Family and Personal Correlates of Academic Achievement

Researchers and educators raise the question of whether pupils' academic performance can be improved through parental involvement in academic activities. The main objective of the following study is to verify whether parental involvement in school activities and family socioeconomic status are associated with children's academic achievement. 150 Spanish seventh grade pupils completed intelligence tests, and their teachers assessed parents' involvement in the school and estimated parents' cultural levels. To measure academic achievement the pupil's overall grade was taken from the Pupils' Final Evaluation Registers. The education and professional level of the mother and father and home size were obtained from the Pupil Personal Register; these variables define the family socioeconomic status. The data, analyzed through application of structural equations, suggest that academic achievement is directly influenced by the cultural level of the family and the child's intelligence but is indirectly influenced by parental involvement in school activities and the socioeconomic status of the child's family.

cific, our intention was to seek answers to: (1) whether family variables are associated with academic achievement, (2) to what extent family socioeconomic status, pupil's intellectual abhties, and parents' involvement in education (as estimated by the teacher) are associated with academic achievement, (3) to what extent parental involvement is associated with ~arental cultural level, both being as perceived by the teacher. These issues are of particular importance to school psychologists who attempt to enlist parents in the schoohg of their children or who work with teachers of children whose parents may not be involved.
Intelligence and intellectual a~citudes are the factors concerning achievement most studied and one of the most stable when prehcting achievement. The variance in achievernenc accounted for by intehgence alone is estimated at around 35% (Castej6n & Navas, 1992). Most authors agree on the claim that family variables influence the development and educational achievements of children (Christenson, Rounds, & Gorney, 1992;Garcia, Clemente, Sorribes, & Vdanueva, 1994;Marjoribanks, 1994). But, which family variables are the most important? Many authors have distinguished between family structure variables and family process variables (Reynolds & Lee, 1991). The first group usually includes parents' professional and educational levels, marital status, family constellation, etc. The process variables, called by Walberg "home curriculum," have been grouped by Christenson, Rounds, & Gorney (1992) into five categories of parents' expectations and attributions, home learning structure, family affective climate, disciplinary styles, and parents' involvement in children's education. Walberg (1984) reported that home curriculum predicted academic learning better than family size and economic resources. Keith, Troutman, Trivette, Keith, Bickley, and Singh (1993) scated that the total effect of socioeconomic status on children's academic achievement at school is higher than that reached by parental involvement. Although current research has emphasized process variables against socioeconomic ones, one must note that many family processes take place in certain family environments that can be described on the basis of certain socioeconomic or structural characteristics. For example, parents with low education prefer to use coercive disciphary techniques, while parents with a university education employ affective techniques (Wranda, 1985); the academic performance of pupils from large families is slightly, but consistently, lower than that of children from smaller families; in general, nvo-parent farnihes are more effective in helping their children to comply with their roles as pupils than other family structures. In earlier studies, we have shown that family socioeconomic status variables, especially the mother's education, are narrowly related to children's development and academic performance (Garcia, et al., 1994).
Even given these results, it is generally agreed that parental involvement is associated effectively with improving pupils' learning (Christenson, Rounds, & Franklin, 1992). However, parental involvement is not a unitary construct (Keith, 1991). Some accept that the direct effect of parental involvement on achievement is greater than that of socioeconomic status but that parental involvement is partly based on socioeconomic status (Keith, et al., 1993). Others have indicated that parental involvement explains almost aU the influence parents' education has on achievement (Stevenson & Baker, 1987). Not all types of parental involvement, however, have had the same relation with achievement, neither does a given family variable have the same effect on the different types of parental involvement. Singh, Bickley, Trivette, Keith, Keith, and Anderson (1995) pointed out that parents' expectations with regard to their children's educational achievements have some influence on the academic attainments of eighth-grade pupils, but the incidence of parent-child communication is small or nonexistent. On the other hand, Miller (1986) indicated that parents' education is associated with the~r interest in school matters but not their participation in parents' associauons Among the main factors playing a role in parental involvement are family socioeconomic variables (Keith, et al., 1993;Grolnick & Slowiaczek, 1994).
Lastly, how do teachers influence parental involvement? While Dauber and Epstein (1993) claimed that teachers' practices in promoting and guiding parental involvement and their sensibhty toward parents' differential needs are the best predictors of parental involvement, Seeley (1989) observed that some teachers think promoting parental involvement is not part of their professional role.
