Download presentation

Presentation is loading. Please wait.

Published bySara Law Modified over 5 years ago

1
Visual Analogue Scales in Online Surveys Session 19: Scale Construction & Methods' Effects Frederik FunkeMarch 23 rd, 2005 General Online Research 2005

2
Objective 1/20 Comparison of 4- & 8-point categorial scales and Visual Analogue Scales (VAS) Does type of scale affect… … means & variance of responses? … response times? … nonresponse & dropout? What is the appropriate way of categorizing VAS values?

3
Overview Visual Analogue Scales (VAS) Design & sample Results Conclusion 2/20

4
First description in 1921 Main disadvantages –not suitable for all populations –JavaScript has to be enabled Main advantages –high reliability and validity –point of socially desired responses is not easy to identify 3/20 Visual Analogue Scales

5
Scales to be compared: VAS (80 discrete values) delivered by JavaScript Verbal and animated instruction: inappropriate (trifft überhaupt nicht zu) appropriate (trifft voll und ganz zu) VAS inappropriate (trifft überhaupt nicht zu) appropriate (trifft voll und ganz zu) 4-point inappropriate (trifft überhaupt nicht zu) appropriate (trifft voll und ganz zu) 8-point 4/20 Design

6
Questionnaire: –16 items (behaviour in groups) –experimental design –dichotomizing of questionnaire 9 groups: 4-point–4-point4-point –8-point4-point–VAS 8-point–4-point8-point –8-point8-point–VAS VAS–4-pointVAS–8-pointVAS–VAS –blind Sample –self selected –recruitment via email, websites & newsletters –n = 667 5/20

7
Results Mean, standard deviation Response times Nonresponse Dropout Categorizing VAS values 6/20

8
Results 7/20 VAS8-point4-point

9
Results Mean differences are significant (p < 0,01) ( 4-point)= 58,1 ( 8-point)= 60,0 ( VAS)= 57,5 8/20

10
Results Standard Deviation differences are significant (p < 0,01) s (4-point)=34,3 s (8-point)=31,3 s (VAS)=28,1 9/20

11
Results ( 4-point)=7,0 ( 8-point)=7,1 ( VAS)=7,8 differences between VAS & categorial scales are significant (p < 0,01) Response Times 10/20

12
Results s (4-point)=4,4 s (8-point)=4,0 s (VAS)=5,2 differences are significant (p < 0,01) Standard Deviation (Response Times) 11/20

13
Results Item Nonresponse 12/20 without lurkers

14
Results 13/20 Dropout

15
How to categorize VAS-Values? When comparing frequencies, VAS values have to be categorized: 2 ways of categorizing VAS: (1) linear transformation: equal intervals form one category (2) transformation with reduced extremes: extreme categories width is 2/3 of inner categories width 14/20

16
Middle categories can be described in 3 ways: –fits perfectly (e.g. 2) –fits, but tendency to the right side (e.g. 2-) –fits, but tendency to the left side (e.g. 2+) Extreme categories only can be described in 2 ways: –fits perfectly (e.g. 1) –fits, but tendency to adjoining category (e.g. 1-) Model of Reduced Extremes 11-2+23+2-33-4+4 2 sections3 sections 2 sections 15/20

17
Linear transformation VAS 4-point: Transformation with reduced extremes VAS 4-point: Linear transformation VAS 8-point: Transformation with reduced extremes VAS 8-point: Transformation of VAS 16/20 1 80 1-1617-4041-6465-80 1234 180 1-78-1819-2930-4041-5152-6263-7374-80 87654321 1 1-2021-4041-6061-80 1234 80 1 1-1011-2021-3031-4041-5051-6061-7071-80 86754321 80

18
Reducing extreme categories width leads for every category to greater correspondence between 4-point scale and VAS index of all 16 items VAS (linearly transformed)4-pointVAS (reduced extremes) Transformation of VAS Values 17/20

19
Greater correspondence between 8-point and VAS after transformation with reduced extremes Only at 1 category advantages for linear transformation index of all 16 items VAS (linearly transformed)8-pointVAS (reduced extremes) Transformation of VAS Values 18/20

20
Conclusion Modest differences in mean Lower variance for VAS higher reliability Increase of response time when using VAS Higher dropout rate when using VAS More lurkers when using VAS More nonresponse with VAS Transforming VAS with reduced extremes leads to greater correspondence with categorial scales 19/20

21
Thanks for your time! Further thanks for support: –Ulf-Dietrich Reips (University of Zurich) –Dagmar Krebs (University of Gießen) –Timo Gnambs –gir-l, WEXTOR & Experimental Psychology Lab this presentations URL: http://www.FrederikFunke.de/papers/gor2005.htm

Similar presentations

OK

Feichter_DPG-SYKL03_Bild-01. Feichter_DPG-SYKL03_Bild-02.

Feichter_DPG-SYKL03_Bild-01. Feichter_DPG-SYKL03_Bild-02.

© 2019 SlidePlayer.com Inc.

All rights reserved.

To make this website work, we log user data and share it with processors. To use this website, you must agree to our Privacy Policy, including cookie policy.

Ads by Google