9 The relation between control and decisions
9.1 Do they go together?
thus not limited only to the relation with her partner, but includes her children as well.
To sum up, the analysis shows that it is not problematic to model the relation between division of labor and power outcomes as was done in the multinomial logistic regression models.
Labor and power are not interrelated, which means that the coefficients are not distorted by an endogeneity bias. This is confirmed in the bivariate probit models.
The following chapter will investigate in more detail another interrelation : the interrelation between control over the income and decision-making. This will be done in two steps. First, the power outcome of one model will be introduced into the other model as an explanatory variable, and vice versa. Second, bivariate probit models will test the correlation between both outcomes. In addition, in the case of a correlation between the error terms, the bivariate probit model allows a comparison of the effects of the power bases between the power outcomes. Are the power bases related to a larger extent to control over the income or to decision-making?
This question will be answered in the following sections.
Control over income ref(Joint Pool) Women Men Separat
I decide 0.801 1.367
(0.33) (0.41) Partner decides 1.356 0.625 (0.40) (0.28) I control
I decide 2.996* 3.306*
(1.38) (1.57) Partner decides 0.992 1.647 (0.80) (1.26) Partner controls
I decide 1.950 0.700
(1.39) (0.46) Partner decides 3.268* 4.864***
(1.56) (2.01)
N 1579 1579
Pseudo-R2 0.25 0.23
Note: Multinomial logistic regression model; Expo-nentiated coefficients (RRR); Clustered standard er-rors in parentheses; Unweighted; Dependent variable:
control over income; Basecategory: joint pool; Only selected covariats presented; Male and female respon-dents; *p <0.05, **p <0.01, ***p <0.001; Data source: SOEP 2005 and 2008
Table 20: First model for control over the income
Control over income ref(Separat) Women Men Pool
I decide 1.248 0.731
(0.52) (0.22) Partner decides 0.737 1.601 (0.22) (0.72) I control
I decide 3.741* 2.418
(2.04) (1.17) Partner decides 0.731 2.637 (0.61) (2.22) Partner controls
I decide 2.434 0.512
(1.84) (0.35) Partner decides 2.410 7.786***
(1.22) (4.02)
N 1579 1579
Pseudo-R2 0.25 0.23
Note: Multinomial logistic regression model; Expo-nentiated coefficients (RRR); Clustered standard er-rors in parentheses; Unweighted; Dependent variable:
control over income; Basecategory: seperate system;
Only selected covariats presented; Male and female respondents; *p < 0.05, **p < 0.01, ***p <
0.001; Data source: SOEP 2005 and 2008
Table 21: Second model for control over the income
Table 20, 21, 22, and 23 show the odds ratios of the multinomial logistic regressions for the power outcomes only. The first model estimates the chance of the three different outcome categories of control over the income, with the joint pool as the reference category (Table 20).
Here decision-making is the explanatory variable. In the second model, the separate system is the reference category for control (Table 21). The third model estimates the chance of the two outcome categories for decision-making, with joint decision-making as the reference cat-egory (Table 22). In the fourth model, the separate system is the reference catcat-egory for control (Table 23). Covariates are the same as in the previous models. Again, models are run sepa-rately for the male and the female samples. The gender-specific explanatory variables fit the gender-specific dependent variable: her perception of control over the finances is regressed on her perception of decision-making, and vice versa. The same is done with the male sample.
Note that my previous study did the analysis only for the female respondents. This study ex-tends previous research through allowing for a comparison between the male and the female partners’ perceptions of their power allocations.