Some studes have shown that teachers have a negative vision of the family, which leads to family variables being seen as the cause of children's educational needs (Moses & Croll, 1987). Indeed, there are many teachers who believe that parents wdl continue a low involvement even when given opportunities to increase it (Leitch & Tangri, 1988). Leitch and Tangri (1988) added that teachers believe that uninvolved parents have a negative influence on home-school relations. However, Rich (1987) claimed that the emphasis teachers give to family conditions should be interpreted more as a recognition of the importance of the home as an educational environment than as an attempt to blame the family.
The main purpose of this study was to understand how family socioeconomic status and pupils' intellectual ability may be related to the pupils' academic achievement, taking parents' involvement in education and parents' cultural level (as perceived by the teacher) as nlediator variables.

Model
The proposed model is based on Keith, et al. (1993) and is designed to test the influence of parental involvement perceived by the teacher on seventh-grade pupils' academic achievement, after controlling for relevant background variables (family socioeconomic status variables and pupils' intellectual abhty). The model contains original components: achievement is more strongly related to the parental involvement perceived by teachers rather than as reported by parents. According to Hulsebosch (1991), teachers may support parental involvement or not, and also as Grolnick and Slowiaczek (1994) pointed out, teachers may act as the mechanism through which the positive effects of parental involvement are manifest. The model considers that the multiple influences on academic achievement can be mediated by the evaluation the teacher makes of the pupil and family. The opinion the teacher has of the pupil is strongly associated with the opinion of the family.
The Parental Involvement Perceived by Teachers Questionnaire (Garcia, 1989) presented an item which asked teachers to estimate parents' cultural levels. This variable was included in the model because it was assumed to have a global character and because it is supposed that the cultural level of the family acts as a model of learning for the child. The hypothesis was that it is a variable in which the teacher summarizes the dfferent information obtained about the pupil and the family environment. This item has a high discriminating capacity in measurement of a child's competence and parental involvement (Garcia & Rosel, 1997). It correlates significantly with measurements of academic achievement, with measurements of intelligence, and with parental involvement other than that explained by socioeconomic variables. Correlations of this item with family socioeconomic variables range between .21 and 3 4 , which suggest that this variable is independent of them.
Although the model is especially focused on the effects of family variables on academic achievement, the effect of pupils' intellectual abhty on academic achievement cannot be ignored (Alexander, Pallas, & Cook, 1981). Consequently, in the model, academic achievement is claimed to be the direct result of the pupils' personal resources (intellectual ability) and the resources to which the pupils have access-determined by social origin and parental involvement.

Sample
The sample was all the pupils in the seventh grade (N= 163), their parents, and their teachers from three public primary schools in the urban area of Castekn (a Spanish town on the Mediterranean coast with approximately 110,000 mhabitants) selected by the Local Education Authority as being rep-resentative of schools in the town. Participants took part in the research voluntarily and were guaranteed that their identity would not be disclosed.
The pupils were from five classes, two in School A (n = 27 and 34), two in School B (n=40 and 391, and one in School C (n=24). The sample was reduced to 150 pupils given incomplete records for 13 pupils; thus the final sample was made up of 85 boys and 65 girls, with numbers per classroom of 25, 32, 34, 35, and 24, respectively. The distribution by sex and age was homogeneous in the five classrooms. All were Caucasian, and 92% of pupils were 13 years old. There were 57 fathers and 93 mothers (one per pupil). Ln this sample, 10% were single-parent families. Mothers and fathers were distributed equally over the different educational levels: 19% had not finished their primary studies, 50% had completed primary studies, and 31% had completed secondary or university studes. As to the professional level of mothers, 41% were housewives, 34% were cleaners or pieceworkers, 16% were clerks or middle grade technicians, only 7% pursued careers requiring university qualifications, and 2 % ran a small business. In the case of fathers, 16% were unemployed or were unskilled workers, 20% were self-employed (with their own business, but without employees), 40% were speciahzed technicians or had a small business with employees, and 24% pursued careers requir~ng university qualifications.

Variables in the Model
Family socioeconomic status.-The education and professional status of the mother and father and home size were obtained from the Pupil Personal Register. These variables define the family socioeconomic status. When data were not avadable in the Pupil Personal Register, we asked the pupils or telephoned parents.
-As to the number of rooms in the home (excluding the dining/living room, the lutchen, and the washrooms), 23% of families had 3 rooms, 52% had 4, and 17% had 5. Only 6% had more than 5 rooms and 2% fewer than 3.
