Publications>
Research Bulletin

Phi Delta Kappa Center for Evaluation, Development, and Research
December 1998, No. 22

Is Parent Involvement Related to Student Achievement? Exploring the Evidence
by Ron Thorkildsen and Melanie R. Scott Stein

After conducting a fairly extensive review of the literature on the influence of parent involvement in education, we conclude that many researchers overinterpret their findings. In many cases, undue importance is given to findings of statistical significance, or causal relationships are implied from correlational studies. In this Research Bulletin we focus specifically on the relationship between parent involvement and student achievement, interpreting these findings more cautiously. We also explain some of the problems in determining cause and effect relationships and in judging the importance of research findings.

Benjamin Bloom once wrote, "If we are convinced that a good education is necessary for all who live in modern society, then we must search for the alterable variables which can make a difference in the learning of children. . . . Our basic research task is to understand further how such alterable variables can be altered and their consequent effect on students, teachers, and learning."(1) Determining cause and effect relationships is the goal of most research. But isolating causative variables is very difficult when humans are the subject of study and the study is conducted in less than a laboratory environment. In the field of education, we are interested in knowing the causes of educational outcomes, such as what causes increases in student achievement, so we can use this information to improve our educational practices. And while it is possible to alter or change some variables that affect students' learning, as Bloom advises, it is not possible to change others. For example, it is possible to change the extent to which parents become involved in their children's education, but changing a parent's educational background or income level are beyond the realm of a school's influence. This distinction is particularly important, because children's academic achievement appears to be more strongly related to their parents' level of involvement than to their parents' level of education or income.(2)

Researchers will readily admit that cause and effect is never established based on the results of a single study, particularly in educational research. One way we can try to determine cause and effect is through a strategy called triangulation. Triangulation simply means that we try to determine cause through more than one method. For example, for this Research Bulletin, over 50 articles, including reviews and primary reports of studies, were examined to explore the relationship between parent involvement and children's academic achievement. Most of these articles were reported in refereed journals; that is, the study or review was evaluated by independent, professional critics and approved for publication. These studies and reviews are divided into experimental and quasi-experimental studies and correlational studies, wherever possible.

EXPERIMENTAL STUDIES ON PARENT INVOLVEMENT

When designing and conducting experimental studies, researchers attempt to control all of the variables that could account for the results. For example, a researcher may want to determine if implementing a set of parent involvement strategies results in higher reading achievement. He or she might administer a reading test to all students in the classroom before implementing the program (pretest), assign students randomly to one of two groups, implement the parent involvement strategies with parents and children in one group but not the other, and then administer a second reading test (posttest) to all the students. The researcher wants to determine if any differences between pretest and posttest scores were the result of the parent involvement strategies and not some other variable or variables, such as the child's school reading program or natural maturation. Controlling for the influence of these other variables is typically achieved by including a "control" group that does everything the experimental group does except for participating in the experimental program or "treatment."

To help insure equivalence between the experimental group and the control group, the researcher randomly assigns students to each group. Random assignment and equivalency between groups are extremely important, because the researcher wants to say that all variables that may influence the outcome of the study affect each group of students identically. Random assignment to groups, or the lack thereof, is the primary difference between experimental and quasi-experimental studies.(3) Experimental studies involve random assignment and are clearly the stronger of the two. However, quasi-experimental studies provide corroborating evidence.

Experimental researchers often use inferential statistics and report that their results are or are not "statistically significant." If a finding is statistically significant, the researcher is saying that results as large as those that were found (e.g., group differences in reading scores) rarely happen by chance. Statistical significance provides some information about the probability of chance occurrences with particular sample sizes (number of study participants), but it does not provide information about the importance of the results. This distinction is crucial, because many researchers appear to equate statistical significance with importance.

Other statistics, such as effect sizes, are independent of sample size and provide information about the importance or educational significance of a study's result. Effect sizes are important because they can be compared across studies, allowing comparisons between group differences and relationships between sets of scores from different studies. Determining the importance of a particular effect size is somewhat arbitrary; however, some guidance exists. For example, Jacob Cohen suggests that 0.20 is a small effect, 0.50 is a medium effect, and 0.80 is a large effect for correlation coefficients and mean differences.(4) G. Kasten Tallmadge considers an effect size of 0.25 to be educationally significant for achievement tests.(5) These numbers apply to both correlation coefficients and differences between group scores. The opinions of other researchers about the magnitude of an important effect size is also considered.

