0 vues

Titre original : Asa Snaintro

Transféré par Tatheer Fatima

- Brass 2004
- White-Network Analysis and Ethnographic Problems
- 1069 - Behind Acquisitions of Alliance Partners - Exploratory Learning and Network Embeddedness
- Baranyi Et Al 2011 Ecological Indicators
- tmp70D5
- WOM on Marketing.pdf
- Clinical Pharmacology & Therapeutics Volume Issue 2013 [Doi 10.1038_clpt.2013.176] Harrold, JM; Ramanathan, M; Mager, De -- Network-Based Approaches in Drug Discovery and Early Development
- net work analysis of a tourism destination
- 1005.2122
- 50 DynProbit (Montreal2006)
- Ok Network 5
- ACS07 Tania Experim
- Company Background
- Resources for Understanding Network Effectiveness
- Media Disruption Exacerbates Revolutionary Unrest
- 1222518975
- Modelo Canvas-uol Host
- Strategic management:stakeholders
- Social Networking Website
- Survey of Privacy Issues in Social Networking

Vous êtes sur la page 1sur 147

San Francisco, August 2004

James Moody

Introduction

their associating in work and in play, in love and in war, to trade or to

worship, to help or to hinder. It is in the social relations men establish that

their interests find expression and their desires become realized.”

Peter M. Blau

Exchange and Power in Social Life, 1964

"If we ever get to the point of charting a whole city or a whole nation, we

would have … a picture of a vast solar system of intangible structures,

powerfully influencing conduct, as gravitation does in space. Such an

invisible structure underlies society and has its influence in determining the

conduct of society as a whole."

J.L. Moreno, New York Times, April 13, 1933

These patterns of connection form a social space, that can be seen in multiple

contexts:

Introduction

Source: Linton Freeman “See you in the funny pages” Connections, 23, 2000, 32-42.

Introduction

High Schools as Networks

Introduction

And yet, standard social science analysis methods do not take this space

into account.

“For the last thirty years, empirical social research has been

dominated by the sample survey. But as usually practiced, …, the

survey is a sociological meat grinder, tearing the individual from his

social context and guaranteeing that nobody in the study interacts

with anyone else in it.”

Allen Barton, 1968 (Quoted in Freeman 2004)

identify social connectivity using only our intuitive understanding.

extend our theoretical intuition of the patterns that construct social

structure.

Introduction Why do Networks Matter? Local vision

Introduction Why do Networks Matter? Local vision

Introduction

which can reflect a power distribution or influence attitudes

and behaviors. Our understanding of social life improves if

we account for this social space.

on the spread of “goods” or power dynamics that could not be

seen focusing only on individual behavior.

Introduction

and identifying connections among actors. SNA

•is motivated by a structural intuition based on ties linking

social actors

•is grounded in systematic empirical data

•draws heavily on graphic imagery

•relies on the use of mathematical and/or computational

models.

relating types of observable social spaces and their relation to

individual and group behavior.

1. Introduction

2. Social Network Data

a. Basic data Elements

b. Collecting network data

c. Basic data structures

3. Measuring Networks

a. Flows within of goods in networks

1) Topology

2) Time

b. Structure of Social Space

1) Small Worlds, Scale-Free, Triads

2) Cohesive Groups

3) Role Positions

4. Modeling with Networks

a. Modeling Behaviors with Networks

1) Peer attribute models

2) Network Autocorrelation Models

3) Dyad / QAP Models

b. Modeling Network Network Structure

1) QAP for network structure

2) Exponential Random Graph Models

5. SNA Computer Programs

Social Network Data

actors and their relations.

Actors are referred to variously as:

Nodes, vertices or points

Relations are referred to variously as:

Edges, Arcs, Lines, Ties

Example:

b d

a c e

Social Network Data

Binary or Valued

Directed or Undirected

b d b d

a c e a c e

Undirected, binary Directed, binary

b d b d

1 3 1 2

a c 4

e a c e

Undirected, Valued Directed, Valued

Social Network Data

Social network data are substantively divided by the number of

modes in the data.

actors in the network. All the nodes are of the same type (people,

organization, ideas, etc). Examples:

Communication, friendship, giving orders, sending email.

their friends), but you can use multiple-informant data, which is

more common in child development research (Cairns and

Cairns).

Social Network Data

Social network data are substantively divided by the number of

modes in the data.

all ties are across classes. Examples:

People as members of groups

People as authors on papers

Words used often by people

Events in the life history of people

The two modes of the data represent a duality: you can project

the data as people connected to people through joint membership

in a group, or groups to each other through common membership

your nodes.

Social Network Data

1) Ego-network

- Have data on a respondent (ego) and the people they are connected to

(alters). Example: 1985 GSS module

2) Partial network

- Ego networks plus some amount of tracing to reach contacts of

contacts

actors in the relevant population

Social Network Data

- Data on all actors within a particular (relevant) boundary

- Never exactly complete (due to missing data), but boundaries are set

sciences, friendships among all students in a classroom

networks today, though I will briefly mention some standard uses of

ego-network data.

Social Network Data

Collecting Network Data

surveys, published accounts, special informants, etc.

In general, you can only make conclusions about relations among the

set of nodes you have collected, so it is important to observe as

much of the network as possible.

Social Network Data

Collecting Network Data

have breadth over depth. Having detailed information on <50% of the

sample will make it very difficult to draw conclusions about the general

network structure.

b) Question format:

• If you ask people to recall names (an open list format), fatigue will

result in under-reporting

• If you ask people to check off names from a full list, you can often get

over-reporting

c) It is common to limit people to ~5 nominations. This will bias network stats

for stars, but is sometimes the best choice to avoid fatigue.

d) Concrete relational indicators are best (who did you talk to?) over attitudes

that are harder to define (who do you like?)

Social Network Data

Collecting Network Data

2) Many secondary sources (particularly for 2-mode data)

3) National Longitudinal Survey of Adolescent Health (Add Health)

Social Network Data

Basic Data Structures

Working with pictures.

No standard way to draw a sociogram: each of these are equal:

Social Network Data

Basic Data Structures

great deal of good work to be done on using visualization to build network

intuition.

I recommend using layouts that optimize on the feature you are most interested

in, and find that either a hierarchical layout or a force-directed layout are best.