Pupils' intellectual ability.-AU pupils answered Yuste's DLfferential and General Aptitudes Battery-M form (1988), which measures general verbal and nonverbal intelligence. The Verbal Lntelligence scale has three subtests (Verbal Mental Ability, Verbal Comprehension, and Numerical Aptitude). The reliabhty obtained by the split-half method was .92, and the reliabhty of the test-retest in a year was .85. As to validity, the correlations obtained were between .81 and .86 with scores on Otis's test (1992) and .55 with school achievement. The Nonverbal Intelligence scale also has three subtests (Nonverbal Mental Ability, Logical Reasoning, Spatial Aptitude). The reliabhty obtained by the split-half method was .94, and the reliability of the test-retest was .83. As to validity, the correlations obtained were between .51 and .56 with scores on the Raven Coloured Progressive Matrices Test (1971) and .3 1 with school achievement. The correlation between scores on the verbal and nonverbal scales was .74.
Parental involvement.-Parental involvement was of primary interest in this study. Teachers estimated the parental involvement of their-pupils' parents using the items of the Parental Lnvolvernent Perceived by Teachers Questionnaire (Garcia, 1989). This provides information about two Iunds of parental involvement in the school: teacher-parent contacts (7 items rated on a 4-point scale with anchors of always, often, sometimes, never, e.g., attend meetings with the teacher, collaborate with the teacher, and parental participation in school activities and 6 items rated on a 4-point scale using as anchors always, often, sometimes, never, e.g., they express opinions about the organization of the school, they are active members of the Parent-Teacher Association or school councils). These two parental involvement modes have a certain similarity with two of the types of parental involvement proposed by Epstein (1988), namely, at school and in governance and advocacy. They are measures of parental involvement perceived by teachers (Keith, 1991).
As to the validity of the Parental Involvement Perceived by Teachers Questionnaire, scores were significantly correlated with parents' conceptions of education and parents' reports of participation in the school (values between r = .25 and r = .46, for n=75). Garcia (1989) correctly classified 54 of 62 pupils (87%) by social status using the two measurements from the Parental Involvement Perceived by Teachers Questionnaire, and another nine measurements from teachers' evaluations of pupils. The coefficient alpha reliability was .88 (N= 150).
Parentx' cultural level.-Teachers were asked to estimate parents' cultural levels using a scale with four categories of very low, low, medium, and high. The measurement sought was teachers' perception of family cultural level, taking "culture" as meaning "humanistic" knowledge (literature, music, history, etc.) and (oral, written, and gestural) expressive style. The idea behind incorporating this variable was to provide a global indicator of the perception teachers have of their pupils' famhes. Garcia and Rose1 (1997) classified parents according to teacher-rated cultural levels and reported significant differences in intellectual level, academic self-esteem, social status, estimated achievement, and effort by pupils, and in various measurements of parental involvement.
We are giving priority to perception measurements over objective ones. Bronfenbrenner (1979) claimed that physical and objective characteristics are not the only important ones in the description of the environment, far more important is the meaning these characteristics take on for the people living in it.
Academic achievement.-To measure academic achievement, the pupil's overall grade was used, which was taken from the Pupils' Final ~v d u a i i o n Registers, the only official document registering academic results. The overall grade is not an arithmetical mean of the grades in each subject; it is more a joint qualitative evaluation by all teachers who have given classes to a pupil.

Procedure
Once the schools had been chosen, the research team informed principals of the aims of the research and asked for the participation of students in Grade 7, their parents, and their teachers. AN the seventh-grade students were administered an intelligence test. Then, the teachers were asked to describe the parental involvement of each pupil's family. Some teachers clainled not to feel ¶uaLfied to give their opinions of the famhes. Others were not keen co report on families. Given the situation, the research team informed them that the aim was to discover the perception they had of the relationship a particular family had with them and the school, independently of whether that information coincided with information obtained from other sources, for example, parents or the principals. We had made sure, beforehand, that all teachers knew students' families from the pre\rious academic year.
Findy, the education and professional level of the parents and the home size were obtained from the Pupil Personal Register. These variables represent an index of family socioeconomic status and, therefore, of the support or resources parents can offer their children. In Spain, there are no reliable statistics on family income and, culturally, there is reticence to disclosing real income, which is the reason why parents were not dlrectly asked about this. In Spain, the percentage of f a d e s having their own home is much higher than the percent of those living in rented accommodation. Having one's own home, then, is not such a discriminating socioeconomic status item as home size.