An important experimental study that examined the relationship between parent involvement and mathematics achievement was one conducted by John Fantuzzo and his colleagues at the University of Pennsylvania.(6) This 1995 study was well designed and implemented, and it is a replication of studies conducted in 1992 and 1993.(7) Evidence for cause and effect relationships is built most conclusively through experimentation and replication; that is, when experimental studies are repeated by other researchers and similar results emerge, there is strong evidence that a causal relationship exists. Replications rarely occur in educational research, however.

For this study there were two treatments, a parent involvement program and a reciprocal peer tutoring (RPT) program. For the parent involvement program, staff communicated with parents and parents with staff through notes and telephone calls. At the end of each week, parents were informed of their child's achievements through "celebration" notes, and parents then engaged in reward activities with their children, such as going to movies, shopping, and playing games. The RPT program involved dyads of students setting team goals and working together in alternating roles as teacher or student. The student dyads worked computational problems under the supervision of classroom assistants. Students participating in the study were randomly assigned to one of three groups: parent involvement only, parent involvement plus RPT, or a control group that did not participate in either treatment.

Students' scores on a mathematics achievement test were compared to determine if there were differences between the three groups. The results of this carefully designed study revealed both statistically and educationally significant differences. Scores of students in the parent involvement program plus RPT were statistically significantly higher than scores of students in the control group. Scores of students in the parent involvement only group were not statistically significantly different from the control group. However, keep in mind that with small sample sizes (12 and 13 students), large differences are required for statistical significance. With regard to the study's practical importance, scores of students in the parent involvement only group were educationally significantly different from those of the control group. These findings are very similar to the findings of the 1992 and 1993 studies, from which the 1995 study was replicated.

The importance of school personnel communicating with parents is supported by a study conducted by Sadie Grimmett and Mae McCoy in 1980.(8) They implemented a parent involvement program in which parents received communication describing their child's reading program and the child's progress in the program. They found both statistically and educationally positive results for the program, with effect sizes ranging from .73 to 1.02.

These studies have limited generalizibility, because they were conducted with at-risk, mostly black children from families who were considered as having low socioeconomic status (SES). That is, we cannot say with certainty that the findings from these studies would apply to all parents and all public school children nationwide.

Another well-designed study conducted in 1973 with a similar population of students examined the effectiveness of a parent involvement program that used geographical games directly related to the reading instruction provided to the students in school.(9) Parents played the game with their children at home. The researcher reported significant achievement gains in composite reading and vocabulary scores through the use of these games.

Finally, a large-scale, longitudinal study conducted in rural Appalachia involved parents as the primary teachers for their three- to five-year-old children.(10) The parents used a combination of special TV programs and parent guides. The students who participated in the program showed statistically significantly higher achievement gains on psycho-linguistic ability tests than did children in a control group.

QUASI-EXPERIMENTAL STUDIES ON PARENT INVOLVEMENT

Four quasi-experimental studies on the effects of parent involvement on achievement were reviewed for this bulletin.(11) One general finding of these studies was an increase in parent's positive attitudes toward parent involvement and the schools as a result of having participated in experimental parent involvement programs. With regard to achievement, only two of the studies reported statistically significant differences between the scores of students participating in programs using parent involvement and those of students in control groups or "control" school districts.(12) Where it was possible to calculate effect sizes from the statistics reported about achievement gains, effect sizes ranged from .42 to .50.

These findings are supported by a review of 29 studies by M. Elizabeth Graue and her colleagues.(13) Overall, parent involvement had a moderate positive effect on student achievement. In a study of 846 black, inner-city, elementary school children, Herbert Walberg and his colleagues noted, "The most important result is that classes whose parents were intensively involved in the program gained an estimated 1.1 grade equivalents; classes whose parents were less intensively involved gained only .5 grade equivalents."(14)

Common themes reported in these studies include the following:

• Parent involvement programs need to be multi-year programs.
• Parents support parent involvement programs.
• Parents' attitudes toward parent involvement programs and schools improve when they are involved.
• Younger children generally show greater achievement gains in both reading and math than do older children when their parents are involved in their schooling.

CORRELATIONAL STUDIES

When one behavior or trait accompanies another, we say the two correlate. A correlation is a statistical measure that expresses a degree of relationship between two variables, such as the height and weight of a group of people. The symbol used most often for a correlation coefficient is r, and values of r range between -1.0 and 1.0. An r value of 0 indicates no relationship between a pair of variables. A positive correlation means that two variables increase together or decrease together. An r value of 1.0 is perfectly positive, indicating that an increase in one variable corresponds directly to an increase in the other. An r value less than 1.0, such as 0.8, is still positive but less than perfect. For example, if height and weight correlate at r = 0.8, it means that weight tends to follow height but not perfectly.

