Yannick Kurvers, 8008701
3c)
It is not correct to calculate indegree in the same way as outdegree by simply
substituting alter for ego, because indegree measures how often others (alter)
consider a specific student (ego) a friend. This is different from outdegree, which
measures how many friends ego has named itself.
Social relationships are often asymmetric, meaning ego may see a particular
alter as a friend, while this need not be mutual. Because the variable friend in the
dataset is reported from ego's perspective, it does not provide direct information on
how others (alters) view ego.
Therefore, to calculate indegree correctly, it is necessary to reconstruct others'
perspective on ego by organising the dataset in a different way or performing
additional calculations.
4a)
Regression table 1.
Variable Coefficient (SE)
Friend 0.4265**
(0.0234)
Befriend 0.1696***
(0.0234)
Indegree 0.0096*
(0.0039)
Outdegree -0.0217***
(0.0029)
Indegree_alter -0.0032
(0.0027)
Outdegree_alter -0.0125
(0.0033)
Money_ego -0.0086
(0.0042)
Money_alter -0.0002
(0.0004)
Sex_ego -0.0095
(0.00037)
Sex_alter -0.1210
(0.0425)
Gender_similarity 0.0156
(0.0136)
_Cons 0.0054
(0.0352)
Number of observations 3732
R-squared 0.2203
*p < 0.05, **p < 0.01, *** p < 0.001
In the first regression, we see that both friend and befriend are strongly
significant with coefficients of 0.4265 and 0.1696 (p < 0.001), respectively. This
means that trust (trust) of ego in alter is positively affected when ego nominates
alter as friend, and also when alter nominates ego as friend. This confirms that
mutual friendship plays an important role in trust. Indegree (how often ego is
named as a friend) has a small but significant positive effect (p < 0.05), while
outdegree (how many friends ego names) has a negative effect (p < 0.001).
Interestingly, alter's network positions (indegree_alter and outdegree_alter) have
no significant effect on trust. Control variables such as pocket money and gender
also show little effect, although gender_similarity is almost significant (p = 0.055).
The model explains 22% of the variance in trust, which is fairly high for this type
of analysis.
1
, Yannick Kurvers, 8008701
4b)
Regression table 2.
Variable coefficient
Tertii_ego -0.0027
(0.0028)
Tertii_alter 0.0312***
(0.0040)
Money_ego 0.0000
(0.0000)
Money_alter -0.0001
(0.0006)
Sex_ego -0.0016
(0.0141)
Sex_alter -0.1008***
(0.0144)
Gender_similarity -0.1575***
(0.0139)
_cons -0.0464
(0.0351)
Number of observations 3732
R-squared 0.0705
*p < 0.05, **p < 0.01, *** p < 0.001
The second regression included the variables tertii_ego (how many third parties
were named as friends by ego) and tertii_alter (how many third parties were named
by alter). Interestingly, only tertii_alter is significant (p < 0.001) with a coefficient of
0.0312, suggesting that the more friends alter has, the greater ego's trust in alter. In
contrast, tertii_ego has no significant effect. This highlights that not only the direct
relationship, but also the broader social position of alter is important for trust. Control
variables show interesting trends: gender_similarity has a strong negative effect (-
0.1575, p < 0.001), suggesting that trust is lower when ego and alter are of the same
gender. The model explains 7% of the variance, which is quite low but appropriate
given the limited number of independent variables.
4c)
Regression table 3.
Coefficient
Variable
(SE)
tertii_indegree_e 0.0155***
go (0.0041)
tertii_indegree_al
0.0065* (0.0028)
ter
money_ego -0.0001 (0.0003)
money_alter 0.0002 (0.0002)
sex_ego 0.0126 (0.0144)
-0.1267***
sex_alter
(0.0137)
0.1568***
gender_similarity
(0.0136)
_cons -0.0214 (0.0353)
N: 3732
R-squared: 0.0614
2
3c)
It is not correct to calculate indegree in the same way as outdegree by simply
substituting alter for ego, because indegree measures how often others (alter)
consider a specific student (ego) a friend. This is different from outdegree, which
measures how many friends ego has named itself.
Social relationships are often asymmetric, meaning ego may see a particular
alter as a friend, while this need not be mutual. Because the variable friend in the
dataset is reported from ego's perspective, it does not provide direct information on
how others (alters) view ego.
Therefore, to calculate indegree correctly, it is necessary to reconstruct others'
perspective on ego by organising the dataset in a different way or performing
additional calculations.
4a)
Regression table 1.
Variable Coefficient (SE)
Friend 0.4265**
(0.0234)
Befriend 0.1696***
(0.0234)
Indegree 0.0096*
(0.0039)
Outdegree -0.0217***
(0.0029)
Indegree_alter -0.0032
(0.0027)
Outdegree_alter -0.0125
(0.0033)
Money_ego -0.0086
(0.0042)
Money_alter -0.0002
(0.0004)
Sex_ego -0.0095
(0.00037)
Sex_alter -0.1210
(0.0425)
Gender_similarity 0.0156
(0.0136)
_Cons 0.0054
(0.0352)
Number of observations 3732
R-squared 0.2203
*p < 0.05, **p < 0.01, *** p < 0.001
In the first regression, we see that both friend and befriend are strongly
significant with coefficients of 0.4265 and 0.1696 (p < 0.001), respectively. This
means that trust (trust) of ego in alter is positively affected when ego nominates
alter as friend, and also when alter nominates ego as friend. This confirms that
mutual friendship plays an important role in trust. Indegree (how often ego is
named as a friend) has a small but significant positive effect (p < 0.05), while
outdegree (how many friends ego names) has a negative effect (p < 0.001).
Interestingly, alter's network positions (indegree_alter and outdegree_alter) have
no significant effect on trust. Control variables such as pocket money and gender
also show little effect, although gender_similarity is almost significant (p = 0.055).
The model explains 22% of the variance in trust, which is fairly high for this type
of analysis.
1
, Yannick Kurvers, 8008701
4b)
Regression table 2.
Variable coefficient
Tertii_ego -0.0027
(0.0028)
Tertii_alter 0.0312***
(0.0040)
Money_ego 0.0000
(0.0000)
Money_alter -0.0001
(0.0006)
Sex_ego -0.0016
(0.0141)
Sex_alter -0.1008***
(0.0144)
Gender_similarity -0.1575***
(0.0139)
_cons -0.0464
(0.0351)
Number of observations 3732
R-squared 0.0705
*p < 0.05, **p < 0.01, *** p < 0.001
The second regression included the variables tertii_ego (how many third parties
were named as friends by ego) and tertii_alter (how many third parties were named
by alter). Interestingly, only tertii_alter is significant (p < 0.001) with a coefficient of
0.0312, suggesting that the more friends alter has, the greater ego's trust in alter. In
contrast, tertii_ego has no significant effect. This highlights that not only the direct
relationship, but also the broader social position of alter is important for trust. Control
variables show interesting trends: gender_similarity has a strong negative effect (-
0.1575, p < 0.001), suggesting that trust is lower when ego and alter are of the same
gender. The model explains 7% of the variance, which is quite low but appropriate
given the limited number of independent variables.
4c)
Regression table 3.
Coefficient
Variable
(SE)
tertii_indegree_e 0.0155***
go (0.0041)
tertii_indegree_al
0.0065* (0.0028)
ter
money_ego -0.0001 (0.0003)
money_alter 0.0002 (0.0002)
sex_ego 0.0126 (0.0144)
-0.1267***
sex_alter
(0.0137)
0.1568***
gender_similarity
(0.0136)
_cons -0.0214 (0.0353)
N: 3732
R-squared: 0.0614
2
