
Distribution based on simulation
  

Variable

Mean

Median

Standard deviation

5^{th}percentile

95^{th}percentile

Sample values

p value


Change commitment (4 items)
       
AD
_{
M(J)
} mean

1.15

1.15

0.05

1.06

1.23

0.72

0.000

r
_{
WG(J)
} mean

0.20

0.19

0.08

0.07

0.33

0.82

0.000

Change Efficacy (5 items)
       
AD
_{
M(J)
} mean

1.15

1.15

0.05

1.07

1.22

0.76

0.000

r
_{
WG(J)
} mean

0.21

0.20

0.08

0.08

0.34

0.82

0.000


Note: For each of the four statistics (AD
_{
M(J)
} mean and r
_{
WG(J)
} mean for each scale), we obtained an empirical distribution based on 100,000 simulated random samples. The distributions are summarized in terms of their means, medians, standard deviations, and 5^{th} and 95^{th} percentiles. Consider the r
_{
WG(J)
} mean for the fouritem Change Commitment scale. Its sample value was 0.82. Simulations under the uniform (rectangular) null distribution indicate that the empirical distribution has a mean of 0.20 and a 95^{th} percentile of 0.33, which is much lower than the observed value of 0.82. Thus, the corresponding pvalue is 0.00. Therefore, the conclusion is that based on the mean r
_{
WG(J)
} we reject the null hypothesis that there is no agreement in the ensemble of 10 INGOs.