A perfectly negative correlation, expressed by -1.0, indicates that an increase in one trait or behavior corresponds to a decrease in the other. An example of a negative correlation is rainfall and forest fire damage. As the amount of rainfall increases, damage by forest fires tends to decrease. Another example of a negative relationship is that between parent involvement and student dropout rate. As parent involvement increases, dropout rate tends to decrease.

Much of the research on parent involvement and student achievement is correlational. Researchers conducting correlational studies do not always attempt to control variables; they are interested in relationships between variables. For example, one might ask the question, is the degree to which parents are involved in their child's education related to their child's achievement on, say, a reading test? If we could obtain a score for degree of involvement and then measure the child's reading ability, we could calculate a correlation statistic and estimate the relationship. Let's say we found a relatively strong relationship, for example, r = .80. That is, for those parents who had a higher degree of involvement, their children tended to have higher reading scores. There is a relationship, but is there evidence that high parent involvement resulted in high student achievement? No. Causation can never be determined from just knowing a relationship. We never know for sure if there are other potentially casual variables at work, and we do not know for sure if the direction of the relationship is what we would expect. For instance, in our example above, there is evidence that parents tend to be more involved if their children are doing well. But did involvement cause achievement or did achievement cause involvement? We don't know. Well, then, is correlational research valuable? Yes. From correlational research we can detect simple and complex relationships. These relationships provide a base for additional research and information to help determine the strength of causal relationships.

Correlational studies use either bivariate or multivariate analyses. Bivariate analyses examine the relationship between a pair of variables, without regard to the relationship of other variables to the variables in the pair. A correlation between two variables can be influenced by one or more other variables. Not considering other relevant variables is termed "under specification." An example of under specification is the relationship between Head Start and first-grade achievement, which has been estimated at r = -0.19. This negative correlation misleadingly implies that if children attend Head Start, their first-grade achievement will decrease. However, if another variable termed "educational advantage" -- usually associated with high SES -- is included in the analysis, the relationship between attending Head Start and first-grade achievement is r = .30.(15) This change in the r value is because educational advantage is negatively correlated with Head Start (r = -0.70). That is, children who have a lower educational advantage are likely to be eligible for Head Start, and typically they are academically behind children who do not qualify for Head Start. (This type of analysis is only possible using the multivariate statistics used in multivariate studies.)

Bivariate Analyses of Parent Involvement. One of the better correlational studies involved a relatively large representative random sample (328 students).(16) The researchers included correlational and qualitative indices, such as parent, teacher, and student interviews and case studies. They found that "few significant relationships appeared to exist between parent involvement activities and children's school performance. Where a link was found, it generally related to home-based reinforcement provided by parents."(17) The statistically significant correlations ranged between .22 and .28 -- effect sizes normally considered small but educationally significant. Home-based reinforcement included such activities as parents encouraging reading, caring what happens in class, keeping track of school progress, and finding children a place to study. Findings from this study also indicated that parents, students, and teachers were positive about parent involvement. These results are generally supported by other such correlational studies. Researchers have reported statistically significant relationships between parent involvement and achievement with correlation coefficients ranging from 0.16 to 0.49.(18) Other common findings among these studies are:

• parents with higher education levels are generally more involved;
• younger students are more influenced than older students by parent involvement;
• home-based reinforcement has an educationally significant affect on achievement; and
• parent expectations have the strongest relationship with higher student achievement.

The latter finding is the most prevalent and is supported by the strongest relationships. Interestingly, there did not appear to be a positive correlation between parents helping with homework and student achievement. Another finding of interest is that parents of older children (middle and high school) tend to be more reactive to problems, that is, they get involved only when problems arise. This implies that parents of middle and high school students need to be more proactive.(19)

Multivariate Analyses of Parent Involvement. Multivariate studies examine the interrelationships between more than two variables. Through the use of multivariate analysis techniques, such as causal modeling, evidence revealing patterns of behavior can be built. Multivariate studies are important, because the relationships between a number of pertinent variables are considered.