At the end of the academic year, we asked the schools' permission to consult their Pupils' Final Evaluation Registers. This record only reflects the overall grade obtained by each student and, when the overall grade is insufficient or very deficient, failed subjects were also noted.

RESULTS
The variable means, standard deviations, and intercorrelations among the variables in the model are shown in Table 1 First of all, we tested the measurement models for each latent variable. The models were estimated through exploratory factorial analysis of principal components, varimax orthogonal rotation, with the number of factors limited to the eigenvalues greater than 1.00, according to Kaiser's criterion. Analyses were made using the SPSS for Windows 6.1 program (Noruiis, 1994).
The variables of mother's education, mother's professional level, father's education, father's professional level, and home size are factorized; there is one factor (Bartlett test, ~,~= 2 1 0 . 1 8 , p < ,011 with mother's education, mother's professional level, father's education, and home size which accounts for 65% of the variance of the latent variable Family Socioeconomic Status. The variables General Verbal Intehgence and General Nonverbal Intelligence are gathered in only one factor (Bartlett test, x,* = 557.50, p < .01). This factor, which includes the two intelligence variables, explained 78% of the variance of the latent variable Pupils' Intellectual Abhty. The variables  (Bentler, 1995). The results of the structural equation analysis are shown in Fig. 1 Related to pupils' academic achievement are parental cultural level (0 = .47; t=2.18) and the pupils' intellectual abhty (P=.69: t=4.11). The pupils' intellectual ability also was significantly correlated w r h pupils' achievement, ~arental involvement, and parents' cultural levels (p = 20; t =2.12); see Table   3 . These results indicate that the best grades are obtained by the most intelligent pupils or by pupils whose parents are perceived to have the highest cultural levels.
Parental involvement is associated with parental cultural levels rated by the teacher (P= .74; t=7.04). Given that both factors were assessed by the teacher and bearing in mind the two measurements of parental involvemenc included in the model, this result suggests that one of the sources teachers use to estimate parents' cultural levels is the amount of time parents spend at the school. Parental involvemenc is related to family socioeconomic variables (0 = .24; t = 2.43) and children's intellectual abhty (P = .57; t = 4.60). In the teachers' opinion, parents with higher education and professional levels and with more intelligent children are more involved in the school. Parental involvement also has an indrect association with pupils' academic achievement, through ~erceived parents' cultural level (P = .35; t = 2.1 1). There is no evidence of a direct association of family socioeconom~c variables with pupils' educational achievement. Family socioeconomic vanables influence children's school achievement through parental involvement (p= .08; t = 1.51). But, given the positive and significant covariance between children's intellectual abihty and family socioeconomic variables (@ = 3 9 ; t=3.34) and the wide agreement among researchers about h k i n g high intellectual ab~Lity of children with high social class (Walberg & Marjoribanks, 1976), their contribution to educational results should not be ignored. Garcia and Rose1 (1997) detected differences in children's intellectual abhty associated with mothers' education and the professional level of both parents. Lastly, the model proposed fits the data but is only one of several that could have been significant. More research based on theories and on the observation of the relationships between parents, teachers, and pupils in schools, therefore, needs to be carried out.

Drscussro~
The obtained results show that family variables are associated with children's educational achievements. The sum of the hrect and mdrect effects of the three family factors on pupils' grades in the model is very high (P= .90), in fact slightly higher than the total effect of children's intellectual ability (p=.89). What is more, we must consider that part of the effect of intellectual abhty on educational achievement is produced through parental involvement and parents' perceived cultural levels (p = 2 0 ) . According to the model, teachers perceived that parental involvement had a significant effect, albeit indirect, on pupils' academic success. But, if we consider that parental involvement is the only varlable used in the model that can be modified by the intervention of teachers and school psychologists, its importance increases. Teachers not only believe that ~arental nonin-volvement negatively affects family-school relations and children's school outcomes (Leitch & Tangri, 1988) but also that parental involvement positively affects children's school achievement. This observation is consistent with the idea that teachers evaluate their pupils' famhes and assume that these influence academic achievement. In the model tested here only the measurements of perceived parental involvement in schools have been included ("teacherparent contact," and "parental participation at school"). This could indicate that teachers only place value on parental involvement in the school (Grolnick & Slowiaczek, 1994) or that they underestimate or ignore parental involvement at home (Johnson, 1991 ).
