Determinants and Perceptions of Social Mobility in Poland, 1992-2008

The purpose of this article is to analyse the factors affecting social mobility in Poland, based on results obtained from the estimation of logit models for the years 1992-2008 as well as from interviews asking about the necessary conditions to achieve success in life. Data used in this work come from the Polish General Social Survey carried out by the Institute of Social Studies of the University of Warsaw on a representative sample of adult household members. This set contains interviews conducted in the years 1992-2008 and includes 16,234 respondents. Studies using logit models to answer the question posed in the present study show that the probability of success in life in Poland is determined by both ability and level of education; interview responses to the same question, however, show that family origins, parents’ level of education and connections are just as important. Most interestingly, the percentage of people indicating the importance of these factors was higher in 2002 than it was in 1992.


Introduction
The concept of social mobility (Saunders, 2010, p. 1-2;Social mobility, 2008) concerns an individual or social group's ability to change his or her position in the social system. In the case of intragenerational mobility, movement is measured in relation to the life of a single person. If a person's current position is discussed in reference to the position held by his or her parents at the same age, it is described as intergenerational mobility. Profession, income level, education, and other measures of class or socio-economic status can be compared. When a person moves from a lower to a higher social position, he or she achieves social promotion, whereas when a person moves from a higher to a lower position, he or she undergoes social degradation. The factors influencing social mobility are a subject of research in many scientific disciplines, including sociology, psychology, epidemiology, and social medicine.
In fact, the social status achieved in adulthood is an important indicator of access to material goods and health status (Adams et al, 2004;Boyle, Norman & Popham, 2009;Smith et al., 1998). The existence of social mobility also has consequences for social cohesion, equality, economic stability and happiness (Aldridge, 2003). Therefore, it is important to isolate the factors that facilitate mobility as well as those that may pose a barrier to it in order to target material resources that will support the former (Forrest et al., 2011). Most of the research on this subject shows that social, educational, material, cultural and physical factors in individuals' childhoods or early youth impact the direc-