A number of multivariate studies have examined the National Education Longitudinal Data of 1988 (NELS-88), a data set that is generally considered to be representative of U.S. students. The researchers in one of these studies concluded that the "strongest influence on achievement was that of previous achievement."(20) In addition, parents' expectations of their child's success had a strong positive relationship with student achievement. The researchers suggested that helping parents develop practical ways to convey higher aspirations to their middle school students may be one way to improve academic achievement and motivate children to work to meet the higher expectations.

Another study that used the NELS-88 data set examined the relationship between parent involvement and homework. Timothy Keith and his colleagues found a strong positive correlation between parent involvement and the amount of homework completed.(21) They also found a strong positive correlation between students' previous achievement and parent involvement, which implies that parents of high achievers are more involved, and more involvement increases achievement. The researchers noted that "parents do want to become involved in order to help their children achieve, but they need assistance and guidance in order to do so."(22) The strong relationship between students' previous achievement and parents' level of involvement further implies that parents of low achievers need more encouragement, help, and guidance than parents of high achievers. This finding is extended by a study of 481 low SES, minority second-graders.(23) The researchers found a relatively small but positive correlation between parent involvement and achievement and noted that parental involvement alone is insufficient to yield the kind of results often expected.

Of all the components of parent involvement studied in the correlational research reviewed here, parental expectations (or parental aspirations) has the strongest positive correlation with achievement. Furthermore, the results of these correlational studies suggest that parent communication with the school is important, as is communication between the parent and child about school. There is a small positive correlation between parents helping with homework and achievement; however, a stronger positive correlation exists between achievement and parents encouraging homework as well as setting aside time and a place to do homework.

SUMMARY

Many of the studies reviewed here involved large numbers of participants, allowing relatively insignificant findings to achieve statistical significance, which can give undue importance to relatively insignificant findings. Consequently, both statistical and educational significance should be used in judging the importance of a study or group of studies.(24) Both were used in drawing conclusions here, and small to moderate but educationally significant effect sizes across many studies were found.

Where other variables are included in a study, student ability and school climate have substantially stronger relationships with achievement than does parent involvement. However, parent involvement appears to account for 10% to 20% of the variance in achievement, which is important. Likewise standardized mean difference effect sizes of 0.70, found in some of the better experimental studies, would be considered important when compared to commonly used standards.

Based on our review of all the aspects of parent involvement that have been studied, parent expectations of their child's success in school consistently has the strongest relationship with achievement. A supportive home environment provided by parents with high expectations for their child's success is consistently related to higher achievement.(25) Another cogent finding is that parents need strong, ongoing support from schools to provide effective parent involvement. As noted by Joyce Epstein, "Helping students succeed in school is an ongoing process, requiring attention every year -- from preschool through high school -- from teachers, families, and the students themselves."(26) This need for attention is especially true with low-income families. These parents are traditionally less involved; therefore, it is particularly important for schools to initiate involvement with these parents.

ENDNOTES

1. Benjamin Bloom, "The Search for Methods of Group Instruction as Effective as One-to-One," Educational Leadership 4 (8): 16 (1980).

2. Kusum Singh et al., "The Effects of Four Components of Parental Involvement on Eighth-Grade Student Achievement: Structural Analysis of NELS-88 Data," School Psychology Review 24 (2): 299-317 (1995); Anne T. Henderson, "Families and Student Achievement," PTA Today 20 (4): 12-14 (1995).

3. Donald T. Campbell and Julian C. Stanley, Experimental and Quasi-Experimental Designs for Research (Chicago: Rand McNally College Publishing,1963), 34. Some quasi-experimental designs employ randomization, but the experimenter does not have full control over the scheduling of the treatment. Lack of randomization and/or the ability to control treatment schedules diminishes experimental control and results in a quasi-experimental study.

4. Jacob Cohen, Statistical Power Analysis for the Behavioral Sciences, 2d ed. (Hillsdale, N.J.: Lawrence Erlbaum Associates, 1988).

5. G. Kasten Tallmadge, Joint Dissemination Review Panel Ideabook (Washington, D.C.: GPO, 1977).

6. John W. Fantuzzo, Gwendolyn Y. Davis, and Marika D. Ginsburg, "Effects of Parent Involvement in Isolation or in Combination with Peer Tutoring on Student Self-Concept and Mathematics Achievement," Journal of Educational Psychology 87 (2): 272-81 (1995).

