Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Social Network Analysis application in formation of students' groups

11,678 views

Published on

Слайды с презентацией предварительных результатов эксперимента по применению анализа социальных сетей для формирования студенческих групп, представленные на IV Международной конференции Российской ассоциации исследователей высшего образования «Университетские традиции: ресурс или бремя?»

Published in: Technology, Design
  • Be the first to comment

Social Network Analysis application in formation of students' groups

  1. 1. «Использование анализа социальных сетей для формирования учебных групп» Александр Пронин, Александр Семёнов IV международная конференция Российской ассоциации исследователей высшего образования «Университетские традиции: ресурс или бремя?» Москва 28.09.2013
  2. 2. Forming students groups seems to be a tough problem!
  3. 3. Challenge Task: regroup 4 student groups into 3 equally sized, homogenous groups with equal average score group A Group 1 group B N1 = N2 = N3 = 26 Group 2 group C group D Group 3 N – group size S – average score S1 = S2 = S3
  4. 4. 2 Approaches Approach1: «Weak link» Approach2: «Melting pot»
  5. 5. “Weak link” principle Boys Girls Subgroup 1 Subgroup 2 Subgroup 3
  6. 6. Approach 1 results Standard Network Analysis Row count Column count Link count Density Components of 1 node (isolates) Components of 2 nodes (dyadic isolates) Components of 3 or more nodes Reciprocity Characteristic path length Clustering coefficient Network levels (diameter) Network fragmentation Krackhardt connectedness Krackhardt efficiency Krackhardt hierarchy Krackhardt upperboundedness Degree centralization Betweenness centralization Closeness centralization Eigenvector centralization Reciprocal (symmetric)? 221_1 18.000 18.000 63.000 0.206 0 0 2 0.575 1.548 0.637 4.000 0.471 0.529 0.631 0.575 0.877 0.099 0.057 0.053 0.468 No (57% of the links are reciprocal) 221_2 18.000 18.000 48.000 0.157 0 0 1 0.333 3.167 0.348 9.000 0.000 1.000 0.860 0.314 1.000 0.121 0.323 0.314 0.301 No (33% of the links are reciprocal) 222_1 19.000 19.000 60.000 0.175 2 0 1 0.500 2.570 0.485 6.000 0.205 0.795 0.800 0.221 1.000 0.145 0.229 0.116 0.325 No (50% of the links are reciprocal) 222_2 19.000 19.000 46.000 0.135 2 0 1 0.533 2.759 0.296 7.000 0.205 0.795 0.883 0.412 1.000 0.191 0.191 0.125 0.305 No (53% of the links are reciprocal) 223_1 22.000 22.000 74.000 0.160 0 0 1 0.423 2.910 0.454 6.000 0.000 1.000 0.852 0.250 1.000 0.112 0.239 0.268 0.258 No (42% of the links are reciprocal) 223_2 22.000 22.000 58.000 0.126 0 0 1 0.349 2.379 0.351 6.000 0.000 1.000 0.895 0.820 0.767 0.176 0.070 0.177 0.425 No (34% of the links are reciprocal) 224_1 19.000 19.000 58.000 0.170 2 0 1 0.611 2.547 0.519 6.000 0.205 0.795 0.833 0.538 1.000 0.121 0.120 0.205 0.457 No (61% of the links are reciprocal) 224_2 19.000 19.000 49.000 0.143 2 0 1 0.531 2.365 0.391 6.000 0.205 0.795 0.867 0.724 1.000 0.119 0.143 0.242 0.482 No (53% of the links are reciprocal)
  7. 7. Approach 2 principle Initial groups Faculty-level ties New Groups Clustering
  8. 8. Group level Sociometric questionnaire Communication Approach 1 Trust Faculty level Approach 2 Information Advice Improvement
  9. 9. Key Results 1st interval Average score Rating position 2nd interval Year later 57% 37% 42% 35% 68% 52% % - percentage of students, who increased their average score and position in cumulative rating
  10. 10. Comparative results 1st interval Type of groups Score change Score increase 2nd interval Score change Score increase Year later Score change Score increase 48% +0,24 48% +0,84 48% +1 63% +0,49 34% +0,44 41% +0,45 Score change -- % of students, who increased their average score according to current rating Avg. score increase – mean increase of average score
  11. 11. Future steps • • • • Finalize data analysis Conduct a survey Publish a paper with results Present a methodology
  12. 12. Thank you! • Александр Пронин aspronin@hse.ru • Александр Семёнов semenoffalex@gmail.com http://jarens.ru

×