Determinants and perceptions of social mobility in Poland, 1992-2008
tion of social mobility in adulthood (Blane, Smith & Hart, 1999). Depending on the data and methodology used in a particular study, however, this effect is different. In his study of American society, Jencks (1979) showed that the results of youths' mental ability tests are a strong predictor of educational attainment and professional status, even after accounting for variables such as education, occupation and parental income that reduce the strength of this relationship. Similarly, Herrnstein and Murray (1994) found a significant relationship between the results of the AFQT (Armed Forces Qualification Test) and the living conditions of people surveyed in the United States 10 years later.
Studies carried out by Korenmana and Winship (2000) on the same sample, however, show that the socioeconomic status of one's parents is equally important as intelligence. Similarly, Breen and Goldthorpe (1999;, who examined English society, found that the social class in which a person originates continues to exert an influence on the social class in which a person will be at the age of 33, even after taking into account the results of the General Ability Test carried out at the age of 11 and effort invested in education. These conclusions mirror similar, earlier findings from Halsey, Heath, and Ridge (1980). Using the same data as Breen and Goldthorpe, however, Saunders (2002) found individuals' class of origin explains about a quarter of the variance in social class at age 33, while individual abilities, motivations and qualifications explain more than 60% of the variance. A comparable impact on both children's mental abilities and their class of origin was confirmed in studies conducted by Deary and others (2005) in Scotland. On the other hand, Iannelli and Paterson (2005), using data from the 2001 Scottish Household survey, came to the conclusion that the effect of education on attaining a professional position grew for people entering the labour market from 1950-1980, but was on the decline at the time of their study. They also noted that parents' social position determines the social advancement of people with higher education levels to a lesser extent than it does those with lower education. In other words, in the case where children from a higher social class do not achieve the appropriate qualifications, they have family resources (e.g., financial support or social connections) that help them remain in their social class of origin. Studies carried out based on data from Denmark and Canada by Bingley, Corak and Westergard-Nielsen (2011) showed that most employment opportunities, especially in the case of high salaried positions, were found with help of friends and family.
The literature also encompasses studies examining subjective perceptions of social position (measured by respondents' indications of the places they occupy on a ten-rung social ladder) and the factors that influence this position. If these factors are independent of the respondent's demographics, such as gender, age, or ethnic and racial background, they reflect limited opportunities for social mobility within a given society (Lindemann, 2007). Studies conducted in Western countries show that the probability of identification with a higher social class depends on age, in that it is greater for older people. In these countries, the majority of unemployed and low-income individuals are young (Yamaguchi & Wang, 2002), whereas gender does not significantly affect people's self-esteem with regard to social position. In the case of Estonia, there are different dependencies where age is also a decisive factor; in this case, however, young people evaluate their position much more highly. This may be due to the social transformation that gave the advantage to social groups that were better able to adapt to new conditions, manage in the market economy and benefit from new technologies (Lindemann, 2007;Pettai, 2002). Discrimination against women in the Estonian labor market causes Estonian women to evaluate their positions lower in comparison to men. Moreover, the key factor influencing the perception of people's positions in society is profession. Some studies, however, indicate that education and achieved income are even more important (Kelley & Evans, 1995). The influence of education depends on the institutional solutions that exist in a given society. This factor has a strong influence in the Nordic countries and continental Europe, where the level of education has a significant impact on a person's abilities in the labor market (Knudsen, 1988). Much weaker effects are observed in the U.S. and the U.K., where higher education is accessible to many people and therefore does not confer the same level of advantage (Gross, 2003). In Estonia, where education does not guarantee a high income but rather helps people to realise their social ambitions, such as becoming part of the elite (Lindemann, 2007), is not dissimilar. Individuals' status in the labor market Determinants and perceptions of social mobility in Poland, 1992Poland, -2008 in Western countries and Estonia have a comparable effect on people's opinions of their social position; i.e., managers assess their position as being high, while unskilled workers put themselves at the bottom of the social hierarchy.
There are very few studies, however, in the area of subjective social mobility, understood as the evaluation of the respondent's professional position compared to the position held by his or her father when the respondent was 14 years old. A lack of studies does not mean that this topic is insignificant. Taking it into account may help to explain different social attitudes and behaviours, such as attitudes to discrimination or life satisfaction (Kelly & Kelly, 2009 Conversely, parents' education and the high position of one's father on The International Socio-Economic Index of Occupational Status (ISEI) have a negative effect (in the latter case, perhaps because children whose parents are already at the top of the social ladder are less likely to move even higher (Kelly & Kelly, 2009)). Research carried out in Ukraine based on ISSP data from 2009 shows that, as in the work cited above, subjectively perceived social mobility is influenced by factors such as the difference between the economic position of the respondent's profession and the position of his or her father's profession, as well as holding a managerial position. Furthermore, age effects have been indicated (the older the person, the higher their mobility), as well as effects related to one's place of residence (people living in rural areas have a higher subjective mobility) (Oksamytna, 2010).
An analysis of the literature cited above makes it clear that the most frequent topic of study is whether social mobility is affected by factors dependent on the respondent, such as intelligence, ability and educational achievement, or social origins. This was the subject of the author's considerations in a previous article (Sokołowska, 2013), in which the author examined whether Polish society is meritocratic; that is, if all people have the same chance to attain a high social position, which any person can reach if he or she is talented and willing to work hard (Young, 1958).
This study highlights a number of factors, including those testifying to equal opportunities for all in terms of intergenerational mobility, and factors signifying that a given society is far from meritocratic. There was a conclusion that the study results would change if it was possible to account for IQ test results or personal connections in obtaining a job. An examination of these variables was not possible because such questions were not asked in the PGSS; the closest question to this in the survey asks the respondent to assess whether, in his opinion, a given condition for success in life is essential, very important, moderately important, not very important or unimportant. This provides an interesting juxtaposition of how far our judgments about the determinants of success in life are consistent with reality, and whether a particular respondent's situation affects that judgment.

Research methodology and selection of variables
To examine whether individuals' views on the conditions necessary to achieve success in life coincide with what actually impacts social mobility, the author used the results obtained from the article cited above (Sokołowska, 2013). The data on which the article was based were derived from the Polish General Social Survey (Cichomski, Jerzyński & Zielinski, 2009a)  do you think that, compared with the average Polish family income, your own family income is much lower than average, slightly lower, average, slightly higher or much higher than the average? and the responses to a similar question about the respondent's family income in the period when he was about 14 years old.
The choice of explanatory variables was guided by the literature (Breen & Goldthorpe, 2001;Deary et al., 2005;Domanski, 2004;Duncan et al., 2002;Firkowska-Mankiewicz, 2002;Forrest et al. 2011;Halsey et al., 1980, Saunders, 2010Waller, 1971) and the availability of the variables in the data set. Among the initial set of variables, variables relating to the father's, mother's and respondent's professions, the respondent's and respon-dent's parents' levels of education, respondent's sex, age in years, and level of family income when the respondent was 14 years old (only for models 1-4), and rated by the interviewer's understanding of the respondent's PGSS survey questions, were found. The final variable was chosen because of the lack of IQ tests results in the data. Moreover, the literature on the subject allows us to assume that the above factor has significant effect on social mobility, which is why the attempts were made to at least partially compensate for the lack of choice available in the survey questions. In addition, depending on the study year, different sets of variables were available, including those related to the knowledge of foreign languages, completed vocational courses, the number of books in the home when the respondent was 14 years old, literacy, numeracy, and features of the respondent's personality. The Table 1 contains a list of variables attached to the models in particular years.
Due to the form of dependent variables, logit models were used to establish dependencies. These models are used when a random component has a logistic distribution and the dependent variable is qualitative. If it is

