Gintarė Biterytė (UČO: 477937) Ekaterina Volevach (UČO: 460236) Assignment 3 In this assignment we will perform multivariate analysis on ‘‘European Values Study 2008: Integrated Dataset‘‘ in order to answer the question whether married people have more traditional attitude towards gender roles that people who are widowed or divorced. The dependent variable is index of attitude towards gender roles. You are the only ones who tell us what the dependent variable is! Good The independent variables include marital status of respondent (married, divorced, widowed), having kids (yes, no), believing in god (yes, no), age, and interaction between education level (lower, middle, upper) and gender (male, female). The results are presented in table 1. In the model M1, the only independent variable is marital status of a respondent. As it is showed in the table 1, if a respondent is divorced is divorced, he/she will have 75,9% more egalitarian attitude towards gender roles that a person who is divorced. However, the situation is different if person is widowed. Widowed persons will have 44,3% more traditional attitude towards gender roles than married persons. Model M2 shows how much the attitude changes if a married/divorced/widowed person have kids. As you can see from the table 1, divorced person will have even more egalitarian attitude compared to married person if he/she has kids (comparing with married person who also has kids). For a widowed person having kids means more traditional attitude towards gender roles compared to a married person, but as shown in table, percentage in the model M2 is lower than in model M1. But if comparing R-squared stays the same in both models (0,0140) OK, but it is not the percentage, it is just plus minus. So in your index, widowed person will have +0,443 higher values than married. In the model M3, except from marital status and having kids, respondent’s faith (believing in god) is added. Divorced person still has more egalitarian attitude than married person. However, the percentage is slowly decreasing (from 77.9 % to 62 %). Interesting is the decrease of percentage among widowed people, it is not significant (from 39,9 % to 36,3%), but it is not that we expected from people who believe in god. Believing in god makes divorced people to have less egalitarian attitude towards gender roles than married people and at the same time it makes widowed people to have less traditional attitude towards gender roles than married people. However, this model is better than model M2 (R – squared is 0,0337). In the next step we added age. Its influence to divorced people is not big, but to widowed it is huge. The older the person is, the smaller the difference between widowed and married people is (just 18.5% more traditional attitude than among married people). However, the quality of model M4 is not noticeably bigger that the model M3. In the last step we added the interaction between gender and education. The quality of the model strongly increased (R-squared 0,778 no, it is 0,0778 only J). Interesting is the difference between low educated male and low educated female and attitude towards gender roles itself. Less educated woman has more egalitarian attitude than less educated man. NO, you really have to put gender and education into your model, and after that put an interaction between gender and education. You probably did it because I advised it in my e-mails, and margins wouldn’t work without that, but there is no info in your table about that. In summation, index of attitude towards gender roles is not influenced only by marital status. The biggest impact, except from marital status have believing in god and level of education. Adding the level of education makes the model the greatest quality. Table 1. Determinants of attitude towards gender roles index M1 M2 M3 M4 M5 Marital status Married Ref Ref Ref Ref Ref. Divorced -0.759 -0.786 -0.632 -0.620 -0.332 Widowed 0.443 0.399 0.363 0.185 0.315 Having kids No Ref Ref Ref Ref Yes -0.115 -0.143 -0.030 0.176 Believing in God No Ref Ref Ref Yes 1.151 1.152 1.186 Age 0.010 0.008 Gender#education Male#Lower Ref Male#middle -0.661 Male#upper -1.106 female#Lower -0.613 female#Middle -1.510 female#Upper -2.326 Cons 17.226 17.308 16.376 15.778 17.075 N 53,227 52,330 49,550 49,550 49,215 R^2 0.0140 0.0140 0.0337 0.0355 0.0778 Graph 1 shows the marginal effects of age on attitudes towards gender roles index. From the graph and model M5 presented in the table 1, we can answer our research question and test our hypotheses. Hypothesis 1 assumed that married people have more traditional attitude towards gender roles than people that got divorced. As seen from the table, this hypothesis cannot be rejected, because even after adding few other variables, such as having kids, believing in god divorced people stays more egalitarian. Hypothesis 2 assumed that widowed people have more traditional attitude towards gender roles than married. Table 1 shows that even if after adding some other variables, percentage of traditionalism remains significant (from 18.5% to 44.3%). Good work, I take: -1 point for not presenting results for gender and education in your table -1 point for interpreting the model as a per cent change and not change in the index