Social Network Data

Basic Data Structures

b d b d

a c e a c e

Undirected, binary Directed, binary

a b c d e a b c d e

a 1 a 1

b 1 1 b 1

c 1 1 1 c 1 1 1

d 1 1 d

e 1 1 e 1 1

Social Network Data

Basic Data Structures

a 1 ab

b 1 1 ab ba

bac bc

c 1 1 1 cbde cb

d 1 1 dce cd

e 1 1 ecd ce

dc

de

ec

ed

Measuring Networks: Flow

Measuring Networks: Flow

to another (pij), two factors affect flow through a network:

Topology

-the shape, or form, of the network

- Example: one actor cannot pass information to another unless they

are either directly or indirectly connected

Time

- the timing of contact matters

- Example: an actor cannot pass information he has not receive yet

Measuring Networks: Flow

and centrality

Connectivity refers to how actors in one part of the network are connected to

actors in another part of the network.

true if there is a chain of contact from one actor to another.

• Distance: Given they can be reached, how many steps are they from

each other?

Measuring Networks: Flow

Without full network data, you can’t distinguish actors with limited

information potential from those more deeply embedded in a setting.

b

a

Measuring Networks: Flow

Reachability

Indirect connections are what make networks systems. One actor can

reach another if there is a path in the graph connecting them.

b d a

b f

a c e

c

f

d e

components

Measuring Networks: Flow

Reachability

Reachability

then the two are reachable. If there is at least one path connecting

every pair of actors in the graph, the graph is connected and is called

a component.

a chain of relations.

Measuring Networks: Flow

Reachability

This example

contains many

components.

Measuring Networks: Flow

Distance & number of paths

1 step from 4

2 steps from 5

3 steps from 4

4 steps from 3

5 steps from 1

a

Measuring Networks: Flow

Distance & number of paths

Paths are the different routes one can take. Node-independent paths are

particularly important.

paths connecting a and

b.

b

independent paths

a

Measuring Networks: Flow

Distance & number of paths

Probability of transfer

by distance and number of paths, assume a constant pij of 0.6

1.2

1

10 paths

0.8

probability

5 paths

0.6

2 paths

0.4

1 path

0.2

0

2 3 4 5 6

Path distance

Reachability in Colorado Springs

(Sexual contact only) •High-risk actors over 4 years

•695 people represented

•Longest path is 17 steps

•Average distance is about 5 steps

•Average person is within 3 steps

of 75 other people

•137 people connected through 2

independent paths, core of 30

people connected through 4

independent paths

Measuring Networks: Flow

Centrality

resides in a network.

• For example, we can compare actors at the edge of the network to actors

at the center.

distinction between insiders and outsiders.

Measuring Networks: Flow

Centrality

captured through centrality.

which nodes are in the ‘center’ of the network. In practice, identifying

exactly what we mean by ‘center’ is somewhat complicated, but

substantively we often have reason to believe that people at the center

are very important.

“importance” in a network:

•Degree

•Closeness

•Betweenness

Measuring Networks: Flow

Centrality

the number of ties, and the actor with the most ties is the most

important:

C D d (ni ) X i X ij

j

Measuring Networks: Flow

Centrality

we look at the dispersion of centrality:

g

2

S D (CD (ni ) Cd ) / g

2

i 1

Or, using Freeman’s general formula for centralization (which ranges from 0 to 1):

CD

C

g

i 1 D (n ) CD (ni )

*

[( g 1)( g 2)]

Measuring Networks: Flow Degree Centralization Scores

Centrality

Freeman: .02

Variance: 3.9 Variance: 0.0

Variance: .17

Freeman: .07

Variance: .20

Measuring Networks: Flow

Centrality

important if he/she is relatively close to all other actors.

Closeness is based on the inverse of the distance of each actor to every other actor

in the network.

Closeness Centrality:

1

g

Cc (ni ) d (ni , n j )

j 1

Normalized Closeness Centrality

Measuring Networks: Flow

Centrality Closeness Centrality in the examples

C=1.0 C=0.0

C=0.36

C=0.28

Measuring Networks: Flow

Centrality

Betweenness Centrality:

Model based on communication flow: A person who lies on

communication paths can control communication flow, and is thus important.

Betweenness centrality counts the number of shortest paths between i and k

that actor j resides on.

b

a

C d e f g h

Measuring Networks: Flow

Centrality

Betweenness Centrality:

C B (ni ) g jk (ni ) / g jk

j k

gjk(ni) = the number that actor i is on.

'

B

Measuring Networks: Flow

Centrality

Betweenness Centrality:

Centralization: .31

Measuring Networks: Flow

Centrality

Actors that appear very

different when seen

individually, are

comparable in the global

network.

Measuring Networks: Flow

Time

Topology

- the shape, or form, of the network

- simple example: one actor cannot pass information to

another unless they are either directly or indirectly

connected

Time

- the timing of contacts matters

- simple example: an actor cannot pass information he has

not yet received.

Measuring Networks: Flow

Time

Timing in networks

dynamics,though a number of recent pieces are addressing this.

1) The structure itself evolves, in ways that will affect the topology an

thus flow.

Measuring Networks: Flow

Time Drug Relations, Colorado Springs, Year 1

Colorado Springs, over

5 years

Measuring Networks: Flow

Time Drug Relations, Colorado Springs, Year 2

Current year in red, past relations in gray

Measuring Networks: Flow

Time Drug Relations, Colorado Springs, Year 3

Current year in red, past relations in gray

Measuring Networks: Flow

Time Drug Relations, Colorado Springs, Year 4

Current year in red, past relations in gray

Measuring Networks: Flow

Time Drug Relations, Colorado Springs, Year 5

Current year in red, past relations in gray

Measuring Networks: Flow

Time

8-9

C E

2-5

A B

3-5

D F

Measuring Networks: Flow

Time

C E

A B

D F

While clearly important, this is not often handled well by current software.

Measuring Networks: Structure & Social Space

The second broad division for measuring networks steps back to

generalized features of the global network.

These factors almost always are of interest because of what they imply

about how goods move through the network, but have resulted in a distinct

line of methods and substantive research.

1) Basic structure of large-scale networks

2) Cohesive Peer Groups

3) Identifying Role positions (blockmodels)

Measuring Networks: Large-Scale Models

Small World Networks

the substantive point is that networks

are structured such that even when

most of our connections are local,

any pair of people can be connected

by a fairly small number of relational

steps.

Works on 2 parameters:

1) The Clustering Coefficient (c) =

average proportion of closed

triangles

2) The average distance (L)

separating nodes in the network

Measuring Networks: Large-Scale Models

Small World Networks

C=Large, L is Small =

SW Graphs

•Small average distance between nodes

Measuring Networks: Large-Scale Models

Small World Networks

network, a single random

connection will create a shortcut

that lowers L dramatically

world properties can occur in

graphs with a surprisingly small

number of shortcuts

unclear, but seem similar to a

random graphs where local

clusters are reduced to a single

point.

