Presentation_1375280653597

Friendship and assistance ties of students:

effect of GPA

D.Valeeva, O.Poldin, M.YudkevichCenter for Institutional Studies (Higher School of Economics)

Moscow, Russia

XXXIII Sunbelt Social Network Conference,May 21-26, 2013, Hamburg, Germany

Network as dependent variable Network as independent variable

Student Network effects Network effects Student

characteristics (homophily, (peer group abilities

(race, sex etc.) propinquity etc.) effects) (academic

achievement)

DeFour & Hirsch (1990) Antrobus (1988)D’Augelli & Hershberger (1993) Thomas (2000)Mayer & Puller (2008) Sacerdote (2001) Zimmerman

(2003)

Our contribution:• friendship and study assistance networks• p* modeling of student networks• administratively formed groups• Russian sample

Main research areas of student social networks

Main network effects

• Reciprocity: tendency of friendship ties to be mutual

• Transitivity: tendency of ties to be closed in triads (“friend of my friend is also my friend”)

• Homophily: tendency of ties to be formed between individuals with the same characteristics and abilities

• Propinquity: tendency of ties to be formed between individuals situated in same geographical and institutional settings

• Popularity: tendency of more popular and active individuals to form more toes and to be more successful

Hypotheses

About reciprocity and transitivity effects:H1. Friendship ties are more mutual than assistance tiesH2. Friendship ties are more transitive than assistance ties

About homophily and propinquity effects:H3. There are propinquity effects between students in study

group and type of tuition (in both friendship and assistance networks)

H4. There are homophily effects between students in their gender (only in friendship network)

H5. There are homophily effects between students in their GPA (in both friendship and assistance networks)

About popularity effects:H6. Academically successful students are more popular (in

both friendship and assistance networks)

Data

Students of Economics Department in Higher School of Economics

• 2nd year students: 94 students• 3rd year students: 118 students • ~80% of the whole network is described• Students from 7 study groups (administratively formed)• Tuition free and full tuition students study together

Questions in the questionnaire:1. Please indicate up to 5 of your classmates with whom

you spend most of your time2. Please indicate up to 5 of your classmates to whom

you address on some educational help + Additional information about gender and type of tuition of

students, their study group and GPA

Descriptive statistics-1

Mean Min Max No. of obs.

No. of friendship ties

3.49(1.31)

0 5 94

No. of assistance ties

3.35(1.47)

0 5 94

GPA 6.89(0.93)

4.70 9.52 94

% of men 37% 94

% of tuition free students

68% 94

Friendship network

Assistance network

Descriptive statistics-2

Friendship network

Assistance network

Density 0.04 0.04

Reciprocity 0.55 0.40

Transitivity 0.36 0.35

Methods

• ERGM or p* => estimation of probability of tie in network

Variables:a) Network characteristics: density, reciprocity,

transitivityb) Student characteristics: homophily, propinquity and

popularity on gender, group, tuition, GPA

Network characteristics

Network characteristics

Picture

Reciprocity

GWESP(geometrically weighted edgewise shared partner)

Mixed 2-star

Effects Friendship network

Assistance network

Effects Estimate(SE) Estimate(SE)

Density -2.112* (0.895) -11.971*** (1.249)

Reciprocity 2.215*** (0.244) 1.013*** (0.286)

Mixed 2-star -0.193*** (0.037) -0.126*** (0.026)

Gwesp 1.231*** (0.120) 1.131*** (0.127)

Friendship network 3.282*** (0.210)

Assistance network 3.205*** (0.199)

Homophily on gender (women) 0.452* (0.180) 0.334 (0.204)

Homophily on gender (men) 0.502*** (0.152) 0.023 (0.155)

Propinquity on study group 0.822*** (0.129) 1.101*** (0.158)

Propinquity on tuition (free) 0.137 (0.154) 0.229 (0.161)

Propinquity on tuition (full) 0.435* (0.197) 0.198 (0.305)

Abs. diff. in GPA -0.114 (0.089) -0.185+ (0.111)

Popularity on GPA -0.334** (0.123) 1.029*** (0.163)

Conclusions

Reciprocity and transitivity effects:1. Friendship ties are more mutual than assistance ties2. There is no significant differences between networks in their

transitivity measures

Homophily and propinquity effects:3. Students connections are highly defined by their study groups4. Homophily on gender is significant only in friendship ties5. Propinquity on tuition is significant only in friendship ties (for

full tuition students)6. Difference in GPA doesn’t determine friendship or assistance

ties of students

Popularity effects:7. Academically successful students are more popular (have

more in-going ties) in assistance networks, but less popular in friendship networks

GOF plots for friendship network-1

GOF plots for friendship network-2

GOF plots for assistance network-1

GOF plots for assistance network-2