Lang methodological innovations-in_the_identification_and_modeling-194

241 views

Published on

A GOR presentation

Published in: Business, Technology, Education
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
241
On SlideShare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
0
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Lang methodological innovations-in_the_identification_and_modeling-194

  1. 1. 03/16/2011Methodological innovations in the identificationand modeling of Internet user profilesMeike Han / Volker Lang / Steffen Hillmert
  2. 2. Content of our presentation................................................................................................................1. Context of our study2. Theoretical background & empirical research so far3. Operationalisation, measurement & data4. Item analysis & preliminary models5. Conclusions & outlookGOR – 03/16/2011 1
  3. 3. 1. Informational environments, educational trajectories & life course decisions................................................................................................................Our research project …• is part of the Science Campus Tübingen (Cluster 3) 3). http://www.wissenschaftscampus- tuebingen.de/www/de/index.html?ref=folder41• is a longitudinal study (panel) of bachelor students.• constructs and tests instruments for assessing decision making processes in real life contexts. (e.g. choice of a field of study, subject related and private decisions, like flat hunting during the bachelor) g g )• focuses on the role of using, knowledge about and assessment of digital media (esp. the Internet).GOR – 03/16/2011 2
  4. 4. 2. Theoretical background................................................................................................................As internet access has become more and more widespread(the “first level digital divide” closes), first divide• questions regarding differences in knowledge about, competencies, attitudes towards and emotions related to using the internet (the “second level digital divide”) and• their influence on decision processes and decision behavior (the “third level digital divide”) become virulent.Students can be regarded as homogenous referring to internet access andfrequency of internet use (Hargittai, 2010a; Richter/Naumann/Horz, 2010).GOR – 03/16/2011 3
  5. 5. 2. Empirical studies so far focused on …................................................................................................................• Computer related knowledge, competencies, attitudes, anxiety (Richter/Naumann/Horz, 2010; Senkbeil/Wittwer, 2007; Campbell, 1990),• Internet related competencies (van Deursen/van Dijk, 2009), often self- reported ( p (Bonfadelli, 2002; Hargittai, 2007; Hargittai, 2010), , ; g , ; g , ),• types of internet related activities, locations of internet usage (Maar, 2004; Hargittai, 2010; DiMaggio et al., 2004),• additionally including an overall measurement for attitudes towards the internet (Bonfadelli, 2002).→ B ildi Building on thi research we focus on the relations between social this h f th l ti b t i l background, gender and internet related attitudes & emotions.GOR – 03/16/2011 4
  6. 6. 2. A social structural model of differences regarding the second and third level digital divide................................................................................................................ social background decision (e.g. parental education) behavior internet related media socialisation knowledge, gender regarding the internet competencies, (environment & education) ( ) attitudes & emotionsGOR – 03/16/2011 5
  7. 7. 3. Operationalisation................................................................................................................We measure internet related attitudes & emotions using two item batteriescontaining 12 items each.It can be assumed that the diffusion of the internet affects different areas of life(Bonfadelli,(Bonfadelli 2004; Hargittai 2010a) Hargittai, 2010a).We differentiate four aspects regarding these effects:1.1 knowledge: searching processing and validating information [4 items] searching,2. social relations: forms of interacting with others [6 items]3. in-/exclusion: potentials to integrate or discriminate against certain social groups (e g handicaped people) [10 items] (e.g.4. life style: means of improving or impairing individual behavior (e.g. enabling environmentally friendly habits) [4 items]GOR – 03/16/2011 6