Measuring Networks: Large-Scale Models

Scale-Free Networks

settings, Barabási points out that the

distribution of network involvement

(degree) is highly and characteristically

skewed.

Measuring Networks: Large-Scale Models

Scale Free Networks

number of partners (degree)

Measuring Networks: Large-Scale Models

Scale Free Networks

number of partners (degree)

p(k ) ~ k

Measuring Networks: Large-Scale Models

Scale Free Networks

capacity of high-degree nodes:

Measuring Networks: Large-Scale Models

Scale Free Networks

degree nodes, as ‘hubs’ create shortcuts that carry network flow.

Measuring Networks: Large-Scale Models

Scale Free Networks

(Sexual contact only) •Network is approximately

scale-free, with = -1.3

depend on the hubs.

Measuring Networks: Large-Scale Models

Social Cohesion

in Social Networks: Node Connectivity and Conditional

Density." Sociological Methodology 31:305-59.

Cohesion and Embeddedness: A hierarchical Conception of

Social Groups” American Sociological Review 68:103-127

Walter W. Powell (2004) "Networks, Fields, and

Organizations: Scale, Topology and Cohesive

Embeddings." Computational and Mathematical

Organization Theory. 10:95-117

Collaboration Network: Disciplinary Cohesion from

1963 to 1999" American Sociological Review. 69:213-

238

Measuring Networks: Large-Scale Models

Social Cohesion

(a) A group’s structural cohesion is equal to the minimum number of actors who,

if removed from the group, would disconnect the group.

paths linking each pair of actors in the group.

Measuring Networks: Large-Scale Models

Social Cohesion

nodes are removed

0 1 2 3

Node Connectivity

Measuring Networks: Large-Scale Models

Social Cohesion

embeddedness, since cohesive sets nest inside of each other.

2

1 3

9

4 8 10

11

5 7 12

13

6 14

15

17

18 16

19

20

2

22

23

Measuring Networks: Large-Scale Models

Social Cohesion

3-Component (n=58)

Measuring Networks: Large-Scale Models

Social Cohesion

Largest BC: 247 Bicomponents

k > 4: 318

Max k: 12

Structural Cohesion

simultaneously gives

us a positional and

subgroup analysis.

Measuring Networks:

Cohesive Sub Groups

“significant social subgroups” – some smaller collection of nodes in

the graph that can be considered, at least in some senses, as a “unit”

based on the pattern, strength, or frequency of ties.

There are many ways to identify groups. They all insist on a group

being in a connected component, but other than that the variation is

wide.

Measuring Networks:

Cohesive Sub Groups

member of the graph is connected to every other member of the graph.

Cliques are collections of nodes where density = 1.0.

Properties of cliques:

• Density: 1.0

• Everyone connected to n-1 alters

• Distance between every pair is 1

• Ratio of within group ties to between

group ties is infinite

• All triads are transitive

Measuring Networks:

Cohesive Sub Groups

In practice, complete cliques are not very useful. They tend to overlap

heavily and are limited in their size.

relaxed the complete

connectivity requirement

(with varying degrees of

success). See the Moody

& White (2003) for a

discussion of these

attempts.

Measuring Networks:

Cohesive Sub Groups

1) Measures of fit

To identify a primary group, we need some measure of how clustered

the network is. Usually, this is a function of the number of ties that

fall within group to the number of ties that fall between group.

Once we have such an index, we need a method for searching through

the network to maximize the fit.

In addition to maximizing a group function such as (1) we can use the

relational distance directly, and look for clusters in the data. We next

go over two different styles of cluster analysis

Measuring Networks:

Cohesive Sub Groups

Segregation Index

(Freeman, L. C. 1972. "Segregation in Social Networks." Sociological Methods and

Research 6411-30.)

Theoretically, he argues, if a given attribute (group label) does not matter for

social relations, then relations should be distributed randomly with respect to the

attribute. Thus, the difference between the number of cross-group ties expected

by chance and the number observed measures segregation.

E( X ) X

Seg

E( X )

Measuring Networks:

Cohesive Sub Groups

attributes in this network: people with Blue eyes and Brown eyes

and people who are square or not (they must be hip).

Measuring Networks:

Cohesive Sub Groups

Segregation Index

Mixing Matrix:

Blue Brown

Blue 6 17

Brown 17 16

Seg = -0.25

Hip Square

Hip 20 3

Square 3 30

Seg = 0.78

Measuring Networks:

Cohesive Sub Groups

The segregation index is one metric used to identify groups. Others include:

a) The ratio of in-group to out-group ties (Negopy, UCINET Factions)

b) Maximizing the probability of in-group contact (CliqueFinder)

c) The Segregation Matrix Index (SMI)

d) The dyadic factor loadings for overlapping groups (akin to a latent

class model)

e) Minimize the within-group distance

Once a metric has been chosen, some algorithm is needed to search through

the graph to identify clusters. These algorithms range from very sophisticated

“graph-intelligent” algorithms, such as NEGOPY, to simple cluster analysis

of distance matrices.

In most cases, you have to pre-set the number of groups to use (the exceptions

are NEGOPY and CliqueFinder. Moody’s CROWDS algorithm also has

automatic stopping criteria, but you have to give it starting values.

Measuring Networks:

Cohesive Sub Groups

algorithms will give

different results.

NEGOPY results to the

RNM results. NEGOPY

returned one large group,

RNM found many smaller,

denser groups.

explore multiple solutions

and algorithms.

Measuring Networks: Gangon Prison Network

Cohesive Sub Groups

algorithms will give

different results.

Here, I compare

NEGOPY, FACTIONS

and RNM. Groups A and

B are identical, C is close.

F, E and D differ.

explore multiple solutions

and algorithms.

Measuring Networks:

Role Positions

Overview

•Social life can be described (at least in part) through social roles.

•To the extent that roles can be characterized by regular interaction

patterns, we can summarize roles through common relational patterns.

•Identifying these sets is the goal of block-model analyses.

•Background ideas for White, Boorman and Brieger. Social life as

interconnected system of roles

•Important feature: thinking of roles as connected in a role system =

social structure

White, Harrison C.; Boorman, Scott A., and Breiger, Ronald L. Social

Structure from Multiple Networks I. American Journal of Sociology.

1976; 81730-780.

