WPRI Report Continued:
the Milwaukee Public Schools:
By David Dodenhoff, PhD.
Using similar methods, this section presents estimates of various kinds of parental involvement within the MPS system. These can be broken into two primary types: on-site involvement (at the child’s school), and at-home involvement.
The first task in this section is to construct a measure indicating high levels of on-site parental involvement. (Given the comparatively limited impact of on-site involvement on student achievement, only high levels of parental involvement are worth considering.) At a bare minimum, a highly involved parent would be expected to:
Considering that participation in just these two activities might require as little as two or three hours per year, a parent who participated in only these activities could not be considered highly involved at the school. The NHES survey, however, asked two other questions pertaining to on-site involvement. These addressed: a) parental attendance at general school meetings, and b) parental service as a volunteer at the school.
Similar to the two activities above, attending a general school meeting is a relatively low-cost form of involvement. One might not be comfortable classifying a parent as highly involved if he or she did nothing more than attend a general meeting, attend an event in support of his or her child, and attend a parent-teacher conference. Again, over the course of a school year, this might require no more than three or four hours. On the other hand, requiring that a parent serve as a school volunteer in order to be classified as highly involved seems too restrictive. For purposes of the estimate below, then, a parent is considered highly involved at the school site if he or she:
(The interpolation of the frequency of this third, hypothetical activity is explained in the Methodology Appendix.)
Based on the foregoing criteria, one arrives at the following estimate:
An estimated 34 percent of MPS parents can be considered “highly involved” at their child’s school site.
Estimating levels of at-home parental involvement is somewhat more complicated, in that the nature of meaningful involvement differs among children of different ages. Accordingly, estimates of parental involvement are presented separately for three different groups of students:
1. those nine years old or younger,
2. those ranging in age from 10 to 13, and
3. those from 14 to 17 years old.For each of these groups, a set of activities was identified that, if engaged in by parents, would constitute moderate to high levels of at-home involvement. The specific activities were determined by a review of the relevant research literature and, of course, by the availability of questions addressing specific parental activities in the NHES data.[ii]
For the youngest group (students nine years old or younger), parents were considered moderately to highly involved in their child’s education at home if the parent:
An estimated 49.2 percent of MPS parents with children under age 10 meet all four of these criteria, and therefore can be considered moderately to highly involved at home.
For students ages 10 to 13, parents were considered moderately to highly involved in their child’s education at home if the parent:
An estimated 42.5 percent of MPS parents with children from age 10 to age 13 meet all five of these criteria, and therefore can be considered moderately to highly involved at home.
Finally, for students ages 14 through 17, parents were considered moderately to highly involved in their child’s education at home if the parent:
An estimated 39.7 percent of MPS parents with children from age 14 to age 17 meet all five of these criteria, and therefore can be considered moderately to highly involved at home.
The numbers above can be combined to estimate the percentage of MPS parents who are highly involved at the school site and moderately to highly involved in their children’s learning at home.[vi] The children of these parents would expect to enjoy the most significant boost from parental involvement. The estimates for each age group are as follows:
A Note on Social Desirability Bias
The estimates in this study were derived in large part from survey data. Unfortunately, survey questions of the sort used here are notoriously prone to what is known as “social desirability bias.” This kind of bias can enter a survey when certain questions have a “right,” or socially desirable, answer that respondents may give even when it is not accurate.
Imagine, for example, a survey of political attitudes and participation. After a standard battery of questions on party affiliation, news sources, and opinions on Congress and the president, the interviewer might ask, “Did you vote in the last election?” There is clearly a socially desirable answer to this question—“yes.” Voting is what all good citizens do. Recognizing this, many people tell survey interviewers that they voted in the most recent election, even when they did not. In the 2002 mid-term election, for example, about 40 percent of eligible voters cast a ballot. In the National Election Study survey for 2002, however, 64.9 percent reported having done so.[vii] This over-reporting by a margin of nearly 25 points is an example of social desirability bias.
Questions about parental involvement in education also have a “right,” or socially desirable, answer. Responsible parents are supposed to go to school meetings, attend parent-teacher conferences, help their child do his or her homework, and so on. Because parents may be uncomfortable acknowledging that they have fallen short of these standards, some simply might not acknowledge it. In other words, they might say that they engaged in a particular activity related to their child’s education when in fact they did not.