Decision-making ref(Joint decision) Women Men I decide
Separat control 0.924 1.076 (0.35) (0.36)
I control 3.906** 3.118*
(1.84) (1.50) Partner controls 1.469 0.715 (1.10) (0.50) Partner decides
Separat control 1.275 0.592 (0.42) (0.25)
I control 0.712 1.341
(0.49) (1.15) Partner controls 2.119 5.714***
(1.06) (2.40)
N 1579 1579
Pseudo-R2 0.22 0.25
Note: Multinomial logistic regression model; Exponen-tiated coefficients (RRR); Clustered standard errors in parentheses; Unweighted; Dependent variable: decision-making; Basecategory: joint decision; Only selected co-variats presented; Male and female respondents; *p <
0.05, **p <0.01, ***p <0.001; Data source: SOEP 2005 and 2008
Table 22: First model for decision-making
Decision-making
ref(Both) Women Men
I decide
Pool 1.083 0.929
(0.41) (0.31) I control 4.229** 2.897*
(2.16) (1.52) Partner controls 1.591 0.664 (1.23) (0.50) Partner decides
Pool 0.785 1.689
(0.26) (0.71)
I control 0.559 2.264
(0.42) (2.12) Partner controls 1.663 9.650***
(0.89) (4.79)
N 1579 1579
Pseudo-R2 0.22 0.25
Note: Multinomial logistic regression model; Exponen-tiated coefficients (RRR); Clustered standard errors in parentheses; Unweighted; Dependent variable: decision-making; Basecategory: joint decision; Only selected co-variats presented; Male and female respondents; *p <
0.05, **p <0.01, ***p <0.001; Data source: SOEP 2005 and 2008
Table 23: Second model for decision-making
First of all, the results show that the power outcomes are related to each other. Comparing the male and the female samples, we see that the relation between control and decision-making is significant for both partners. For the female partners’ power outcomes, the effects are even more significant (Tables 22, 23, and 20). Hence, my first conclusion that the relation between the power outcomes is stronger for men than for women needs to be rejected here. Actu-ally, the opposite seems to be the case. The relation between control and decision-making is stronger for the female than for the male partners.
Interestingly, the men in particular perceive the power of their partners. The coefficients for the partners’ control and decision-making are highly significant in the male sample in all of the models. With values between around 5 and 10, the relative risk ratios indicate a strong relation between her power outcomes perceived by men. Note that the number of observations in these categories is sufficient for an estimation of the coefficients. One explanation could be that since male partners’ gender identities still conform to the male breadwinner model, he might react much more sensitively to a situation where she has power. As a reaction, he perceives her power outcomes as more strongly related to each other than his power outcomes – as if to state that “she wears the breeches”. This finding is in line with the previous result that male partners are more sensitive to financial power – at least to the allocation of power bases.
Although the coefficients are not significant, all the models indicate that the male partners’
control is related not only to his, but also to his partner’s decision-making. Hence, whereas he is unlikely to make the financial decisions if she controls the income, she is likely to have power in one dimension while he has power in the other dimension. Not only is the relation between both of these power outcomes stronger for the female than for the male partners, she is also more likely to make the decisions if he controls the income than vice versa.
The previous chapter discussed if the joint pool is an “equal” gender arrangement or whether one of the partners still has a power advantage. Since pooling the money can be considered a rather traditional arrangement used most often by married couples with a conventional distri-bution of resources and labor, we might expect that men are the decision-makers if the income is pooled. Women might then be decision-makers in more progressive forms of power allo-cation, i.e. if the partners separate their incomes. Surprisingly, we observe the opposite. The separate system is associated with the male partner’s power and the joint pool is related to the female partners’ power – in both power dimensions. Note, however, that the coefficients are not significant.
This finding is rather unexpected since we know that partners separate their earnings if she has a greater share of income. Hence we could assume that since she has more money, she is also the decision-maker. The same is to be expected for the joint pool. Since she has less money if the partners pool their incomes, she is less likely to be the decision-maker. Although the results are surprising, they are supported by a study based on British data. For British couples, Vogler et al. (2008) also found that the male partner is more likely to make the decisions if the partners separate their incomes. In order to explain the relation between financial power outcomes, future research will have to allow a much more detailed analysis of partners’ power allocations.
Control over the income and financial decision-making are interrelated power outcomes. Their correlation can be tested further by estimating a bivariate probit model. This model also allows to test whether the power bases are more strongly related to one of the power outcomes. Which do the power bases explain better: control over the income or financial decision-making?