•The key article describing the theoretical and technical elements of

block-modeling

Measuring Networks:

Role Positions

Elements of a Role:

people

occupant acts with respect to

Examples:

Parent - child, Teacher - student, Lover - lover, Friend - Friend,

Husband - Wife, etc.

‘logical’ types of roles, and then examines how they can be linked together.

Measuring Networks:

Role Positions

Start with some basic ideas of what a role is: An exchange of something (support,

ideas, commands, etc) between actors. Thus, we might represent a family as:

H W

C

C C

Romantic Love

Provides food for

Bickers with

(and there are, of course, many other relations inside a family!)

Measuring Networks:

Role Positions

The key idea, is that we can express a role through a relation (or set of relations)

and thus a social system by the inventory of roles. If roles equate to positions in

an exchange system, then we need only identify particular aspects of a position.

But what aspect?

Structural Equivalence

types of ties to the same people.

Measuring Networks:

Role Positions

Structural Equivalence

A single relation

Measuring Networks:

Role Positions

Structural Equivalence

Measuring Networks:

Role Positions

2) Measure the degree to which pairs of actors are equivalent

3) Develop a representation of the equivalencies

4) Assess the adequacy of the representation

Measuring Networks:

Role Positions

Structural Equivalence:

Two actors are equivalent if they have the same type of ties to the same people.

Measuring Networks:

Role Positions

Automorphic Equivalence:

Actors occupy indistinguishable structural locations in the network. That is,

that they are in isomorphic positions in the network.

all graph theoretic properties (I.e. degree, number of people reachable,

centrality, etc.)

Measuring Networks:

Role Positions

Automorphic Equivalence:

Measuring Networks:

Role Positions

Regular Equivalence:

Regular equivalence does not require actors to have identical

ties to identical actors or to be structurally indistinguishable.

from equivalent actors.

and for all actors, if i k, then there exists some actor l such

that j l and k is regularly equivalent to l.

Measuring Networks:

Role Positions

Regular Equivalence:

There may be multiple regular equivalence partitions in a network, and thus we tend

to want to find the maximal regular equivalence position, the one with the fewest

positions.

Measuring Networks:

Role Positions

Role or Local Equivalence:

While most equivalence measures focus on position within the full network, some

measures focus only on the patters within the local tie neighborhood. These have

been called ‘local role’ equivalence.

Note that:

Structurally equivalent actors are automorphically equivalent,

Automorphically equivalent actors are regularly equivalent.

Structurally equivalent and automorphically equivalent actors are role equivalent

In practice, we tend to ignore some of these distinctions, as they get blurred quickly

once we have to operationalize them in real-world graphs. It turns out that few

people are ever exactly equivalent, and thus we approximate the links between the

types.

In all cases, the procedure can work over multiple relations simultaneously.

a measure of similarity among nodes.

Measuring Networks:

Role Positions

Once you identify equivalent actors, block them in the matrix and reduce it, based on the number of ties

in the cell of interest. The key values are a zero block (no ties) and a one-block (all ties present):

1 2 3 4 5 6 1 2 3 4 5 6

1 . 1 1 0 1 1 0 0 0

1 1 0 0 0 0 0 0 0 0 0 0

2 1 0 0 1 0 0

2 1 . 0 0 1 1 0 0 0 0 0 0 0 0

3 1 0 1 0 1 0

1 0 . 1 0 0 1 1 1 1 0 0 0 0

3 4 0 1 0 1 0 1

1 0 1 . 0 0 1 1 1 1 0 0 0 0

5 0 0 1 0 0 0

0 1 0 0 . 1 0 0 0 0 1 1 1 1

4 6 0 0 0 1 0 0

0 1 0 0 1 . 0 0 0 0 1 1 1 1

0 0 1 1 0 0 . 0 0 0 0 0 0 0

5 0 0 1 1 0 0 0 . 0 0 0 0 0 0

0 0 1 1 0 0 0 0 . 0 0 0 0 0

0 0 1 1 0 0 0 0 0 . 0 0 0 0

0 0 0 0 1 1 0 0 0 0 . 0 0 0

6 0 0 0 0 1 1 0 0 0 0 0 . 0 0

0 0 0 0 1 1 0 0 0 0 0 0 . 0

0 0 0 0 1 1 0 0 0 0 0 0 0 .

Measuring Networks:

Role Positions

Once you partition the matrix, reduce it:

. 1 1 1 0 0 0 0 0 0 0 0 0 0 1 2 3

1 . 0 0 1 1 0 0 0 0 0 0 0 0

1 0 . 1 0 0 1 1 1 1 0 0 0 0

1 1 1 0

1 0 1 . 0 0 1 1 1 1 0 0 0 0 2 1 1 1

0 1 0 0 . 1 0 0 0 0 1 1 1 1 3 0 1 0

0 1 0 0 1 . 0 0 0 0 1 1 1 1

0 0 1 1 0 0 . 0 0 0 0 0 0 0

0 0 1 1 0 0 0 . 0 0 0 0 0 0

0 0 1 1 0 0 0 0 . 0 0 0 0 0

0 0 1 1 0 0 0 0 0 . 0 0 0 0

0 0 0 0 1 1 0 0 0 0 . 0 0 0

0 0 0 0 1 1 0 0 0 0 0 . 0 0

0 0 0 0 1 1 0 0 0 0 0 0 . 0 1 2

0 0 0 0 1 1 0 0 0 0 0 0 0 .

3

Regular equivalence

(here I placed a one in the image matrix if there were any ties in the ij block)

Measuring Networks:

Role Positions

Operationally, you have to measure the similarity between actors. If two actors

are structurally equivalent, then they will have identical ties to other people.

Consider the example again:

C and D match on all

1 2 3 4 5 6 12 other people, and

C D Match

1 . 1 1 1 0 0 0 0 0 0 0 0 0 0 1 1 1 are thus structurally

2 1 . 0 0 1 1 0 0 0 0 0 0 0 0 0 0 1 equivalent.

1 0 . 1 0 0 1 1 1 1 0 0 0 0 . 1 .

3

1 0 1 . 0 0 1 1 1 1 0 0 0 0 1 . .

