Upload
abigail-carr
View
216
Download
3
Tags:
Embed Size (px)
Citation preview
Effectiveness of Monetary Incentives and Other Stimuli Across Establishment Survey
Populations
ICES III 2007
Montreal, Quebec Canada 4.30.07
Danna Moore and Mike Ollinger
(1) Why is this study important?
(2) Gaps in Establishment Survey Literature
(3) Provide suggestions for implementing establishment surveys
Objective
Establishment Respondents: What type of respondent are they?
Household Respondente.g. small farmer, small business owner
Large business or Org Respondent Multiple locations Gate keepers Record system
Can vary along the continuum
Why is this important?
Questionnaire variation Converting Mandatory reporting R’s Vary over time Risk/Difficulty -- cash incentives Population based sample sizes Experimental design Cost/Benefit Generalize
Stimuli Tested
• Cash Incentives
• Cash plus special postage/package
• Multiple modes
• Mode sequencing
• Mode preferences
• Visual Design— color background vs. none• Answer boxes stand out
Some Answers Towards the Big Question
Do incentives help or hinder in obtaining survey responses from businesses?
• Crosses types of establishments and industries.
• Experiment based • Population based• Random assignment
Understanding Why People ParticipateSeveral theories (Dillman, 1978; Gouldner, 1960; Biner and Kidd, 1994; Groves
et al., 1999).• Social exchange
Leverage Saliency Theory—Groves et al. (1999)Decision to participate is a series of interactive additive
factors.
Some are survey specific and others are person specific.
Incentives are viewed as an inducement used to compensate for absence of some factors (i.e., saliency or sense of duty).
1999 POQ vol 64
Leverage Saliency TheoryOf Survey Participation
Features of Establishment Surveys That Often Lead to Survey Errors
1. One person selected to
represent
establishment.
2. Burden increases as
they answer as a
representative.
3. There is a respondent
questionnaire
interaction.
4.Respondent’s characteristics in relation to the establishment influences their ability to respond.
5. intermediary between the questionnaire and the characteristics of the record system.
6. Organizational environment
7. Extenuating survey situation
• Businesses differ across industries by size, structure, and organizational environment.
• Each survey may have a situation or
circumstance that impedes contact.
• These differences and circumstances often
influence how well a survey request can
penetrate an establishment
Factors Influencing Response
Gaps
• Lack of monetary incentive studies:
Few experimental treatments for definitive comparisons and outcomes.
Few comparisons of cash versus “cash like” (e.g. checks, ATM cards) studies—could be especially important for government.
No empirical demonstrations of effectiveness of incentives across types of industries and firm size.
Few comparing effects of various size incentives. Not much on effectiveness of survey modes and mode
sequencing Response attributable to survey mode sequencing.
2001 HMO Survey Physicians N=1474
43%
57% 63% 66% 64%
80%70%
0%
20%
40%
60%
80%
100%
1Control T1: FC, FC T2:Priority No, FC $10 T3: Priority $10, FC T4; Fc, Priority $10T5 Priority $10,FC T6: Priority $10 2x 2004 Priority $10 2x
***
**
2003 USDA Nationwide Meat Manufacturers N=1,705 Response rates achieved by experimental treatment group
44%
23%
47% 45%
56%
67%70%
57%
0%
10%
20%
30%
40%
50%
60%
70%
80%
Pilot-- Redmeat Mf.
Pilot--PoultryMf.
Full-- RedmeatMf
Full Poultry Mf
Control: 2x FC No $ Trt 2: 2x priority Trt 3: 2x Priority $5
******
2006 Snake River Grain Warehouse and Shipper Survey Response by Experimental Group (n=424 elevators)
58%
76%
64%63%
49%
10%
57%
72%
43%
0%
10%
20%
30%
40%
50%
60%
70%
80%
Graincompanies
Port Shippers Grain Elevators
Control Trt 1: 2X FC No $ Trt 2: FC $5 , FC No $ Trt 3: 2x FC $5
Chi SQ 6.7P < .01
***
N.S.
N.S.
