a social influence analysis of perceived organizational support

More documents

Recommendations

Info

Analysis CHAPTER FOUR: ANALYSIS AND RESULTS Unlike most social science research in which the individual level of analysis is examined, the level of analysis in this study is the dyad. Therefore, each variable is represented as a matrix in which rows and columns represent actors and cells represent a relational state between actors (Raider & Krackhardt, 2001: 68). The fact that the level of analysis in this study is the dyad requires special analysis techniques because dyadic relations are not independent of one another, as are observations in most social science research (Raider & Krackhardt, 2001). As a result, there may be high levels of autocorrelation among the error terms in regular statistical models of this data. Accordingly, it is inappropriate to analyze data such as this using Ordinary Least Squares Regression, PLS or LISREL. Social networks researchers suggest utilization of a test that is robust against autocorrelation. Krackhardt (1988) suggests using Quadratic Assignment Procedure (QAP) regression to deal with autocorrelation problems associated with network data. QAP offers permutation-based tests of significance which are more resistant to autocorrelation problems than are ordinary least-squares regression models 3 (Raider & Krackhardt, 2001). Therefore, data analysis was conducted using UCINET 6 for Windows, a network analysis program developed by Borgatti, Everett, and Freeman (2002). Quadratic assignment procedure correlation analysis was utilized to generate a bivariate correlation matrix and quadratic assignment procedure regression (QAP) will be used to test the hypotheses. QAP 3 Currently there is some debate between Krackhardt and Wasserman regarding the extent to which QAP regression is resistant to autocorrelation problems (Butler; personal communication; Madhaven, personal communication). Wasserman advocates utilization of the p* model (Wasserman & Pattison, 1996). Somewhat like QAP, p* analysis consists of “generating a set of predictor variables from a network and then employing logistic regression analysis to fit a series of nested models in which the response variable is the presence or absence of a tie between each pair of actors” (Madhaven, Gnyawali, & He, 2004). I selected QAP regression for this study as it is more commonly utilized in the management literature at this time (e.g. Hinds, Carley, Krackhardt, & Wholey, 2000; Umphress et al., 2003). 65
correlation analysis has two steps. In the first step, Pearson’s correlation coefficients for corresponding cells in the two matrices are computed. Then the program permutes the rows and columns of one matrix and calculates the correlation between the matrices. This is repeated 1000 times; each correlation from step 1 is compared with step 2 in an effort to determine the number of times the correlation generated by random permutations is larger or equal to the step 1 correlation. To test all hypotheses, the POS similarity matrix was regressed on the social network matrices and control variable matrices using multiple regression QAP analysis. Multiple regression QAP analysis works in much the same way that QAP correlation analysis works. Initially, the program conducts standard multiple regression across corresponding cells of the POS matrix, the social networks matrices, and the control variable matrices (Borgatti et al., 2002). Next, the all rows and columns from the POS matrix are permuted randomly and the regression coefficient is recomputed. This step occurs 1,000 times in an effort to estimate the standard error. The results from this second step are in the form of R-squared values and regression coefficients. Each of the coefficients from step 2 is compared to the coefficient produced in step 1. Following this, the procedure computes the number of random permutations needed in step 2 to produce results as extreme as those produced in the first step. If a low proportion of similar results are found in step 2 when compared with step 1, a significant relationship is indicated. As mentioned previously, two different analyses were utilized in this dissertation, each with a different set of measures that represented different assumptions concerning the manner in which tie strength is measured in social networks research. Thus, a separate set of regression analyses was conducted for each of these analyses. The results of these analyses are presented in 66
Page 1 and 2:
TITLE PAGE A SOCIAL INFLUENCE ANALY
Page 3 and 4:
Copyright by Thomas J. Zagenczyk 20
Page 5 and 6:
with the expectation that reciproca
Page 7 and 8:
LIST OF TABLES Table 1: Hypothesize
Page 9 and 10:
PREFACE Thank you to my chair, Audr
Page 11 and 12:
epresentatives of the organization
Page 13 and 14:
Researchers also consider the conte
Page 15 and 16:
Exploring the effects of social inf
Page 17 and 18:
CHAPTER TWO: LITERATURE REVIEW Orga
Page 19 and 20:
Levinson’s work explained the lar
Page 21 and 22:
Eisenberger and colleagues (1986) a
Page 23 and 24: In subsequent studies, researchers
Page 25 and 26: the organization is based on percei
Page 27 and 28: esearch exploring the antecedents o
Page 29 and 30: a manager with high formal status s
Page 31 and 32: to carry out tasks on their own, an
Page 33 and 34: occurs because the norm of reciproc
Page 35 and 36: on their daily tasks. Examining thi
Page 37 and 38: coworkers are able to provide infor
Page 39 and 40: when forming perceptions and evalua
Page 41 and 42: identifies with and interacts with
Page 43 and 44: Miles, and Grover (2003) showed tha
Page 45 and 46: instance, a researcher may find a p
Page 47 and 48: likely to discuss sensitive issues
Page 49 and 50: it does not require that the role m
Page 51 and 52: among tellers. In one branch, no se
Page 53 and 54: eside outside of the focal individu
Page 55 and 56: To test the effects of social influ
Page 57 and 58: disseminating organizational inform
Page 59 and 60: less positive. Thus, the POS of the
Page 61 and 62: ties (and other combinations of str
Page 63 and 64: Research Setting Hypotheses were te
Page 65 and 66: network data concerning recurring i
Page 67 and 68: I explored these different requirem
Page 69 and 70: or not the employees with whom they
Page 71 and 72: frequent contact with person j, or
Page 73: employee held a leadership position
Page 77 and 78: Organizational Support was .84, whi
Page 79 and 80: Table 5: Descriptive Statistics for
Page 81 and 82: Correlations. Two different types o
Page 83 and 84: Control Variables Table 7: Pearson
Page 85 and 86: Control Variables Table 9: Pearson
Page 87 and 88: Results of Hypotheses Tests using Q
Page 89 and 90: positions had perceptions of organi
Page 91 and 92: CHAPTER FIVE: DISCUSSION AND CONCLU
Page 93 and 94: elationship may not have been consi
Page 95 and 96: social learning theory (Bandura, 19
Page 97 and 98: to provide false information is les
Page 99 and 100: Fourth, when studying social influe
Page 101 and 102: theoretical foundation of POS is so
Page 103 and 104: consistent with past research (Ibar
Page 105 and 106: employees who were widely regarded
Page 107 and 108: etween employees, only strong frien
Page 109 and 110: of structure and behavior. Academy
Page 111 and 112: organizational support. In J. A-M.
Page 113 and 114: Hammer, M. (1985). Implications of
Page 115 and 116: Kohlberg, L. (1963). Moral developm
Page 117 and 118: Mehra, A., Kilduff, M., & Brass, D.
Page 119 and 120: Rogers, E.M. (1995). Diffusion of i
Page 121 and 122: networks: An introduction to Markov
Page 123 and 124: Name of employee Yes Yes No No Name
show all

a social influence analysis of perceived organizational support

Create successful ePaper yourself

Delete template?

Save as template?