Mar 28, 2015
Networks and job satisfaction
2
Can network ties increase job satisfaction?And if so, how?
Affective ties (trust, friendship)
Instrumental ties (communication)
General Mechanisms
3
Centrality effect Contagion effect
4
Overview hypotheses Popularity/Centrality effects
Advice (weak, instrumental tie) - Hyp. 1Trust (strong, affective tie) - Hyp. 2
Contagion effectsInformation contagion - Hyp. 3Affective contagion - Hyp. 4
Centrality mechanism
5
Centrality effectSocial Capital Makes
Us Feel Good
“social networks serve as a social resource which affects job satisfaction through the provision of social support” (Hurlbert, 1991)
=> Effects of FRIENDSHIP: Baldwin et al. (1997) two ways: 1) important resource for psychosocial support (buffer work problems)2) important for access to crucial resources (i.c., information)
Centrality advice
6
Centrality effect
Instrumental support:Communication network important for
access to crucial resources (i.e. information)
cf. Performance literature (e.g. Sparrowe et al., 2001)
Hyp 1: The higher the number of interpersonal advice ties of a focal actor (outdegree centrality), the more likely it is that the job satisfaction of the focal actor will increase over time
Centrality personal trust
7
Centrality effect
Affective support:Affective (friendship/personal trust)
ties: buffer work problems (Baldwin et al., 1997; Morrison, 2004)-Trust facilitates collaboration and
exchange of information
Hyp 2: The higher the number of interpersonal trust choices received by a focal actor (indegree centrality), the more likely it is that the job satisfaction of the focal actor will increase over time
Contagion Mechanisms
8
Contagion effect
Ties as transmitters of - information about the job- feelings, moods
Contagion advice
9
Contagion effectContagion based social capital:-Social information process theory(Salancik and Pfeffer, 1977; Festinger, 1954):
Evaluation of own situation based on others perception of situation, etc.
=> Employee’s vision about their job is based on information from his/her colleagues…
Contagion advice
10
Contagion effect
Hyp 3: The higher (lower) the mean job satisfaction of those colleagues whom a focal actor asks advice from, the more likely it will be that the job satisfaction of the focal actor will be high (low).
Contagion based social capital:-Social information process theory(Salancik and Pfeffer, 1977; Festinger, 1954):
Evaluation of own situation based on others perception of situation, etc.
=> Employee’s vision about their job is based on information from his/her colleagues…
Contagion personal trust
11
Contagion effectPeople’s mood is influenced by others they are emotionally connected with
-Mood linkage theory: unconscious mimicking
=> emotional contagion (cf. Cote, 2005)
Contagion personal trust
12
Contagion effect
Hyp 4: The higher (lower) the mean job satisfaction of those colleagues whom a focal actor trusts, the more likely it will be that the job satisfaction of the focal actor will be high (low).
People’s mood is influenced by others they are emotionally connected with
-Mood linkage theory: unconscious mimicking
=> emotional contagion (cf. Cote, 2005)
13
Data and method DATA• 30 teams in 2 knowledge-intensive organizations• Teams between 5 and 12 members
Job satisfactionDifferent items: income, job security, autonomy, etc...
Background characteristics: Age, gender, hierarchy, size of team,...
14
Data and method
Missing• Missing data imputed with existing data (from other actors). •Missing network data randomly imputed by given density Done multiple times Reports average of different imputations (and have a look at variation)
15
Data and method Method Network centrality
Indegree (trust received/advice received)Outdegree (trust in many others/advice in others)
Method: Regression and spatial regression Y=b*X + rho*W*Y + e
Results ADVICE
16
Centrality effect b s.e. t pIntercept 7.57 0.66 11.43 ***
hierarch 0.68 0.31 2.23 *sizetcon -0.10 0.05 -2.14 *gender -0.18 0.22 -0.83age 0.02 0.01 1.35ind 0.02 0.05 0.32outd 0.04 0.04 0.88
Global Moran's I for regression residualsMoran I statistic standard deviate = 3.1869 - 4.1075, p-value = 0.0007191 – 0.000002alternative hypothesis: greatersample estimates:Observed Moran's I Expectation Variance 0.146031106 -0.011125034 0.001463894 0.109657376 -0.011566099 0.001446935 0.136249899 -0.011280247 0.001456123
b s.e. t pIntercept 7.37 0.64 11.58 ***
hierarch 0.74 0.29 2.53 *sizetcon -0.07 0.05 -1.44gender -0.19 0.21 -0.89age 0.01 0.01 1.03ind 0.01 0.05 0.24outd -0.31 0.14 -2.22 °/*
Results ADVICE
17
Rho: 0.045722 - 0.052735 LR test value: 7.7238 - 11.708 p-value: 0.0054498 - 0.00062222
Asymptotic standard error: 0.017986 - 0.019132p-value: 0.0033672 - 0.016856 */***
Contagion effect
b s.e. t PIntercept 7.882 0.623 12.654 ***hierarch 0.670 0.276 2.428 *sizetcon -0.141 0.032 -4.397 ***gender -0.141 0.206 -0.687age 0.006 0.011 0.592ind 0.131 0.061 2.137 °/*outd 0.145 0.035 4.199 ***
Results TRUST
18
Centrality effect
Global Moran's I for regression residualsMoran I statistic standard deviate = 0.8797-1.5514, p-value = 0.1895-0.06041alternative hypothesis: greater sample estimates:Observed Moran's I Expectation Variance 0.061405456 -0.013524389 0.002332866 0.042599368 -0.013784516 0.002385922 0.029185915 -0.013648903 0.002370762
Results TRUST
19
b s.e. t pIntercept 7.76 0.61 12.69 ***hierarch 0.68 0.27 2.52 *sizetcon -0.12 0.03 -3.75 **gender -0.13 0.2 -0.64age 0.01 0.01 0.62ind 0.1 0.06 1.56outd -0.19 0.21 -0.9
Rho: 0.041156 - 0.050424 LR test value: 2.7963 - 4.8523 p-value: 0.094482 - 0.027609
Asymptotic standard error: 0.027771 - 0.028047p-value: 0.14227 – 0.070874
Contagion effect
20
Summary results
Mechanism Hyp. Theory• Popularity/centrality
– Advice Social capital/knowledge transfer– Trust Affective social support
• Contagion– Advice Social information process theory– Trust Affect contagion/mood
21
Limitations
Causality?longitudinal analysis (SIENA) on other datasets do support influence rather than selection (Agneessens and Wittek, 2008)
Job satisfactiondistinction between intrinsic and extrinsic aspects? (Flap and Volker, 2001; Agneessens and Wittek = longitudinal)
More complex network effects?Types of ego-networks?