Unfortunately, there are few ways of quantifying social desirability bias in a survey, unless researchers have deliberately built the necessary metrics into the survey itself. In the case of the NHES survey used in this study, they did not. There are two questions within the survey, however, that help illustrate even well-meaning parents’ capacity for exaggeration and wishful thinking. When parents in the survey were asked what kind of grades their children were earning, 77 percent said A’s or B’s. These results would make the citizens of Lake Wobegone proud. Similarly, when parents were asked how far they expected their children to go in their education, more than 70 percent said that they would finish at least a four-year degree. This is highly unlikely, considering that even among recent cohorts, college completion rates level off at about 25 percent.[viii]
To give some sense of the potential magnitude of social desirability bias on questions having to do with parental involvement, consider the results from an earlier study in which parents: a) were asked how often they read to their young children, and b) were asked to keep a “time diary” of their children’s activities during the course of randomly-chosen days. In the survey portion of the study, 47 percent of parents reported reading to their children every day. When researchers examined a random time diary day, however, they found that only 28 percent of parents had recorded that a portion of the child’s day was spent reading with parents. If parents’ survey responses had been accurate, the time diaries should have shown roughly 47 percent of children being read to on the randomly chosen day. Thus, parents may have exaggerated the frequency of daily reading with children by as much as 20 percentage points.[ix]
What are the implications of social desirability bias for the results in this study? Even the relatively low estimated levels of parental involvement reported above may be exaggerated—perhaps significantly. Accordingly, the numbers reported here should be considered upper limits on parental involvement, not averages around which real-world behavior may fluctuate up or down. In fact, one would not be surprised to find actual levels of parental involvement within MPS to be 10 to 20 percentage points lower than indicated by the estimates presented above.
The results presented in the foregoing pages paint a discouraging picture. As noted, an estimated 34 percent of MPS parents actively choose a school for their child(ren), rather than simply settling for the closest option in their neighborhood. This means, of course, that an estimated two-thirds of MPS parents do not participate in the choice system at all. Beyond this, fewer than half of parents who do choose make a choice from among two or more schools (as opposed to considering only one). Of those, about two-thirds consider academic/performance criteria in making their choice.
By the time one arrives at this third cut at the data, only 10 percent of parents remain—that is, only 10 percent of parents consciously choose a school for their child, do so from at least two options, and consider academic/performance criteria in the process. Under the circumstances—with roughly 90 percent of parents either not choosing at all, or choosing but not applying particularly rigorous criteria—it seems unlikely that MPS schools are feeling the pressure of a genuine educational marketplace as a result of public school choice. Their performance certainly suggests not.
The data on parental involvement are equally discouraging. According to the estimates given above, only about one-third of parents are highly involved in their children’s education at the school site. With respect to at-home involvement, the figures vary between roughly 40 and 50 percent, depending on the student’s age. But parents who are at least moderately involved at home and highly involved at school are scarce indeed, constituting no more than one quarter of the parent population, and perhaps as little as 10 percent (again, depending on student age). The reader should recall, too, that these estimates almost certainly overstate the extent of parental involvement within MPS—perhaps dramatically—because of social desirability bias.
None of this is to suggest that MPS should abandon efforts at promoting public school choice and parental involvement. Regardless of their impact on student performance, these two reforms are worthy undertakings. Public school choice helps level the playing field between parents and schools, creates incentives for parents to become more engaged consumers of education, and creates the opportunity to find a better match between student and school than may be available in systems that do not offer choice. Parental involvement explicitly identifies parents as partners in the education enterprise, and makes clear the responsibilities they bear in ensuring that children get a good education. In short, both reforms reject the idea of parents as passive observers of the public education system. These reforms encourage parental empowerment, engagement, and even accountability.
The real appeal of these initiatives, however, lies in their potential to improve student outcomes and the quality of schools. In the MPS system, however, that potential appears not to have been realized. When it comes to public school choice, the estimates presented above indicate that few parents are sufficiently invested in the choice process to create the kind of serious pressure on individual schools that would result in necessary, dramatic improvements. With respect to parental involvement, too, estimated levels of parental engagement within MPS are almost certainly inadequate to have a meaningful impact on student performance. In both cases, then, one sees clear limits to the leverage that parent-oriented reforms can exert over educational outcomes within MPS.