7. John W. Fantuzzo, Judith Alperin King, and Lauren Rio Heller, "Effects of Reciprocal Peer Tutoring on Mathematics and School Adjustment: A Component Analysis," Journal of Educational Psychology 84 (3): 331-39, (1992); Lauren Rio Heller and John W. Fantuzzo, "Reciprocal Peer Tutoring and Parent Partnership: Does Parent Involvement Make A Difference?" School Psychology Review 22 (3): 517-34 (1993).

8. Sadie A. Grimmett and Mae McCoy, "Effects of Parental Communication on Reading Performance of Third-Grade Children," The Reading Teacher 34 (3): 303-8 (1980).

9. Blanche Edwards Clegg, "The Effectiveness of Learning Games Used by Economically Disadvantaged Parents to Increase the Reading Achievement of Their Children," paper presented at the annual meeting of the American Educational Research Association, New Orleans, La., March 1973, ERIC ED 085 656.

10. Edward E. Gotts, "Long-Term Effects of a Home-Oriented Preschool Program," Childhood Education 56 (4): 228-34 (1980).

11. Stan Brodsky et al., "An Urban Family Math Collaborative. Report No. CASE-09094," Center for Advanced Study in Education, City University of New York, 1994; Ronald M. Gillum, Daniel E. Schooley, and Paul D. Novak, "Effects of Parental Involvement on Student Achievement in Three Michigan Performance Contracting Programs," paper presented at the annual meeting of the American Educational Research Association, New York, N.Y., April 1977; Benny B. Wade, "Parent Education: Does It Make A Difference?" paper presented at the Southeastern Regional Conference of the International Reading Association, Nashville, Tenn., February 1980; Herbert J. Walberg, Robert E. Bole, and Hersholt C. C. Waxman, "School-Based Family Socialization and Reading Achievement in the Inner City," Psychology in the School 17 (4): 509-14 (1980).

12. Brodsky et al., "An Urban Family Math Collaborative"; Walberg, Bole, and Waxman, "School-Based Family Socialization."

13. M. Elizabeth Graue, Thomas Weinstein, and Herbert J. Walberg, "School-Based Home Instruction and Learning: A Quantitative Synthesis," Journal of Educational Research 76 (6): 351-60 (1983).

14. Walberg, Bole, and Waxman, "School-Based Family Socialization," 515.

15. Meredith D. Gall and Walter R. Borg, Educational Research: A Guide for Preparing a Thesis or Dissertation Proposal in Education, 5th ed. (New York: Longman, 1989).

16. Kim O. Yap and Donald Y. Enoki, "In Search of the Elusive Magic Bullet: Parent Involvement and Student Outcomes," paper presented at the annual meeting of the American Educational Research Association, New Orleans, La., April 1994.

17. Ibid., 1.

18. See, for example, Ruth E. Geyer and Jianhua Feng, "Parental Involvement and Its Relationship to Second-Graders' Reading Achievement," Muskingum College, Ohio, June 1993, ERIC ED 359 488; and C. R. Bulach, Bobby Malone, and Christy Castleman, "An Investigation of Variables Related to Student Achievement," Mid-Western Educational Researcher 8 (2): 23-29 (1995).

19. Sharon E. Paulson, "Relations of Parenting Style and Parental Involvement with Ninth-Grade Students' Achievement," Journal of Early Adolescence 14 (2): 250-67 (1994).

20. Singh et al., "The Effects of Four Components of Parental Involvement on Eighth-Grade Student Achievement," 306.

21. Timothy Z. Keith et al., "Does Parental Involvement Affect Eighth-Grade Student Achievement? Structural Analysis of National Data," School Psychology Review 22 (3): 474-96 (1993).

22. Ibid., 493.

23. Arthur J. Reynolds, "Comparing Measures of Parental Involvement and Their Effects on Academic Achievement," Early Childhood Research Quarterly 7 (3): 441-62 (1992).

24. James P. Shaver, "What Statistical Significance Testing Is, and What It Is Not," Journal of Experimental Education 61(4): 293-16 (1993).

25. See, for example, Barbara K. Iverson and H. J. Walberg, "Home Environment and School Learning: A Quantitative Synthesis," Journal of Experimental Education 50: 144-51 (1982); Karl R. White, "The Relation between Socioeconomic Status and Academic Achievement," Psychological Bulletin 91 (3): 461-81 (1982).

26. Joyce L. Epstein, Beth S. Simon, and Karen Clark Salinas, "Involving Parents in Homework in the Middle Grades, Research Bulletin No. 18," Center for Evaluation, Development, and Research, Phi Delta Kappa International, 1997, 4.