1992-2008 1999 2002
• Family income when the respondent was 14 (from 1 = "much lower than average" to 5 = "much higher than average" • Occupation of the respondent (0 = "unskilled workers" to 9 = "military / armed forces" • Father' s education level (from 0 = "no education" to 9 = "completed higher") • Mother' s education level (from 0 = "no education" to 9 = "completed higher") • Respondent' s education level (from 0 = "no education" to 9 = "completed higher") • Understanding the PGSS survey questions (from 1 = "poor" to 3 = "good" • Sex (0 = "female", 1 = "male") • Economic position of the respondent' s profession (16 = "labourers, helpers and cleaners" to 90 = "judges") • Age of respondent in years • Completion of vocational courses or further education within the last three years (0 = "no", 1 = "yes") • Number of languages known by the respondent (0,1,2,3) • Number of books at home when the respondent was 14 (1 = "none" to 9 = "1000 or more") • Self-assessment of literacy in Polish and numeracy (1 = "poor" to 4 = "excellent") • Degree of respondent's agreement with the statement, "I am resistant to sustained failure": from 1, "definitely not", to 4, "definitely yes" • Degree of respondent's agreement with the statement, "I set myself generally difficult and ambitious goals": from 1, "definitely not", to 4, "definitely yes" • Degree of respondent's agreement with the statement, "Most of the things that I care about, I do at least as well as others": from 1, "I strongly disagree", to 5, "I strongly agree" • Degree of respondent's agreement with the statement, "When I care about something very much, I usually cannot get it": from 1, "I strongly disagree", to 5, "I strongly agree" Table 1. Sets of explanatory variables available in each period used in the study.
Notes: Variables not shown in italics are only been used in models from (1) to (4) Determinants and perceptions of social mobility in Poland, 1992Poland, -2008 an ordinal variable (models 1-4), it is termed a model for sequenced categories, and in the case that there are only two values (as in models 5-7), it is a dichotomous model. The construction and estimation of logit models is described, inter alia, in the works of Cramer (2001;, Greene (2009) and Gruszczyński (2001;).
The logit model for sequenced categories is based on the assumption that the dependent variable * y is a certain unobserved continuous variable, which is a linear function of explanatory variables are stored in the column vector i x and unknown parameters are stored in the column vector β : A reflection of the unobservable variable * y is an ordinal variable y, which takes, in the case of models from 1 to 4, 10 values corresponding to the dependent variable of self-esteem, social position. The relationship between these variables can be expressed as follows: Where the thresholds j k are usually unknown and are estimated together with the parameters β. Because unobservable variable * y can take on any real value , we can assume that + ∞ = − ∞ = 1 0 0 , k k and the number of thresholds requiring estimation is limited to 9 (Gruszczyński, 2010).
The probability i j p that variable y takes the value j for the i-th person surveyed can be written as: In the case of the logistic distribution of the random component, its cumulative distribution is expressed by the formula: and therefore: The evaluation of models matching was carried out using several ratios (Gruszczyński, 2010). The first ratio used was McFadden's adjusted pseudo-R2 ratio, expressed by the formula