0 1 0 0 . 1 0 0 0 0 1 1 1 1 0 0 1

4

0 1 0 0 1 . 0 0 0 0 1 1 1 1 0 0 1

0 0 1 1 0 0 . 0 0 0 0 0 0 0 1 1 1

5 0 0 1 1 0 0 0 . 0 0 0 0 0 0 1 1 1

0 0 1 1 0 0 0 0 . 0 0 0 0 0 1 1 1

0 0 1 1 0 0 0 0 0 . 0 0 0 0 1 1 1

0 0 0 0 1 1 0 0 0 0 . 0 0 0 0 0 1

6 0 0 0 0 1 1 0 0 0 0 0 . 0 0 0 0 1

0 0 0 0 1 1 0 0 0 0 0 0 . 0 0 0 1

0 0 0 0 1 1 0 0 0 0 0 0 0 . 0 0 1

Sum: 12

Measuring Networks:

Role Positions

If the model is going to be based on asymmetric or multiple relations, you simply stack the

various relations, usually including both “directions” of asymmetric relations:

Stacked

Romance

0 1 0 0 0 0 1 0 0 0

H W 1 0 0 0 0 1 0 0 0 0

0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0

0 0 1 1 1

0 0 1 1 1

Feeds

C 0 0 1 1 1

0 0 0 0 0

0 0 0 0 0

C C 0 0 1 1 1

0 0 0 0 0

0 0 0 0 0

0 0 0 0 0 0 0 0 0 0

Romantic Love 0 0 0 0 0 0 0 0 0 0

Provides food for 1 1 0 0 0

Bickers with Bicker 1 1 0 0 0

0 0 0 0 0 1 1 0 0 0

0 0 0 0 0 0 0 0 0 0

0 0 0 1 1 0 0 0 0 0

0 0 1 0 0 0 0 0 1 1

0 0 1 1 0 0 0 1 0 1

0 0 1 1 0

Measuring Networks:

Role Positions

The metric used to measure structural equivalence by White, Boorman and Brieger is

the correlation between each node’s set of ties. For the example, this would be:

1.00 -0.20 0.08 0.08 -0.19 -0.19 0.77 0.77 0.77 0.77 -0.26 -0.26 -0.26 -0.26

-0.20 1.00 -0.19 -0.19 0.08 0.08 -0.26 -0.26 -0.26 -0.26 0.77 0.77 0.77 0.77

0.08 -0.19 1.00 1.00 -1.00 -1.00 0.36 0.36 0.36 0.36 -0.45 -0.45 -0.45 -0.45

0.08 -0.19 1.00 1.00 -1.00 -1.00 0.36 0.36 0.36 0.36 -0.45 -0.45 -0.45 -0.45

-0.19 0.08 -1.00 -1.00 1.00 1.00 -0.45 -0.45 -0.45 -0.45 0.36 0.36 0.36 0.36

-0.19 0.08 -1.00 -1.00 1.00 1.00 -0.45 -0.45 -0.45 -0.45 0.36 0.36 0.36 0.36

0.77 -0.26 0.36 0.36 -0.45 -0.45 1.00 1.00 1.00 1.00 -0.20 -0.20 -0.20 -0.20

0.77 -0.26 0.36 0.36 -0.45 -0.45 1.00 1.00 1.00 1.00 -0.20 -0.20 -0.20 -0.20

0.77 -0.26 0.36 0.36 -0.45 -0.45 1.00 1.00 1.00 1.00 -0.20 -0.20 -0.20 -0.20

0.77 -0.26 0.36 0.36 -0.45 -0.45 1.00 1.00 1.00 1.00 -0.20 -0.20 -0.20 -0.20

-0.26 0.77 -0.45 -0.45 0.36 0.36 -0.20 -0.20 -0.20 -0.20 1.00 1.00 1.00 1.00

-0.26 0.77 -0.45 -0.45 0.36 0.36 -0.20 -0.20 -0.20 -0.20 1.00 1.00 1.00 1.00

-0.26 0.77 -0.45 -0.45 0.36 0.36 -0.20 -0.20 -0.20 -0.20 1.00 1.00 1.00 1.00

-0.26 0.77 -0.45 -0.45 0.36 0.36 -0.20 -0.20 -0.20 -0.20 1.00 1.00 1.00 1.00

Another common metric is the Euclidean distance between pairs of actors, which you

then use in a standard cluster analysis.

Measuring Networks:

Role Positions

Automorphic and Regular equivalence are more difficult to find, and require

iteratively searching over possible class assignments for sets that have the same

graph theoretic patterns. Usually start with a set of nodes defined as similar on

a number of network measures, then look within these classes for automorphic

equivalence classes.

A theoretically appealing method for finding structures that are very similar to

regular equivalence, role equivalence, uses the triad census. Each node is

involved in (n-1)(n-2)/2 triads, and occupies a particular position in each of

these triads.

Measuring Networks:

Role Positions

number of groups and determining blocks:

partition should be; in other words, when should one stop splitting

positions? Theory and the interpretability of the solution are the

primary consideration in deciding how many positions to produce.”

(W&F, p.378)

“In defining positions of actors, the ‘trick’ is to choose the point along

the series that gives a useful and interpretable partition of the actors

into equivalence classes.” (W&F p.383)

Measuring Networks:

Role Positions

An example:

Padgett, J. F. and Ansell, C. K.

Robust action and the rise of

the Medici, 1400-1434.

American Journal of

Sociology. 1993; 981259-

1319.

sense do not seem to exist, so

P&A turn to the pattern of

network relations among

families.

full 92 family network.

Modeling with Networks: Behaviors

There are two general approaches to modeling behaviors with network data:

1) Using network measures as variables to predict individual outcomes

2) Network autocorrelation / peer influence models

3) Dyad / QAP models of the similarity of actors and their joint network

position

Modeling with Networks: Behaviors

The simplest way to use network data in research is to include the network

measure as a covariate in a standard model:

•Functions of each person’s direct contacts attributes

•Such as: mean income of friends, proportion of friends who are

employed, racial heterogeneity of the friends,etc.

•Structural indicators:

•Such as: Centrality, dummies for group / role membership, etc.

These models are the only option for ego-network data,where information on

network alters is collected from a single respondent’s (ego’s) report.

They can be used from extractions of partial or complete data, but the error term

is – by definition – autocorrelated. Cases are not independent, but connected

through the social relations

Modeling with Networks: Behaviors

Explanations of Social Homogeneity." Sociological Methods and

Research 12:235-61.

———. 1998. A Structural Theory of Social Influence. Cambridge:

Cambridge.

Friedkin, N. E. and E. C. Johnsen. 1990. "Social Influence and

Opinions." Journal of Mathematical Sociology 15(193-205).

———. 1997. "Social Positions in Influence Networks." Social

Networks 19:209-22.

~

Y ()

αWY ()

Xb e

Where W is a direct function of the adjacency matrix, and a is the estimated

value of peer influence.

Modeling with Networks: Behaviors

There are two general ways to test for peer influence in an observed network.