2007 Evaluation of WA Plastic Pesticide Container Use/Recycling Control versus Treatment, N=1,986
35%
52%
22%
42%
0%
10%
20%
30%
40%
50%
60%
CommercialApplicators/Dealers Licensed
for Pesticides
Licensed Producers
Control Trt1: No incentive, FC Trt 2: $5 incentive, Priority
Chi Sq. 15.86***P<.001
Chi Sq. 33.5 ***P< .0001
2006 Oregon Business Environmental Management Survey Type and Experimental Treatment Group Initial sample n=1964
3%
15%
51% 51% 52%47%
28% 25%
8%
18%19%
33%
23% 24%
46% 42% 43%40%
42%41%
0%
10%
20%
30%
40%
50%
60%
Construction: n=394
Manufacturing n=752
Transportation n=343
Accomodation n=475
All Response
Ctrl T1: FC 2x Trt 2: Priority No $ 2x Trt 3: Priority $5, FC Trt 4: Priority $5 2x
***
**
** *** ***
**
***
34%
53%
0%
10%
20%
30%
40%
50%
60%
Percentage
1Control t1: FC 2x No Incentive Ttrt 2: FC $5 2x
2006 Trade Adjustment Assistance Survey, All Qualifying Industries N=6,429Exp. Incentive Treatment vs. Control ---Completion Rates
Sig. Chi Sq. 51.07
P<.001 ***
***
Trade Adjustment Assistance (TAA) 2006 Survey of Program Nationwide – Fisheries N=5,592
29%
58%
31%
48%41%
66%
35%
54%
41%
62%
48%54%
65%
45%
59%
70%
0%
10%
20%
30%
40%
50%
60%
70%
Catfish WASalmon
AKSalmon
TXShrimp
LAShrimp
ALShrimp
FLShrimp
GA-Shrimp
MSShrimp
NCShrimp
SCShrimp
Control T1: FC 2x Trt 2: $5, Priority 2x
**
**
******
***
Trade Adjustment Assistance (TAA) 2006 Survey of Program Nationwide – Commodities
N=837
73%
88%
59%
77%
59%
74% 76%
62%
0%
20%
40%
60%
80%
100%
Grapes CA Olives ID-Potatoes Blueberries Lychees
Control T1: FC 2x Trt 2: $5, Priority 2x
NSNSNS
2%1%
3%
69%
13%12%
Mail--No incentive Mail--$5 priority Web -- No incentive, FC
Web--$5 priority Tel.-No incentive, FC Tel-$5 priority
TAA Survey—Percentage of Completes Associated with Incentive Experimental Assignment and Survey Mode
Mail Completes
Telephone Completes
Web Completes
Suggestions for implementing establishment surveys
Effective Practice:
• Contact respondents multiple times
• Contact respondents in multiple modes.
• Allow for survey mode preference.
• Design surveys that reduce burden • short, conditional branching, ease
• Use leverages
Visual Design Effect – FARW Commercial Dealers N=1600
Background Shading w/ Visible Answer Boxes
Does it make a difference?Is there an interaction effect w/ Incentives?
34%
48%
36%
56%
0%
10%
20%
30%
40%
50%
60%
1
Bkg. Color No $5Bkg. Color $5
No Bkg Color No $5No Bkg Color $5
***
***
Visual Design TestFARW Growers With Pesticide License
**
24%
44%
21%
39%
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
1
Bkg. Color No $5
Bkg. Color $5
No Bkg Color No$5
No Bkg Color $5
******
***
• All 3 survey modes generated completes
• Exper. treatment (Cash incentive & priority mail) stimulated more responses in all 3 survey modes—large interaction effect.
• Offering web as an alternative option garnered 15%
• Telephone last 3% – still effective
Influential CircumstancesSaliency: Topic interest area of business emphasis for entity. High personal interest for respondent. High level of public or political concern
Role of survey sponsor. Regulator or source of certification. Mandatory reporting. Source of program $$ or sponsorship
Response Burden Complexity, length, multiple reports
What was learned from experimental trials
• Token cash incentives were effective across types of establishment and
industry populations.
• 2-day Priority mail was more effective than first class mail.
• Priority mail alone just slightly better than First class
• Cash incentives combined with priority mail was synergistic.
• Mixed mode strategies are very helpful and work.
• Respondents may have mode preferences.
• Establishment population characteristics and the selected respondent characteristics need to be considered jointly in explaining response.
• The survey circumstances and situation impact establishment response.