Presentation is loading. Please wait.

Presentation is loading. Please wait.

The Effect of Interviewer And Personal Visits on Response Consistency

Similar presentations


Presentation on theme: "The Effect of Interviewer And Personal Visits on Response Consistency"— Presentation transcript:

1 The Effect of Interviewer And Personal Visits on Response Consistency
Jonathan Katz QDET2 Conference Miami, FL November 10th, 2016 Disclaimer: Any views expressed are those of the author and not necessarily those of the U.S. Census Bureau.

2 Literature Reliability
Optimal goal is to design questions where responses will be replicated with the same true values Unfeasible “parallel measures” Pre-testing Re-interviews (Groves, 1989)

3 Literature continued…
Interviewer Variability Had different findings depending on whether interviewers were included as a random effect (Dijkstra,1983) (O’Muircheartaigh & Campanelli, 1998) Hard-To-Reach Respondents Mixed research if increasing effort for hard-to-reach respondents impact data quality Peytchev, Peytcheva, & Groves (2010) found abortion experiences to be underreported Hox, de Leeuw, & Chang (2012) found “data quality” not to be affected between “eager” and “reluctant” respondents Social/Political Attitudinal Items Olson (2006) discussed that length of marriage item may have had less measurement bias (compared to time since divorce) because it’s a more “salient” event

4 Motivation Measure how interviewers and number of contact attempts each contribute to reliability of responses to attitudinal items ( using European Social Survey Round 7 Data ) Journal papers have focused upon the relationship between reliability and number of contact attempts (or recruiting hard-to-reach survey non-respondents) the relationship between interviewers and reliability

5 Data Used for Study European Social Survey (ESS) Round 7
40,185 cases 22 countries participated Supplementary questionnaire administered face-to-face following main questionnaire (Country response rates ranged from 30 % to 80 %) Final Data Set N = 28,415 cases 1,819 interviewers 21 countries

6 Items of Interest Six attitude items on an 11 point scale (0-10) administered both in main questionnaire and supplementary questionnaire 3 social attitude items 3 political attitude items

7 Social Attitude Items Racial Attitude And how important do you think being white should be in deciding whether someone born, brought up and living outside [country] should be able to come and live here? Extremely unimportant – – Extremely important Language Attitude Please tell me how important you think being able to speak [country’s official language(s)] should be in deciding whether someone born, brought up and living outside [country] should be able to come and live here. Country Way of Life Attitude Now, how important do you think being committed to the way of life in [country] should be in deciding whether someone born, brought up and living outside [country] should be able to come and live here?

8 Political Attitude Items
Role in Political Issues Attitude How able do you think you are to take an active role in a group involved with political issues? Completely Unable – – Completely Able Participation in Politics Attitude How confident are you in your own ability to participate in politics? Completely Unconfident – – Completely Confident Ease in Politics Attitude How easy do you personally find it to take part in politics? Extremely Difficult – – Extremely Easy

9 Methods Response Score Differences for Histograms
Multilevel Logistic Regression Models Dependent Variable Response Score Difference of ‘0’ or ‘1’  Reliability Response Score Difference of ‘>1’  Low reliability Fixed Effects Respondent- age, education, gender Interviewer- age, gender Number of Personal Visits to Completion Random Effects Country of Interview (Level 1) and Interviewer ID (Level 2)

10 Hypotheses Interviewers and the number of contact attempts will both significantly affect reliability in items from main questionnaire to supplementary questionnaire Variability between interviewers will be significantly different than zero An increase of contact attempts to interview completion (higher “burden”) would be associated with less reliability across responses to identical items Association between personal visits and reliability would be more pronounced for social attitude items compared to political attitude items

11 Social Attitude Response Score Difference Histograms

12 Political Attitude Response Score Difference Histograms

13 Percentage of Personal Visits to Interview Completion Histogram
Mean: 2.85 Median: 2

14 Multilevel Logistic Regression Model of Social Attitudes Items
Characteristic Racial Attitude Language Attitude Country Way of Life Attitude Estimate (Standard Error) Odds Ratio Fixed Effect Number of Contact Attempts 0.04 (0.01)*** 1.04 0.01 (0.01)* 1.01 0.00 (0.01) 1.00 Variance Components Random Effect Interviewer 0.24 (0.02)*** -- 0.16 (0.02)*** 0.17 (0.02)*** Country 0.14 (0.05)*** 0.06 (0.02)*** ***p<0.01 **p<0.05 *p<0.10

15 Multilevel Logistic Regression Model of Political Attitudes Items
Characteristic Role in Political Issues Attitude Participation in Politics Attitude Ease in Politics Attitude Estimate (Standard Error) Odds Ratio Fixed Effect Number of Contact Attempts 0.01 (0.01) 1.01 0.00 1.00 Variance Components Random Effect Interviewer 0.08 (0.01)*** -- 0.07 (0.01)*** 0.10 (0.01)*** Country 0.11 (0.05)** 0.14 0.10 (0.04)** ***p<0.01 **p<0.05 *p<0.10

16 Testing the Variability Between Interviewers Equal to 0
Category Item Chi-Square Statistic Social Attitudes Racial Attitude 305.68*** Language Attitude 187.35*** Country Way of Life Attitude 228.82*** Political Attitudes Role in Political Issues Attitude 98.22*** Participation in Politics Attitude 56.42*** Ease in Politics Attitude 65.87*** ***p<0.01 **p<0.05 *p<0.10

17 Findings for Hypotheses
Interviewers Significant between-interviewer variability at p<0.01 for all six items Contact Attempts Statistically significant at p<0.10 in two of the social attitude items in multilevel logistic regression models However, both these regression coefficients were positive (opposite direction than hypothesized!) None of the “contact attempts” regression coefficients for political attitude items were significant

18 Future Research Goal is to examine if there is significant differences between social and political attitude items in terms of variability across interviewers. In other words, is variability between interviewers more significant for social attitude items compared to political attitude items?