The first estimates the parameters (a and b) of the peer influence model directly,

the second transforms the network into a dyadic model, predicting similarity

among actors.

See Doreian, Patrick. “Maximum likelihood methods for linear models Spatial

Effects and Spatial Disturbances Terms.” Sociological Methods and Research.

1982; 10243-269.

1871. American Sociological Review. 1991; 56716-729. (applied example)

~

Y ()

αWY ()

Xb e

Modeling with Networks: Behaviors

The basic model says that people’s opinions are a function of the opinions of

others and their characteristics.

~

Y ()

αWY ()

Xb e

WY = A simple vector which can be added to your model. That is, multiply

Y by a W matrix, and run the regression with WY as a new variable, and the

regression coefficient is an estimate of a.

This is what Doriean calls the QAD (“Quick and Dirty” estimate of peer

influence, and is equivalent (under certain assumptions) to adding the mean of

ego’s friends to the model.

Modeling with Networks: Behaviors

The problem with the above regression is that cases are, by definition, not

independent. In fact, WY is also known as the ‘network autocorrelation’

coefficient, since a ‘peer influence’ effect is an autocorrelation effect -- your value

is a function of the people you are connected to. In general, OLS is not the best

way to estimate this equation. That is, QAD = Quick and Dirty, and your results

will not be exact.

In practice, the QAD approach (perhaps combined with a GLS estimator) results in

empirical estimates that are “virtually indistinguishable” from MLE (Doreian et al,

1984)

The proper way to estimate the peer equation is to use maximum likelihood

estimates, and Doreian gives the formulas for this in his paper.

Assignment Procedure, to estimate the effects.

Modeling with Networks: Behaviors

Each person was asked to rank their satisfaction with the program, which is the dependent variable

in this analysis.

I constructed two W matrices, one from HELP the other from Best Friend. I treat relations as

symmetric and valued, such that:

1 if Aijt 1 or A jit 1

Wijt 2 if Aijt 1 and A jit 1

0 otherwise

Wij 1

j

Wii 0

I also include Race (white/Non-white, Gender and Cohort Year as exogenous variables in the model.

Modeling with Networks: Behaviors

Distribution of Satisfaction with the department.

Modeling with Networks: Behaviors

Parameter Estimates

Parameter Standardized

Variable Estimate Pr > |t| Estimate

FEMALE -1.07540 0.0142 -0.25455

NONWHITE -0.22087 0.5975 -0.05491

y00 0.93176 0.0798 0.21627

y99 -0.19375 0.7052 -0.04586

y98 -0.45912 0.4637 -0.08289

y97 0.60670 0.3060 0.11919

PEER_BF 0.23936 0.0002 0.42084

PEER_H 0.50668 0.0277 0.23321

Modeling with Networks: Behaviors

Dyad QAP models

Another way to get at peer influence is not through the level of Y, but through the

extent to which actors are similar with respect to Y.

k

matrix of similarities on attributes

Modeling with Networks: Behaviors

Dyad QAP models

1 0 1 1 1 0 0 0 0 0 0 1 0 0 1 0 0 0 1 0 0 1 1 0 0 1 1 0

2 1 0 1 0 0 0 1 0 0 1 0 0 0 1 0 0 0 1 0 0 0 0 1 1 0 0 1

3 1 1 0 0 1 0 1 0 0 0 0 0 1 0 1 1 1 0 1 0 0 1 0 0 1 1 0

4 1 0 0 0 1 0 0 0 0 0 0 1 0 0 1 1 1 0 1 0 1 0 0 0 1 1 0

5 0 0 1 1 0 1 0 1 0 1 1 0 0 0 0 0 0 1 0 1 0 0 0 1 0 0 1

6 0 0 0 0 1 0 0 1 1 0 0 1 1 0 0 1 1 0 0 1 0 0 1 0 0 0 1

7 0 1 1 0 0 0 0 0 0 0 0 1 1 0 1 0 1 0 1 0 1 1 0 0 0 1 0

8 0 0 0 0 1 1 0 0 1 0 0 1 1 0 1 1 0 0 1 0 1 1 0 0 1 0 0

9 0 0 0 0 0 1 0 1 0 1 1 0 0 1 0 0 0 0 0 1 0 0 1 1 0 0 0

Modeling with Networks: Behaviors

Dyad QAP models

Y Distance (Dij=abs(Yi-Yj)

0.32 .000 .277 .228 .181 .278 .298 .095 .307 .481

0.59 .277 .000 .049 .096 .555 .575 .182 .584 .758

0.54 .228 .049 .000 .047 .506 .526 .134 .535 .710

0.50 .181 .096 .047 .000 .459 .479 .087 .488 .663

0.04 .278 .555 .506 .459 .000 .020 .372 .029 .204

0.02 .298 .575 .526 .479 .020 .000 .392 .009 .184

0.41 .095 .182 .134 .087 .372 .392 .000 .401 .576

0.01 .307 .584 .535 .488 .029 .009 .401 .000 .175

-0.17 .481 .758 .710 .663 .204 .184 .576 .175 .000

Modeling with Networks: Behaviors

Dyad QAP models

The REG Procedure

Model: MODEL1

Dependent Variable: SIM

Analysis of Variance

Sum of Mean

Source DF Squares Square F Value Pr > F

Error 31 0.75591 0.02438

Corrected Total 35 1.66248

Dependent Mean 0.33161 Adj R-Sq 0.4866

Coeff Var 47.08929

Parameter Estimates

Parameter Standard

Variable DF Estimate Error t Value Pr > |t|

NOM 1 -0.17054 0.05963 -2.86 0.0075

SAMERCE 1 0.05387 0.05916 0.91 0.3696

SAMESEX 1 -0.06535 0.05365 -1.22 0.2324

NCOMFND 1 -0.16134 0.03862 -4.18 0.0002

Modeling with Networks: Behaviors

Dyad QAP models

Like the basic Peer influence model, cases in dyad models are not

independent. However, the non-independence now comes from two sources:

(1) the fact that the same person is represented in (n-1) dyads and (2) that i and

j are linked through relations.

Procedure. A non-parametric procedure for significance testing.

QAP runs the model of interest on the real data, then randomly permutes the

rows/cols of the data matrix and estimates the model again. In so doing, it

generates an empirical distribution of the coefficients, generating n levels of

the coefficients at ‘chance’ levels, which you then compare to the observed

data. This is implemented in UCINET for regression, and in DAMN for

logistic regression (J.L. Martin).