  8. 8. 3. Measurement................................................................................................................These aspects are covered by our items for internet related attitudes & emotions:Attitude-item referring to the in-/exclusion aspect:“Das Internet trägt zu einer Befreiung des Einzelnen von gesellschaftlicher Bevormundung bei ” bei.(“The Internet facilitates the liberation of the individual from societalpaternalism.”)Emotion-item referring to the knowledge aspect:“Ich fühle mich von der Informationsflut im Internet bisweilen überfordert.”( Sometimes(“Sometimes I feel overstrained by the flood of information in the Internet ”) Internet. )All items are measured on a five-point scale of approval.GOR – 03/16/2011 7
  9. 9. 3. Data................................................................................................................• Online survey of bachelor students of the University of Tübingen (still running).• Recruited from the following fields of study: Biology, Business & Economics, Chemistry, Computer Sciences, Educational Theory, History of Art, Protestant Theology, Sociology, Spanish.• First wave of data collection since January 2011 focuses on … - internet related types of activities, attitudes, emotions and media related socialisation (environment & education). - social background information.• target population: 1017 (~ 70% women) contact sample: 448• realized sample (so far): 90 (started) / 62 (completed)• target sample: ~ 135 (started) / ~ 90 (completed)• analysis sample: 58 (~ 70% women)GOR – 03/16/2011 8
  10. 10. 4. Item analysis I: individual sum scores................................................................................................................Sum scores are calculated 5after reducing the five-point g pto a three-point scale of 4approval because of smalln’s in the end-points. ations mber of observa 3 2 num 1 0 0 5 10 15 20 25 30 35 40 45 48 sum of item1-item24 over individuals (mean = 23.5; s.d. = 6.1)GOR – 03/16/2011 9
  11. 11. 4. Item analysis II: overall response profiles................................................................................................................GOR – 03/16/2011 10
  12. 12. 4. Item analysis III: discrimination coefficients vs item difficulties vs................................................................................................................. item number 1 .8 ination coefficient 12 18 19 .6 22 2 5 9 3 15 14 6 .4 4 discrimi 20 13 8 4 10 21 7 1 23 .2 16 17 11 24 0 0 .2 .4 .6 .8 1 item difficultyGOR – 03/16/2011 11
  13. 13. 4. Item analysis IV: scale construction ................................................................................................................• All 24 items together have a Cronbach‘s α of .7551 (an average IIC of .1138).→ Drop items with a discrimination coefficient below .2.• The remaining 20 items have a Cronbach‘s α of .7758 (an average IIC of .1475). Cronbach s→ Iteratively drop the items which reduce Cronbach‘s α the most until no further increase is possible.• The remaining 10 items have a Cronbach‘s α of .8177 (an average IIC of .3097).• These 10 items all load on one factor with an eigenvalue of 2.99 (> 1).• These 10 items are used to construct a reduced internet related attitudes & emotions scale (IRAE reduced scale). GOR – 03/16/2011 12
  14. 14. 4. Model for our preliminary analysis of the second level digital divide................................................................................................................ measured using an a priori scale social background of parental education in years (e.g. parental education) (of the parent with the highest degree in the family) measured using the IRAE scale (or items) + + internet related gender attitudes & - emotionsGOR – 03/16/2011 13
  15. 15. 4. Response profiles by social background and gender................................................................................................................GOR – 03/16/2011 14
  16. 16. 4. OLS regression model of IRAE reduced scale on social background and gender................................................................................................................ raw values, predicted means & 95%-CIs: men women 1.5 5 P(TypeI-Error) < .01 for all ind. variables; n = 58; R^2 = .2; adj.-R^2 = .18; AIC = 96.13962; BIC = 100.2605; LR-test against model with interaction: P > ^2 = .87 1 IRAE reduced scalee 0 .5 E -.5 independent variables: -1 y years of edu.: .0247928(.0077144) ( ) women: -.585483(.1576625) -1.5 12 13 14 15 16 17 18 19 years of education (of the parent with the highest degree in the family)GOR – 03/16/2011 15
  17. 17. 4. An IRT approach to scale construction................................................................................................................• We estimate a multinomial logit model with item and person specific random intercepts (graded response model or 2PL model for multiple response categories) using one continuous random effect. ) d ff n (person level) = 58; n (item level) = 530; AIC = 1118; BIC = 1241.914• Including gender and our social background indicator does not improve the identification of the model (LR-test: P > χ^2 = .8).• Models using two continuous or an discrete random effect with two mass points d not converge using ML, indicating one-dimensionality (b this i do i ML i di i di i li (but hi can also be due to the small n on the person level).• We use this model to predict responses using MCMC simulation. The correlation between the “actual” and the predicted sum score is .1989. The in-sample predictive validity of our IRT based simulation is ~ .33.GOR – 03/16/2011 16
