Download presentation
Presentation is loading. Please wait.
Published byGervase Russell Modified over 9 years ago
1
Fieldwork efforts Monitoring fieldwork efforts Monitoring fieldwork efforts: Did interviewers /survey organisations implement fieldwork guidelines and procedures? Evaluation of fieldwork efforts Evaluation of fieldwork efforts: Did fieldwork efforts result in increased response rates and more representative surveys? Understanding Understanding the reasons of success/failure by combining information on monitoring and evaluation
2
1. Data Call record data from 12 countries: The Netherlands, Hungary, Poland, Portugal, Finland, Switzerland, Great Britain, Ireland, Hungary, Spain, Greece, Israël each visiteach sample unit Information on timing, outcome and mode of each visit made to each sample unit
3
2. Reducing noncontact rates Reduce noncontact rate by: – making many calls or visits – making visits at varied times of day and week Contacting procedure – minimum of four calls – minimum of 1 weekend call – minimum of 1 evening call – all visits before first contact face-to-face
4
Average number of calls/visits made to noncontacts Very High Very High: Switzerland (73), Great Britain (9.4), Spain (7.8) High High: Finland (5.0), Greece (5.0), Poland (4.8), Slovenia (4.9), The Netherlands (5.1), Portugal (5.1) Moderate Moderate: Hungary (4.0) Low Low: Ireland (3.3), Israël (2.6)
5
Number of call attempts Countries with rather high noncontact rates and low number of call attempts (e.g. Ireland) might increase call attempts No “clear-cut” relationship between number of call attempts and noncontact rates Some countries achieve the target noncontact rate with only moderate or low efforts extended interviewer efforts Other countries rely heavily on extended interviewer efforts
6
Noncontact rates before and after extended interviewer efforts (>4 calls)
7
Number of call attempts Necessity to make many calls/visits is especially high in Great Britain, Spain, Portugal and Switzerland: Why? at-home patterns Less favourable “at-home patterns” ? timing of visits Less optimal timing of visits?
8
Percentage of visits made on a weekday morning or afternoon for the first four visits
9
Percentage of sample units successfully contacted at first visit by timing of visit
10
Timing of visits Countries with less favourable at-home patterns can adapt strategies and make more evening/weekend calls Portugal : high benefits of weekend visits, high % weekend visits Great Britain & Ireland: high benefits of evening visits but low % of evening visits However, making many evening and weekend calls is not necessarily more efficient
11
3. Increasing survey participation Advance letter Respondent Incentives Interviewer training Refusal conversion
12
Refusal conversion procedures Re-approaching reluctant sample units and asking them to reconsider participation ESS recommendations: reissue all soft and as many hard refusals as possible to another interviewer Implementation will vary because of differences in ressources, necessity and definition of “soft” refusal
13
Percentage of refusals re-approached Very High Very High: The Netherlands (88%), Switzerland (84%), Great Britain (77%) High High : Finland (50%), Greece (54%) Moderate Moderate: Spain (34%), Slovenia (33%), Poland (24%), Israël (17%) Low Low: Ireland (1.9%), Hungary (5.3%)
14
Conversion success rate 40%: The Netherlands 30-40% : Slovenia 20-30%: Israël, Poland, Finland 10-20%: Greece, Great Britain <10%: Switzerland
15
Response rates before and after refusal conversion
16
How to explain success in the Netherlands? Second letter to reluctant sample units Incentives increased with financial donations up to 5 Euro Incentives supplemented with quiz Highly motivated survey organisation and interviewers?
17
Does refusal conversion reduce nonresponse bias? Evaluation of refusal conversion procedures should also take into account relationship between increasing response rates and reduction of bias (Stoop,2003) This can be done by comparing converted refusals with cooperative respondents The Netherlands The Netherlands represents an interesting case: what happens with survey estimates if response rates increase from 54% to 68%
18
Effects on survey estimates % Cooperative respondents Converted refusals Total EDUCATION (HIGH)24.019.223.3 INCOME (HIGH)42.834.141.0 SEX (MALE)55.657.255.9 AGE (+65)18.121.118.7 SINGLE23.022.322.9 MEMBER ORG. (NO)15.720.216.6 FRIENDS (NO)33.440.134.7 SOC. TRUST (LOW)22.424.622.8 POL. INTEREST (HIGH)69.058.966.9 IMMIGRANTS65.760.964.7 N 18804842364
19
Effects on survey estimates Most differences are in the expected direction Large & significant differences for political interest (cfr. Voogt & Saris, 2003) Increasing response rate from 54% to 68% seems to reduce bias and improve at least some survey estimates
20
Evaluation of refusal conversion Efforts in Great Britain & Switzerland were not cost-effective: hardly any effect on response rates Refusal conversion did work in the Netherlands, probably due to a range of conversion strategies In the Netherlands, the profile of converted refusals indicates that they are different on a number of key survey variables Results suggest that countries with low response rates might reach higher response rates and more representative samples by adopting the successfull approach of the Netherlands
21
4. Conclusions Call record data are a very usefull and essential tool for monitoring and evaluating fieldwork strategies The analysis raises some interesting questions: e.g. about refusal conversion and timing of visits Results should feed back into survey process and lead to actions to improve fieldwork procedures
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.