Modeling with Networks: Behaviors

Dyad QAP models

Procedure:

1. Calculate the observed association / model

2. for K iterations do:

a) randomly sort one of the matrices

b) recalculate the association / model

c) store the outcome

3. compare the observed outcome to the distribution of

outcomes created by the random permutations.

Modeling with Networks: Behaviors

Dyad QAP models

Modeling with Networks: Behaviors

Dyad QAP models

Modeling with Networks: Behaviors

Dyad QAP models

--------------------------------------------------------------------------------

# of permutations: 2000

Diagonal valid? NO

Random seed: 533

Dependent variable: EX_SIM

Expected values: c:\moody\Classes\soc884\examples\UCINET\mrqap-predicted

Independent variables: EX_NCOM

EX_ADJ

EX_SRCE

EX_SSEX

N = 72

Number of permutations performed: 1999

MODEL FIT

R-square Adj R-Sqr Probability # of Obs

-------- --------- ----------- -----------

0.545 0.525 0.029 72

REGRESSION COEFFICIENTS

Un-stdized Stdized Proportion Proportion

Independent Coefficient Coefficient Significance As Large As Small

----------- ----------- ----------- ------------ ----------- -----------

Intercept 0.519314 0.000000 0.012 0.012 0.988

EX_NCOM -0.161337 -0.541828 0.011 0.989 0.011

EX_ADJ -0.170539 -0.381186 0.020 0.980 0.020

EX_SRCE 0.053864 0.124551 0.236 0.236 0.764

EX_SSEX -0.065364 -0.151144 0.180 0.820 0.180

Note that the coefficient values will be identical, but the p values differ

Modeling with Networks: Behaviors

Dyad QAP models

model is whether observed associations between network structure and

behaviors is due to selection or influence.

distinguish the two.

fixed effect models (sometimes random effects models), where the

network features vary over time. This removes any stable characteristic

that might account for selection into a particular group.

Modeling with Networks: Structure

Dyad QAP models

While the most common way to use QAP models is to predict the

similarity on some substantive variable, one can just as easily predict the

presence/absence of a relation given attribute similarity.

This makes it possible to model the network itself, and ask questions about

how particular structures form.

Modeling with Networks: Structure

Exponential Random Graph Models (p*)

A long research tradition in statistics and random graph theory has lead to

parametric models of networks.

These are models of the entire graph, though as we will see they often work on

the dyads in the graph to be estimated.

of the class of all random graphs with the given known elements. For example,

all graphs with 5 nodes and 3 edges, or, put probabilistically, the probability of

observing the current graph given the conditions.

Modeling with Networks: Structure

Exponential Random Graph Models (p*)

The earliest approaches are based on simple random graph theory, but there’s

been a flurry of activity in the last 10 years or so.

Key references:

- Holland and Leinhardt (1981) JASA

- Frank and Strauss (1986) JASA

- Wasserman and Faust (1994) – Chap 15 & 16

- Wasserman and Pattison (1996)

Thanks to Mark Handcock for sharing some figures/slides about these models.

Modeling with Networks: Structure

Exponential Random Graph Models (p*)

exp{ z ( x)}

p( X x)

( )

Where:

is a vector of parameters (like regression coefficients)

z is a vector of network statistics, conditioning the graph

is a normalizing constant, to ensure the probabilities sum to 1.

Modeling with Networks: Structure

Exponential Random Graph Models (p*)

independent:

exp{ ij xij }

p( X x)

i, j

( )

Where:

ij = logit[P(Xij = 1)]

() =P[1 + exp(ij )]

Note this is one of the few cases where () can be written.

Modeling with Networks: Structure

Exponential Random Graph Models (p*)

graphs are equally likely. The homogeneous bernulli graph model:

exp { xij }

p( X x)

i, j

( )

Where:

() =[1 + exp()]g

Modeling with Networks: Structure

Exponential Random Graph Models (p*)

normalizing constant ends up being a problem. We need a way to express the

probability of the graph that doesn’t depend on that constant. It turns out we

can do this by conditioning on a ‘complement’ graph.

X i, j Sociomatri x with ij element forced to 0

X ic, j Sociomatri x with no tie between i and j

Modeling with Networks: Structure

Exponential Random Graph Models (p*)

p( X ij 1 | X ijc )

ij log c

[ z ( xij ) z ( xij )]

p( X ij 0 | X ij )

Note that we can now model the conditional probability of the graph,

as a function of a set of difference statistics, without reference to the

normalizing constant.

Modeling with Networks: Structure

Exponential Random Graph Models (p*)

Fitting p* models

found at:

http://kentucky.psych.uiuc.edu/pstar/index.html

Including:

A Practical Guide To Fitting p* Social Network Models

Via Logistic Regression

The site includes the PREPSTAR program for creating the difference variables

of interest.

Modeling with Networks: Structure

Exponential Random Graph Models (p*)

1 2 3 |4 5 6

1 1 1 1

2

2 1 1

3 1 1 1 3

6

x

4 1 1 4

5 1 5

6 1 1

We can model this network based on parameters for overall degree of Choice

(), Differential Choice Within Positions (W), Mutuality(), Differential

Mutuality Within Positions (W), and Transitivity (T).

Modeling with Networks: Structure

Exponential Random Graph Models (p*) 1

2

tie = l lw m mw tt / noint; 3

6

run;

4

L = Choice

LW = Within Group

M = Mutuality

MW = Mutual within Group

TT = Transitivity

Substantively, this graph is likely from the random class of graphs with similar mutuality and size

Modeling with Networks: Structure

Exponential Random Graph Models (p*)

One practical problem is that the resulting values are often quite correlated,

making estimation difficult. This is particularly difficult with “star”

parameters.

lw m mw tt

0.0007 <.0001 0.4034

0.0007 <.0001 0.8984

<.0001 <.0001 0.5375

0.4034 0.8984 0.5375

Modeling with Networks: Structure

Exponential Random Graph Models (p*)

1) Expansiveness and attractiveness parameters. = dummies for

each sender/receiver in the network

2) Degree distribution

3) Mutuality

4) Group membership (and all other parameters by group)

5) Transitivity / Intransitivity

6) K-in-stars, k-out-stars

7) Cyclicity

Modeling with Networks: Structure

Comparing to Random Graphs

A conceptual merge between random graph models and QAP models is to identify a

sample of graphs from the universe you are trying to model. So, instead of

estimating:

exp{ z ( x)}

p( X x)

( )

generate X empirically, then compare z(x) to see how likely a measure on x would

be given X. The difficulty, however, is generating X.

Modeling with Networks: Structure

Comparing to Random Graphs

The first option would be to generate all isomorphic graphs within a given

constraint.