This may be just as well. Relying on public school choice and parental involvement to reclaim MPS may be a distraction from the hard work of fixing the district’s schools. In fact, both theories of reform assume the existence of a sizeable core of good schools from which parents can choose, and on which parents can believe that their time and effort are not being wasted. There is no shortage of ideas for developing such a core of schools, even in a district the size of MPS, even in a major urban center with all of its challenges.[x] The question is whether the district, its schools, and its supporters in Madison are prepared to embrace reforms more radical than public school choice and parental involvement. Given the high stakes involved, district parents should insist on nothing less.
The data analysis in this study was based on two sources: the Census Bureau’s 2005 American Community Survey (ACS), and the 2003 Parent and Family Involvement in Education survey, which is part of the National Household Education Surveys Program (NHES) funded by the United States Department of Education. The ACS was the source for the Milwaukee Public Schools (MPS) demographic data used in the study. The NHES survey was the source for national data on parental choice and parental involvement in education for children in grades kindergarten through eight.
As noted in the text, previous research has shown a variety of demographic variables to be correlated with important aspects of both parental choice and parent involvement in education. The intent of this study was to explore the impact of those variables on choice and involvement in a specific education context—the MPS system. The particular demographic variables considered in the study were:
These variables were found to be particularly influential determinants of parental involvement in a U.S. Department of Education study using the same survey questions relied upon in developing the estimates in this paper. (See Christine Winquist Nord and Jerry West, Fathers’ and Mothers’ Involvement in Their Children’s Schools by Family Type and Resident Status, United States Department of Education, Office of Education Research and Improvement, NCES 2001-032, May 2001, Table B1, p.83.) Furthermore, the values for these variables within MPS differ significantly from those within the national dataset, meaning that failure to adjust for them in the estimation process could result in significantly biased estimates.
The basic methodological approach of the study was to weight the national data set (the NHES survey) such that its demographic profile on the four variables just identified matched that of the MPS district. The estimates in the text were then derived by performing straight frequency calculations on the variables of interest in the weighted data set.
The weights were calculated using 2005 American Community Survey Public Use Micro Data for the city of Milwaukee. The micro data were filtered to ensure inclusion only of households with related children enrolled in grades one through twelve. Then, each child case in the dataset was classified according to whether or not:
This classification scheme resulted in a total of 16 possible demographic categories for each child case. Weights were determined by comparing the frequency of each of the 16 categories within Milwaukee to the frequency of the same categories within the NHES dataset. Each case within the survey dataset was then weighted, using one of the 16 category weights. Once this weighting process was complete, the demographics of the weighted NHES data matched the actual demographics within MPS on the four variables given above. The estimates generated from the weighted data set, then, can be considered a reasonable approximation of those that would be generated using actual MPS data.
The discussion of on-site parental involvement in the text required interpolation of parent participation in an activity with frequency levels somewhere between parental frequency of: a) attendance at a general school meeting, and b) volunteering at the child’s school. The actual weighted frequency of the former in the NHES data is 83 percent; of the latter, 30 percent. Splitting the difference between these two numbers, one can assume that roughly 55 percent of parents would engage in an activity requiring a level of effort greater than attending a general meeting but less than volunteering at school.
The question, then, becomes how many of these parents would also: a) attend a parent-teacher conference, and b) attend a school event in support of their child. Parents engaged in all three would be classified in the text as “highly involved.”
The NHES data indicate that 58 percent of parents who attend general school meetings also attend parent-teacher conferences and school events in support of their child. Furthermore, 75 percent of parents who volunteer at their child’s school also attend parent-teacher conferences and school events. (This lends support to the idea that parents who volunteer at school are more highly involved overall than parents who attend general school meetings, other things being equal.) Splitting the difference again, one can assume that roughly 65 percent of parents who would engage in a hypothetical activity requiring effort levels halfway between attending a general school meeting and volunteering at the school would also: a) attend a parent-teacher conference, and b) attend a school event in support of their child.
With this number—65 percent (66.5 percent, to be exact)—the estimate in the text can finally be derived. This is done by noting that 51 percent of parents in the NHES dataset attended parent-teacher conferences and attended school events on behalf of their child. If 66.5 percent of these parents also engaged in the hypothetical third activity described above, then:
.51*.665 = .339 = 33.9 percent
of parents would be estimated to engage in all three activities. This, therefore, was the estimate used in the text for the proportion of parents who could be considered highly involved at the school site.