Teachers believe that parents are a good resource for favoring their children's educational achievement if their physical presence in the school is obvious (they attend or participate) and if they have high professional status and education. The teacher could be the channel through which the positive effects of parental involvement on children's achievement and adjustment are produced (Grolnick & Slowiaczek, 1994). Epstein (1988) pointed out that parents who have meetings with their child's teacher could influence the teacher's interest in orienting the pupil in learning. The parents also become more familiar with the teacher's proposed educational goals, which enables them to be of greater help to their children (Epstein, 1986).
The results of the study of the ddferential effect of family socioeconomic variables and parent involvement on children's educational achievements confirm those by Keith, et al. (1993): even though family socioeconomic variables are only indirectly associated with achievement, the thesis that parental involvement is based on socioeconomic status does seem to be confirmed (Garcia & Rosel, 1999). That is to say, parents with high professional status and education tend to be more involved, and this parental involvement (as perceived by teachers), in turn, favors higher achievements. However, Grolnick and Slowiaczek (1994), quite rightly, wanted to draw our attention to the possibihty that parents of different education get involved in different ways.
The results seem to confirm the molarity of the factor of parent cultural level as perceived by the teacher. Firstly, it is the only factor, along with the pupils1 intellectual abhty, that is directly associated with academic achievement. Secondly, all the other factors are seen to be h k e d with achievement -ultimately through cultural channels-and are significant for parental involvement (p = 3 5 ; t = 2.1 1) and pupils' intellectual abhty (P = 2 0 ; t = 2.12). Thirdly, it is significantly related to parental involvement (P = .74; t = 7.04), family socioeconomic status (P = .18; t = 2.041, and pupils' intellectual abhty (p = .42; t = 3.64). This result could be explained by findings made in a variety of research areas. However, we must remember that, as this factor was based on only one variable, it has a serious luniting effect in this study.
Research regarding teachers' expectations about pupils suggest that many of the teachers' opinions about pupils are generalizations (Carr & Kurtz, 1991) based on some salient characteristic of the pupil and on pre-existing theories (quotidian or scientific) belonging to that particular culture. As Pajares (1992) points out, teachers' beliefs are the result of an "enculturation process." Specifically, it may be that the educational values attributed to some characteristics of family life in our society are included in the semantic network of teachers' beliefs about pupils' achievement. In this sense, our work seems to indicate that teachers, when elaborating their beliefs about educational determinants, tend to use those characteristics of the family context that are easily accessible given their externality as, for example, parents' education and professional levels or the parents' presence m h e school.
This possibility is especially important because teachers' impbc~t ideas are quite resistant to changes, have a strong penetration of their cognitive schemas about education, and their effects are observed in the way in which teachers plan their teaching, how they interact with their pupils, and in pupils' achievement (Fuchs, Fuchs, & P h h p s , 1994;Marredo, 1994). As Davles (1989) claimed, teachers believe that parents of low socioeconomic backgrounds have little to offer the school or their own children. Laureau (1989) concluded that parental involvement was linked to parents' socioeconomic status. Connell, Asheden, Kessler, and Dowsett (1982) suggested that teachers have in mind certain combinations of family variables when they referred to a pupil as having "bad" or "good" family backgrounds and that some of these family variables correlate with school success and that school f d u r e increases in famhes with shortcomings in these quahties. Johnson (1991) also detected that teachers tended to underestimate or ignore parental involvement at home, which made them elaborate a poor image about pupils whose parents do not have an acuve presence at school. Teachers need to have increasing knowledge about the main dimensions on which families can vary and about the influence the family sphere can have on pupils' achievement (Christenson, Rounds, & Gorney, 1992;Procidiano & Fisher, 1992). Unfortunately, teacher training in Spain includes few (or no) procedures for working with parents.
To sum up, the main conclusions are that family characteristics may be influential in pupils' school attainments almost as much as their intellectual abhty. Perceived parental involvement in the school seems to play a central role in the evaluation teachers make of their pupils. These aspects highlight the need for research on the ideas and perceptions teachers have of the family, making them especially sensitive toward the different and vahd roles parents can adopt and for research on a variety of procedures through which teachers can further parental involvement. The goal should be to achieve a match between what parents want and what schools perceive as feasible in aidmg pupils' performance (Christenson, Sheridan, Hurley, & Fenstermacher, 1997).