19 Limitations ESS does not use an interpenetrated design.
Income item was not added in the model Interviewer race and respondent race were unavailable End point labels on low end of scale for political attitude items had different wording Did “burden” play a role?

20 jonathan.m.katz@census.gov or (301) 763-5956
Questions? or (301) I would like to thank Andy Peytchev and Luke Larsen for their consulting on this research.

21 Appendix

22 Estimate (Standard Error)
Racial Attitude Characteristic Value Estimate (Standard Error) Odds Ratio Number of Contact Attempts (0.0088)*** 1.042 Gender Female (0.0278)* 1.051 Interviewer Gender (0.0379)*** 1.156 Education Lower secondary (0.0555) 0.982 Lower tier upper secondary (0.0570) 1.065 Upper Tier Upper Secondary (0.0567)*** 1.226 Advanced Vocational, sub-degree (0.0576)*** 1.251 Lower Tertiary Education, BA Level (0.0638)*** 1.518 Higher Tertiary Education, >= MA level (0.0625)*** 1.791 Respondent Age - - (0.0008)*** 0.987 Interviewer Age (0.0015)** 1.003 Variance Components Interviewer (0.0198)*** -- Country (0.0503)*** ***p<0.01 **p<0.05 *p<0.10

23 Estimate (Standard Error)
Language Attitude Characteristic Value Estimate (Standard Error) Odds Ratio Number of Contact Attempts (0.0077)* 1.013 Gender Female (0.0257)* 0.955 Interviewer Gender (0.0336)* 1.060 Education Lower secondary (0.0534) 0.979 Lower tier upper secondary (0.0544)* 1.094 Upper Tier Upper Secondary (0.0539) 1.058 Advanced Vocational, sub-degree (0.0545)* 1.111 Lower Tertiary Education, BA Level (0.0589)*** 1.167 Higher Tertiary Education, >= MA level (0.0573)*** 1.218 Respondent Age - - (0.0007)*** 0.997 Interviewer Age (0.0013)*** 1.004 Variance Components Interviewer (0.0155)*** -- Country (0.0219)*** ***p<0.01 **p<0.05 *p<0.10

24 Country Way of Life Attitude
Characteristic Value Estimate (Standard Error) Odds Ratio Number of Contact Attempts (0.0076) 1.002 Gender Female (0.0254) 0.985 Interviewer Gender (0.0338) 1.045 Education Lower secondary (0.0529) 1.023 Lower tier upper secondary (0.0537) 1.082 Upper Tier Upper Secondary (0.0535)*** 1.186 Advanced Vocational, sub-degree (0.0539)** 1.125 Lower Tertiary Education, BA Level (0.0584)*** 1.312 Higher Tertiary Education, >= MA level (0.0566)*** 1.258 Respondent Age - - (0.0007)*** Interviewer Age (0.0013)** Variance Components Interviewer (0.0161)*** -- Country (0.0146)*** ***p<0.01 **p<0.05 *p<0.10

25 Ease in Politics Attitude
Characteristic Value Estimate (Standard Error) Odds Ratio Number of Contact Attempts (0.0073) 0.997 Gender Female (0.0247)*** 1.167 Interviewer Gender (0.0306)*** 1.126 Education Lower secondary (0.0531)*** 0.829 Lower tier upper secondary (0.0534)*** 0.814 Upper Tier Upper Secondary (0.0533)*** 0.787 Advanced Vocational, sub-degree (0.0535)*** 0.752 Lower Tertiary Education, BA Level (0.0573)*** 0.678 Higher Tertiary Education, >= MA level (0.0556)*** 0.633 Respondent Age - - (0.0007)** 1.001 Interviewer Age (0.0012) Variance Components Interviewer (0.0126)*** -- Country (0.0426)** ***p<0.01 **p<0.05 *p<0.10

26 Participation in Politics Attitude
Characteristic Value Estimate (Standard Error) Odds Ratio Number of Contact Attempts (0.0073) 0.996 Gender Female (0.0247)*** 1.230 Interviewer Gender (0.0296)*** 1.088 Education Lower secondary (0.0531)*** 0.734 Lower tier upper secondary (0.0535)*** 0.723 Upper Tier Upper Secondary (0.0535)*** 0.684 Advanced Vocational, sub-degree (0.0536)*** 0.660 Lower Tertiary Education, BA Level (0.0573)*** 0.573 Higher Tertiary Education, >= MA level (0.0556)*** 0.522 Respondent Age - - (0.0007)** 1.002 Interviewer Age (0.0012)** 1.003 Variance Components Interviewer (0.0117)*** Country (0.0544)** ***p<0.01 **p<0.05 *p<0.10

27 Role in Political Issues Attitude
Characteristic Value Estimate (Standard Error) Odds Ratio Number of Contact Attempts (0.0073) 1.006 Gender Female (0.0247)*** 1.230 Interviewer Gender (0.0299)** 1.088 Education Lower secondary (0.0533)*** 0.799 Lower tier upper secondary (0.0536)*** 0.780 Upper Tier Upper Secondary (0.0535)*** 0.738 Advanced Vocational, sub-degree (0.0537)*** 0.691 Lower Tertiary Education, BA Level (0.0574)*** 0.579 Higher Tertiary Education, >= MA level (0.0557)*** 0.551 Respondent Age - - (0.0007)*** 1.004 Interviewer Age (0.0012) 1.001 Variance Components Interviewer ( )*** -- Country ( )** ***p<0.01 **p<0.05 *p<0.10


Download ppt "The Effect of Interviewer And Personal Visits on Response Consistency"

Similar presentations


Ads by Google