The question of which parents would engage in a hypothetical third activity (above and beyond attending a parent-teacher conference and a school event) becomes relevant when one attempts to calculate the percentage of parents who are highly involved at school and moderately to highly involved at home. This calculation requires one to sort parents into two groups—those who are highly involved at school, and those who are moderately to highly involved at home—and determine the extent to which the groups overlap. Establishing the size of the latter group is simple; it can be determined based on parents’ survey responses in the NHES data. Establishing the size of the former group is problematic, however; obviously, parents did not respond to a question about whether or not they had participated in the hypothetical activity postulated in this study.
In order to address this issue, participation in the hypothetical activity was assigned at random to exactly 66.5 percent of parents who reported having attended a parent-teacher conference and having attended a school event in support of their child. This random assignment allowed for an unbiased (albeit crude) designation of all parents in the data set as members of one of two groups—those who exhibit high involvement at school, and those who do not. With the size of the former group estimated in this manner, it became possible to estimate the size of the parent population exhibiting high levels of involvement at school and moderate to high levels of involvement at home.
[i] The American Community Survey data classify households by, among other criteria, the presence or absence of related children under 18 years of age. For this reason, the analysis in the text was not extended to children older than 17 years.
[ii] The research literature consulted for this portion of the study included Chandra Muller, “Parents and Schools,” Discovery Magazine, University of Texas, Vol. 14, No.3, pp.31 – 35; Douglas Downey, “Parental and Family Involvement in Education,” Chapter 6 in School Reform Proposals: The Research Evidence (Charlotte, NC: Information Age Publishing, 2002), edited by Alex Molnar; State of Michigan Department of Education, ‘What Research Says About Parent Involvement in Children’s Education,” March 2002, available on-line at: http://tinyurl.com/9p9jj; Chad Nye, Herbert Turner, and Jamie Schwartz, Approaches to parental involvement for improving the academic performance of elementary school children in grades K-6 (London: The Campbell Collaboration, 2006), available on-line at: http://tinyurl.com/ywbbp6; William H. Jeynes, Research Digest: Parental Involvement and Student Achievement: A Meta-Analysis, FINE Network: Harvard Family Research Project, December 2005, available on-line at: http://tinyurl.com/2hay8g; and Desforges, “The Impact of Parental Involvement,” op. cit.
[iii] The specific activities were: telling the child a story; doing arts and crafts with the child; involving the child in household chores; working on a project with the child; and playing board games or puzzles with the child.
[iv] The specific activities were: visiting a library; going to a play, concert, or other live show; visiting an art gallery, museum, or other historical site; and visiting a zoo or aquarium.
[v] The specific activities were: going to a play, concert, or other live show; visiting an art gallery, museum, or other historical site; attending an event sponsored by a community or ethnic group; and attending an athletic or sporting event outside of school in which the student was not a participant.
[vi] The derivation of this final set of estimates is described in the Methodology Appendix.
[vii] Brian Duff, Michael J. Hanmer, Won-Ho Park and Ismail K. White, “Good Excuses: Understanding Who Votes with an Improved Turnout Question,” Public Opinion Quarterly, Vol. 71, Number 1, 2007, p.69.
[viii] See the 2005 American Community Survey for the United States as a whole, Table B15001, B15001. SEX BY AGE BY EDUCATIONAL ATTAINMENT FOR THE POPULATION 18 YEARS AND OVER.
[ix] See Sandra L. Hofferth, “Response Bias in a Popular Indicator of Reading to Children,” Sociological Methodology 36 (1), December 2006, pp.310-315.
[x] See, for example, Samuel Casey Carter, No Excuses: Lessons from 21 High-Performing, High-Poverty Schools (Washington, D.C. Heritage Foundation, 2001); Sammis White, “The Achievement Gap in Milwaukee Public Schools,” Wisconsin Policy Research Institute; and Alan M. Blankstein, Failure is not an Option (Thousand Oaks, CA: Corwin Press, 2004).
©2007 Wisconsin Policy Research Institute, Inc. P.O. Box 487 Thiensville, WI 53092