This is possible for small graphs, but the number gets large fast. For a

network with 3 nodes, there are 16 possible directed graphs. For a

network with 4 nodes, there are 218, for 5 nodes 9608, for 6

nodes1,540,944, and so on…

So, the best approach is to sample from the universe, but, of course, if you

had the universe you wouldn’t need to sample from it. How do you

sample from a population you haven’t observed?

constraints.

Modeling with Networks: Structure

Comparing to Random Graphs

Example: Bearman, Peter S., James Moody and Katherine Stovel (2004) “Chains of Affection:

The Structure of Adolescent Romantic and Sexual Networks” American Journal of Sociology

110:44:92

Romantic Relations in Jefferson High

Modeling with Networks: Structure

Comparing to Random Graphs

Modeling with Networks: Structure

Comparing to Random Graphs

Simulated networks preserve observed degree, isolated dyad

distribution, and four-cycle constraint

Modeling with Networks: Structure

Comparing to Random Graphs

Simulated networks preserve observed degree, isolated dyad

distribution, and four-cycle constraint: 4 examples from the

simulated set

Social Network Software

UCINET

•The Standard network analysis program, runs in Windows

•Good for computing measures of network topography for single nets

•Input-Output of data is a special 2-file format, but is now able to read

PAJEK files directly.

•Not optimal for large networks

•Available from:

Analytic Technologies

Social Network Software

PAJEK

•Program for analyzing and plotting very large networks

•Intuitive windows interface

•Used for most of the real data plots in this presentation

•Started mainly a graphics program, but has expanded to a wide range of

analytic capabilities

•Can link to the R statistical package

•Free

•Available from:

Social Network Software

Cyram Netminer for Windows

•Newest Product, not yet widely used

•Price range depends on application

•Limited to smaller networks O(100)

http://www.netminer.com/NetMiner/home_01.jsp

Social Network Software

NetDraw

•Also very new, but by one of the best known names

in network analysis software.

•Free

•Limited to smaller networks O(100)

Social Network Software

NEGOPY

•Program designed to identify cohesive sub-groups in a network,

based on the relative density of ties.

•DOS based program, need to have data in arc-list format

•Moving the results back into an analysis program is difficult.

•Available from:

William D. Richards

http://www.sfu.ca/~richards/Pages/negopy.htm

•is a collection of IML and Macro programs that allow one to:

a) create network data structures from nomination data

b) import/export data to/from the other network programs

c) calculate measures of network pattern and composition

d) analyze network models

•Allows one to work with multiple, large networks

•Easy to move from creating measures to analyzing data

•Available by sending an email to:

Moody.77@sociology.osu.edu

- Brass 2004Transféré parJuan Camilo Zapata
- White-Network Analysis and Ethnographic ProblemsTransféré parJ. Rodríguez
- 1069 - Behind Acquisitions of Alliance Partners - Exploratory Learning and Network EmbeddednessTransféré parRheza Azaria Jo
- Baranyi Et Al 2011 Ecological IndicatorsTransféré parFábio Vieira
- tmp70D5Transféré parFrontiers
- WOM on Marketing.pdfTransféré parPrabu PPC
- Clinical Pharmacology & Therapeutics Volume Issue 2013 [Doi 10.1038_clpt.2013.176] Harrold, JM; Ramanathan, M; Mager, De -- Network-Based Approaches in Drug Discovery and Early DevelopmentTransféré parHeena Bhojwani
- net work analysis of a tourism destinationTransféré paralzghoul
- 1005.2122Transféré parsadiqaftab786
- 50 DynProbit (Montreal2006)Transféré parguido gentile
- Ok Network 5Transféré parttungl
- ACS07 Tania ExperimTransféré parcruzreyeslaura
- Company BackgroundTransféré parffdsafara
- Resources for Understanding Network EffectivenessTransféré parbbosker
- Media Disruption Exacerbates Revolutionary UnrestTransféré parventolinmono
- 1222518975Transféré parJorge Alberto Herrera Ramirez
- Modelo Canvas-uol HostTransféré parJorge Colares
- Strategic management:stakeholdersTransféré parAnca Dumitrescu
- Social Networking WebsiteTransféré parPatel Hitesh
- Survey of Privacy Issues in Social NetworkingTransféré parIJSTE
- Chapter 16-Operations ManagementTransféré parAcyslz
- TR-04-3Transféré parkanjai
- Palestine NetworkTransféré parapi-25885642

- Veshire Company 4Transféré parMuhammad Kamil
- Gakhov Time Series Forecasting With PythonTransféré parTariqul Islam Shuvo
- Marine GeodesyTransféré parDiyan Gitawanti Pratiwi
- 1 Stone 2004 Transformational vs Servant LeadershipTransféré parsheila
- introduction to psychology revision quizTransféré parapi-341123529
- CPCU 500 Look Inside the BookTransféré parJesus Kossonou
- Amazon ShortlistedTransféré parHarman Singh
- Creative Strategies in Viral AdvertisingTransféré parAlbu Reka-Abigel
- 919 HypothesisTransféré parJasper Villaflor Paulite
- 痛經針灸報告Transféré parOscarTadlock
- Standar-standar Akreditasi RS 2012 Yang Melibatkan RadiologyTransféré parRSUD dr. H. Soemarno Sosroatmodjo Kuala Kapuas
- Case Study2Transféré parJanice Balabat
- 2101733Transféré parprapin11116404
- reagan web quest lesson planTransféré parapi-315333838
- Convex Hull TutorialTransféré parNick Stauf
- Dissertation 4Transféré parGultekin Binte Azad
- işletmeTransféré parReza Razzaghi
- PresentationTransféré parArka Mitra
- Final Report 07Transféré parmittal2801
- Staying Current During Digital DisruptionTransféré parIDG_Enterprise
- Statistical Quality ControlTransféré parMd. Sayed Ibn Rashid
- Lesson Plan Box PlotsTransféré parJonathan Robinson
- PG_TR_070512_LCLTransféré parAndri Yoga
- Best Practices in Crisis CommunicationTransféré parErez Cohen
- Customer Loyalty ReviewTransféré parpravin_raj_16
- The Effect of Aphasia Upon Personality Traits, Depression and AnxietyTransféré parBrayan Mariscal
- Material Selection MethodologyTransféré parNatesh Rajan
- Official Attitudes Toward Post-Mao Chinese FilmTransféré parilamont
- Module 1 - Introduction to OrTransféré parsrk bhai
- MOHD ARIFTransféré parijw9pcjwc