  18. 18. 5. Conclusions................................................................................................................• Our IRAE reduced scale has a reliable Cronbach’s α of .8177.• Factor and IRT analysis indicate that our IRAE reduced scale is one-dimensional.• There is a strong effect of gender on our IRAE reduced scale.• There is a weak but statistically robust effect of our social b k h i kb i i ll b ff f i l background d indicator on our IRAE reduced scale.• There is no indication for an interaction effect between gender and social b k i l background i our analysis. d in l i• Preliminary analysis suggest that there is no effect of the grade of the university entrance qualification on our IRAE reduced scale.• Our IRT model so far is in line with our item analysis but needs further improvement.GOR – 03/16/2011 17
  19. 19. 5. Outlook................................................................................................................• We replicate our analysis after finishing data collection.• We will further test our scale especially addressing what one scale, one- dimensionality theoretically means with regard to our scale.• We will integrate measures of media socialisation (included in our survey) into our analysis.• We will analyze the effect of our IRAE reduced scale on behavior regarding different types of internet usage.• W will additionally i l d measures of i We ill ddi i ll include f internet related competencies. l d i• In another version of our current survey and upcoming panel waves we collect data on decision processes (e.g. search for an internship). In the long run we will trace out how these processes are i fl l ill t th th influenced by db different types and ways of media usage.GOR – 03/16/2011 18
  20. 20. ................................................................................................................ Thank you f h k for your attention! iGOR – 03/16/2011 19
  21. 21. Literature................................................................................................................• Bonfadelli, H., 2002: The internet and Knowledge Gaps. A Theoretical and Empirical Investigation, in: European Journal of Communication 17(1), pp. 65- 84.• Campbell, N. J., 1990: High School Students Computer Attitudes and Attributions : Gender and Ethnic Group Differences, in: Journal of Adolescent Research 5(4), pp. 485-499.• DiMaggio, P. et al., 2004: From unequal access to differentiated use – A literature review and agenda for research on digital inequality, in: Neckerman, K. M. (ed.): Social Inequality, Russel Sage Foundation: New York, pp. 355- 400.• Hargittai, E., 2007: Characteristics of use differences and their implications for dealing with digital inequality, in: Otto, H. U. (ed.): Grenzenlose Cyberwelt? Zum Verhältnis digitaler Ungleichheit und neuen Bildungszugängen für Jugendliche, VS Verlag: Wiesbaden, pp. 121-136.GOR – 03/16/2011 20
  22. 22. Literature................................................................................................................• Hargittai, E., 2010: Digital Na(t)ives? Variation in Internet Skills and Uses among Members of the ‘‘Net Generation’’, in: Sociological Inquiry 80(1), pp. 92-113. ( )• Marr, M., 2004: Soziale Differenzen im Zugang und in der Nutzung des Internet. Aktuelle Befunde aus der Schweiz, in: Medienheft Dossier, pp. 19-27.• Richter, T.; Naumann, J.; Horz, H., 2010: Eine revidierte Fassung des Inventars zur Computerbildung (INCOBI-R), in: Zeitschrift für Pädagogische Psychologie 24(1), pp. 23-37. 23 37.• Senkbeil, M.; Wittwer, J., 2007: Die Computervertrautheit von Jugendlichen und Wirkungen der Computernutzung auf den fachlichen Kompetenzerwerb, in: i PISA K Konsortium D t hl d (ed.): PISA. Die E ti Deutschland ( d ) PISA Di Ergebnisse d d itt b i der dritten internationalen Vergleichsstudie, Waxmann: Münster, pp. 277-306.GOR – 03/16/2011 21
  23. 23. Literature................................................................................................................• van Deursen, A. J.A.M.; van Dijk, J. A.G.M., 2009: Using the internet: skill related problems in users’ online b h l d bl ’ l behavior, in: Interacting with computers h (preprint). Available online:http://www.utwente.nl/gw/mco/bestanden/Using%20the%20Internet-%20Skill%20related%20problems.pdfGOR – 03/16/2011 22

×