.3%	31.1%	29.7%	20.1%

Table 6.23. Results of SAQ Incentive Experiment by Condition: Community 23

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	208	87	100	187	395
	Undeliverable Addresses	13	4	5	9	22
	Households with at least one completed SAQ	6	12	22	34	40
	Percent Households Responding	2.9%	13.8%	22.0%	18.2%	10.1%

Table 6.24. Results of SAQ Incentive Experiment by Condition: Community 24

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	602	298	303	601	1,203
	Undeliverable Addresses	71	37	32	69	140
	Households with at least one completed SAQ	92	80	91	171	263
	Percent Households Responding	15.3%	26.8%	30.0%	28.5%	21.9%

Table 6.25. Results of SAQ Incentive Experiment by Condition: Community 25

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	157	72	77	149	306
	Undeliverable Addresses	17	4	6	10	27
	Households with at least one completed SAQ	23	21	20	41	64
	Percent Households Responding	14.6%	29.2%	26.0%	27.5%	20.9%

Table 6.26. Results of SAQ Incentive Experiment by Condition: Community 26

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	581	289	306	595	1,176
	Undeliverable Addresses	28	15	17	32	60
	Households with at least one completed SAQ	37	48	72	120	157
	Percent Households Responding	6.4%	16.6%	23.5%	20.2%	13.4%

Table 6.27. Results of SAQ Incentive Experiment by Condition: Community 27

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	735	370	381	751	1,486
	Undeliverable Addresses	83	27	38	65	148
	Households with at least one completed SAQ	89	116	121	237	326
	Percent Households Responding	12.1%	31.4%	31.8%	31.6%	21.9%

Table 6.28. Results of SAQ Incentive Experiment by Condition: Community 28

		Control	Experimental
	Categories	No Incentive Offered	$5 Initial only		Total Across Experimental Conditions	Overall
				$5 Initial & $10 Thank You
Total	Households Mailed	749	385	368	753	1,502
	Undeliverable Addresses	21	15	10	25	46
	Households with at least one completed SAQ	57	67	87	154	211
	Percent Households Responding	7.6%	17.4%	23.6%	20.5%	14.0%

2