Results
Parameter estimates of received models are included in Tables 2 and 3. In the case of the logit models, the estimat-  Table 2, it can be concluded that the decisive factor positively impacting the respondent's high self-esteem concerning his or her social position is that respondent's ability to understand the survey questions, followed by his or her level of education. Due to a lack of questions concerning IQ test results, these variables can be regarded as reflecting, to a certain extent, the person's level of intelligence. In addition, in all models, having a profession with a high value on the International Socio-Economic Index of Occupational Status (ISEI) positively affects self-esteem. This scale was developed in 1996 by Ganzeboom and Treiman (1996)    optimistic when they assess their social position, and perhaps also that young Poles are more successful in life than their older counterparts. This may be due to the period of transition, which, as in Estonia (Lindemann, 2007;Pettai, 2002), provided young people with the opportunity to make a career. The positive symbol of the parameter for the gender variable means that in Polish society, men are more likely to occupy a high social position than women, which demonstrates the disadvantaged situation of Polish women. Another factor that positively influences the probability of high social self-esteem, found in Model 2 (estimated for 1999), is the number of foreign languages spoken by the respondent, as well as the respondent's completion of professional courses within the last 3 years.
In Model 4, measuring the year 2002,character traits such as resistance to failure, setting difficult or ambitious goals, having high self-esteem manifested in respondent's agreement with the statement: "Most of the things that I care about, I do at least as well as others" and persistence in the pursuit of the goals (negative sign for the answer to the question: "When I really want something, I usually cannot get it") positively affects social self-esteem. In most cases, the above-mentioned factors speak to the fact that Polish society is rooted in a meritocracy. The significance of the parameters in Model 1 for variables such as family income when the respondent was 14 years old, parental education level, and in model 3, the number of books at home when the respondent was 14 years old, attest to the fact that social position is partly determined by background. This means that children with more affluent and educated parents will find it easier to achieve a higher social position; however, the parameters of these variables  Table 3. The estimation results of logistic regression models for the dependent variables of expense promotion and income degradation, measured on a scale from 0 (no mobility income) to 1 (income mobility) (Sokołowska, 2013).
Notes: For the year 1999, an additional model with the dependent variable of income promotion was estimated, because that year, questions about foreign language skills, literacy and numeracy were included on the PGSS. For the year 2002 there were neither estimates nor an additional model for the dependent variable of income promotion or degradation (also there was no separate model for this variable for the year 1999) due to the lack of significance of the parameter estimates for attached variables in those years. Estimates of parameters significantly different from zero at a significance level ≤10% are shown in bold.
Determinants and perceptions of social mobility in Poland, 1992Poland, -2008 are statistically insignificant among the variables studied, such as the nature of the respondent's qualities (Model 4), the number of foreign languages they know or vocational courses they have successfully completed (Model 2). Comparing the obtained results to the literature cited above, it can be seen that in the factors influencing social self-esteem, Polish society is similar to Estonian society, with the exception that in Polish society, education levels have a stronger influence than professional position. Estonia displays an inverse relationship (Lindemann, 2007).
As for the probability of promotion (Model 5 in Table   3), the most influential of the variables studied are education, gender (being male increases the probability of success), the economic position of the respondent's pro-  generally lower for models estimated using cross-sectional data than for models estimated on the basis of time series (Gruszczyński, 2010, p. 62).
At this point, the author will examine whether the results obtained in the article discussed above (Sokołowska, 2013) coincide with Poles' views on the conditions necessary to achieve success in life. Respondents were asked to rate the factors that may affect the achievement of success on a 5-point scale from 1, negligible, to 5, crucial. Figure 1 shows the percentage of people who indicated that any given category was crucial or very important. The data were collected first in 1992 and then in 2002. Figure 2  The author next decided to explore whether one's views on success are influenced by experiencing a promotion or degradation of income. Therefore, figure  The answers of individuals with above average incomes compared to answers those with belowaverage incomes led to similar conclusions, and those statements will therefore not be presented in the article.

Conclusion
In conclusion, it is clear that our success in life is decided not only by our abilities, education and work habits but also by the family to which we were born, our gender and our age. These conclusions can be drawn either on the basis of the models used in the study, or on the analysis of respondents' answers. What we consider to be the possibility or likelihood of social mobility in Polish  society does not diverge from reality. In addition, data obtained from answers to the question about the conditions for achieving success shows that connections or political affiliations are also important. These factors were not taken into account in the logit models due to the lack of questions about these factors in the PGSS. Identifying these factors as important is a worrisome phenomenon, proving that Polish society may yet be far from providing equal opportunities for all. The increasing importance of having a rich family for success in life may also indicate that Polish society is not going well. Being from a poor family may, for example, be an inhibiting factor to obtaining a good education. Education is theoretically free in Poland, but parents must be able to afford to house, clothe and feed their school-age children. It is also possible payable learning in extramural system but then it is necessary to have a job that allows getting by and paying tuition.
The studies using logit models show that the possibilities for success in life in Poland are determined in the first place by ability and education. However, the importance of factors such as a rich family, parental education levels and connections attest to the fact that origins are similarly important in determining our life chances.