Nonresponse Reduction in Telephone Surveys

1

• Nonresponse Reduction in Telephone Surveys
• Edith D. de Leeuw
• Non-response Symposium, Amsterdam, June 5, 2000

2
Telephone Surveys the Full Circle Up and

• 1960 Difficult to imagine, radical idea
• 1970 Promising
• 1980 Heir apparent to face-to-face
• 1987 Coming of age
• Telephone Survey Methodology
• CATI
• 80-90 Zenith of importance
• more economical
• faster
• stricter interviewer control
• declining response f-t-f

3
The Full Circle and Down

• 1990-2000 under pressure
• Methods become less effective
• Response decreasing
• Why
• Changes in Technology
• Changes in Society
• Affecting
• Sampling
• Contacting
• Persuading

4
Sampling Coverage

• Multiphone/multiline households (e.g., ISDN)
• Overcoverage
• Mobile (cell) phones personal instead of hsh
• Solely mobile
• young
• single
• students
• lower economical status
• Combination household line personal mobile
• stable one/two person household
• families with children
• Only household line
• elderly (55)

5
Contacting

• Technological changes
• Mobile phone
• Answering machines
• Voice mail
• Caller ID/Call screening
• Societal changes
• Individualization
• At home patterns
• Consequences
• Harder to contact groups
• Higher costs due to increased attempts
• Higher costs when including mobile phones

6
Contacting Empirical findings

• Increased Attempt
• More contacts
• Different times
• Tailoring contact time
• Answering Machine/Voice Mail
• Connectors
• Leave short message
• name goal (prenotification)
• Call ID/ call screening
• First results positive for legitimate research
• 50 blocking marketeers vs 20 scientific polls
• university based more attempts needed, but as
  willing
• different profile TAM vs Call-ID (SES)

7
Theories on (Non) ResponseHelpful principles
from literature

• Establish legitimacy, trust
• Reduce respondent burden (costs)
• Reward respect, self-esteem, appreciation,
  incentive
• Induce reciprocation norm, social exchange,
  tit-for-tat
• Tailoring interviewer-respondent interaction
  Step-back, avoid premature refusal
• Use persuasion strategies
• Social validation (others, subjective norms)
• Authority
• Liking
• Scarcity

8
Legitimacy Trust

• Why so important
• Unsolicited calls increasing
• average 5 a year pro person (panel counted,
  Steenis 98)
• average 23 a year pro household (retro
  selfreport, Wunderink,2000)
• Nuisance
• 33 is not bothered by phone calls (Wunderink,
  2000)
• Wrong script expectation
• Marketering, Selling Under Guise of research
• Reassure, correct wrong expectancies, wrong script

9
Legitimacy Trust continued

• Advance letters
• Legitimacy
• Motivation
• Does have an effect (3-7), but
• Not always feasible
• Increase costs
• To be effective
• Well written
• Short informative

10
Legitimacy trust continued 2

• Special, free phone number
• Positive PR UMT (NL), CASRO (USA)
• Introduction
• More limited than face-to-face
• no visual communication channel
• shorter time
• No selling reassurance
• Effective UMT/RETO experiments

11
No Selling Experiments (RETO)

• 29 split-run experiments
• 10 Market Opinion Research Firms
• Cold-calling
• Varied topics
• varied populations
• 101 625 persons reached
• Combined in meta-analysis
• On Average Increase 2
• Small effect, but simple implementation, not
  costly
• Most effective as early as possible in
  introduction

12
Text Introduction

• Split Ballot Experimental control group, ad
  random
• Experimental Introduction
• Good ......, this is ..... from...... We are
  conducting a survey on..... We are not selling
  anything . To ensure we get a scientific sample,
  I would like to speak to.....
• Control Introduction
• Standard introduction agency
• Good ......, this is ..... from...... We are
  conducting a survey on..... I would like to speak
  to.....

13
After the first 30 seconds...

• Start with short introduction (name, not selling,
  topic)
• Then tailor further agenda-based, conversational
• Only successful with
• Well-trained interviewers
• Who listen and react
• Have tool-bag of good arguments
• Know survey/topic
• Special training needed
• Camburn et al, 1999

14
Some Elements From Toolbag

• Avoid request to participate (May I ask you
  some questions?)
• Houtkoop-Steenstra VandeBergh, 2000
• Avoid repeating negative phrases (you are not
  interested?, You are to old?
• Pondman, 1998 Dijkstra Smith, 2001
• Present positive information (social validation)
• Pondman, 1998 Hox et al, 1998
• React, keep conversation going
• Hox et al, 1998, DijkstaSmith, 2001

15
Some Elements Toolbag continued

• Adept language to respondent
• Hox et al, 1998
• Audibly enthusiastic
• Hox et al, 1998 Oksenbergh Cannel, 1988
• Motivated interviewers
• Pondman, 1998
• Supervision, monitoring on response
• De Heer De Leeuw, 2001

16
Reduce costs/increase rewards

• Short perceived length
• Pleasant atmosphere
• Voice interviewer
• Reactions
• positive frequent
• uh, thats a big help, thank you
• Incentives
• Do work (reciprocity)
• Do not effect data quality adversely
• When asked about 50 says not necessary (I do it
  for free)
• But, hugh majority does have airmiles etc...
• Lesson Should be perceived as thank-you not as
  payment

17
Summary Reducing Non-Response

• Reduce non-contacts
• Optimal call schedules, longer fieldwork
• Increase of costs and efforts
• Limit to what one can do
• Reduce refusals
• Good initial script tailoring
• Persuading interviewer training
• Incentives, reducing burden
• Combination needed (TDM)
• Single measures have only small effect
• Increase costs
• No such thing as refusing personality
• All partial information is valuable (to better
  adjust)

18
Theses for Discussion

• The special attention and publicity focussing on
  nonresponse will only INCREASE the nonresponse
  (social validation-principle).
• Reducing nonresponse effectively will need a
  (tailored) mixed-mode approach.
• Statistical adjustment is only effective if one
  plans on nonresponse and collects effective
  auxiliary information about nonrespondents.
• There aint no such thing as a free lunch!

19
Some useful references

• E.D. de Leeuw (1997). Nonresponse wat weten we en
  wat kunnen we eraan doen. Jaarboek NVMI.
• L. Pondman (1998). The influence of the
  interviewer on the refusal rate in telephone
  surveys. Amsterdam, proefschrift (samenvatting in
  jaarboek 1999)
• ZUMA-Nachrichten, Spezial 4 Nonresponse in
  survey research
• Kuusela Coverage
• Hox et al Successful interviewer strategies in
  telephone surveys.
• Special Issue of Journal of Official Statistics
  (1999)
• http//www.jos.nu
• R.M. Groves M.P. Couper (1998). Nonresponse in
  household interview surveys. New York Wiley
• D.A. Dillman (1978, 2000) TDM, New York Wiley
• Groves et al. (2001). Survey Nonresponse. New
  York Wiley