Homeschool advocates often champion studies they claim show that homeschooled students score in the 87th percentile, far above average, as proof of the success of homeschooling. Unfortunately, these studies generally do not use random samples or correct for background factors. This means that the results cannot be generalized from to all homeschoolers, and that we cannot know whether the data shows the consequence of homeschooling or of some other factor, such as socio-economic status.
In this article, we will be turning to some homeschooling data from Alaska. This data is closer to a random sample than just about any other data out there, and it also allows us to correct for students’ economic background. This data allows us to pose some interesting questions and leads to some interesting findings—findings that may be surprising, and that may challenge some of the current narratives about homeschooling.
I do want to be clear here: I am not a statistician. This data really needs to be examined by a statistician, and analyzed in greater depth than I have either the time or ability to offer. There are lots of factors—race, for one—that I am not taking into consideration here. There is a lot more that can be done with this data than I have the time and ability to do. This piece is in no sense a final word on the Alaska data. It is simply an amateur scholar’s cursory look at some interesting numbers.
Alaska’s Correspondence Schools
Parents homeschooling their children in Alaska have several options. They can homeschool under the state’s homeschool law, which is probably the most minimalist in the entire country—no notification, no parent qualifications, no required days of teaching, no required subjects, no assessments. They can also homeschool under the private school law, which requires paperwork, attendance records, and standardized tests, or under the private tutor law, which requires teacher certification (not surprisingly, homeschooling under these two options is not very common). However, the evidence suggests that the majority of homeschoolers actually choose not to homeschool under the state’s homeschool law and instead use a fourth option—the correspondence program.
In 1997, the Alaska legislature passed the state’s minimalistic homeschool law, completely waiving any and all requirements for homeschoolers. With this change, some enterprising educators saw an opportunity. That same year, the school district in the small interior town of Galena, Alaska, founded Interior Distance Education of Alaska (IDEA), a correspondence program specifically targeted toward homeschool families. More such programs sprung up across the state, each operating as an extension of a public school district or as a charter school and offering perks such as computers, testing, assigned certified teachers for support, and money for textbooks, classes, and lessons. Some homeschool parents resisted these programs, seeing them as problematic extensions of state control, but an estimated 75% of the state’s homeschooled children were quickly enrolled in them. (For more of this history, see here and here.)
These correspondence programs offer some serious perks for homeschooling parents. Parents choose their curriculum, their teaching methods, and the subjects their children study. These are not public school at home programs the way we generally think of them, with students using state curriculum or watching videos on the internet. Rather, these programs are specifically designed for conventional homeschoolers. They are bound by state law, and therefore are fairly homogenous in their requirements, which generally run as follows: First, at the beginning of that year, parents must create an individual education plan for each student—an outline of subjects and curriculum—and have that plan approved by the correspondence program. Certain subjects are required, but what each child’s individual education plan will look like is very much up to the parents. Second, each child is assigned a certified teacher who is required to make contact with the child at least once a month. This teacher is more for guidance than for teaching. Third, parents must submit quarterly progress reports for each child. Fourth, at the end of the school year (for grades 3 through 10) each child must take a standardized test at a testing center arranged by the correspondence school. There is no minimum score, though correspondence programs may choose to dis-enroll students if their parents do not follow the requirements of the program. Finally—and here is where the perks really come in—each parent receives around $2000 per child per year for use on things like textbooks, classes, and tutors.
If you’re interested in reading more about these programs, let me point you in a few directions. The Alaska Department of Education and Early Development maintains a Correspondence Program Directory and a Summary of Parental Rights and Expectations. Next, let me point you to three of the larger of these programs. First, the Interior Distance Education of Alaska program is the largest in the state. Here is its School Report Card for the 2011—2012 School Year. Next, Kenai Peninsula Borough School District Connections Alaska Homeschool Program is a second correspondence program that is fairly representative. Here is its Homeschool Program Handbook for 2012—2013. Finally, Yukon-Koyukuk School District Raven Correspondence School is a good third example. You can read its 2013—2014 Parent-Student Handbook.
How many of Alaska’s homeschooled students are enrolled in these correspondence programs? In the 2012-2013 school year, there were almost exactly 11,000 homeschooled students participating in a total of 28 different correspondence programs across the state. This is a large number—if Alaska had the same rate of homeschooled students as in the rest of the country, we would expect there to be only around 5,000 homeschooled students in the state total. Now, Alaska as a state has some particularities. Many areas are very remote, and schools may not always be available. Given this and other particularities, we would expect Alaska to have a higher number of homeschoolers. What we do not know is how many students are being homeschooled under the conventional homeschool option, as it does not require any form of notification (or assessment, or anything else). Because of this, we cannot know what percentage of the total homeschool population is enrolled in correspondence programs. Still, given that a full 11,000 students are enrolled in these programs, and given that a perusal of homeschooling websites and resources for the state suggests that these programs are widely used by conventional homeschoolers, it is probably safe to assume that the majority of homeschooled students in Alaska are indeed enrolled in these correspondence programs.
Alaska’s correspondence programs are run by public school districts or charter schools. Some (though not all) have local home offices or learning centers that provide classes or various group opportunities. Because these programs technically operate as public schools or charter schools under state law, they are required to put together an annual report at the end of each school year. These reports include testing data for each grade where testing is required—grades 3 through 10—broken down by things like age, gender, and socio-economic status, and are released to the public. This testing data will serve as the basis of several future blog posts.
Many of Alaska’s homeschoolers take advantage of these correspondence programs while others instead homeschool under the minimalist homeschool law. These two populations may be distinctly different, though the trouble is that we cannot say for sure how. It could be that homeschool parents who place less emphasis on their children’s educational wellbeing are more likely to homeschool under the minimalist homeschool law, but it could also be that those parents who feel less confident in their teaching ability are more likely to opt for the correspondence school option. It could be that those enrolled in the correspondence school option are predominantly those forced to homeschool by distance rather than those who choose to homeschool out of an array of options. This seems doubtful, however, given that the major statewide correspondence programs very clearly cater to conventional homeschoolers rather than those who come to homeschooling because it is their only option. It could be that those who choose to homeschool under the minimalist homeschool law are mainly the anti-establishment types who bristle at any thought of government country. The point I’m trying to make here is simply that we don’t know.
While we do have testing data for the large number of students enrolled in state correspondence programs and educated at home, we cannot assume that this data is representative of all Alaska homeschoolers—and we also cannot assume that it is representative of all homeschoolers in the United States. Still, that does not mean that the data has nothing to say. And with that, we’ll start delving into this data.
The Basic Academic Data
To start with, here is a listing of all Alaska correspondence schools that enrolled students during the 2011-2012 school year, and the number of students enrolled in each (there were 11007 students enrolled total):
As you can see, Interior Distance Education of Alaska (IDEA), which is located in the small interior town of Galena, Alaska, is by far the largest correspondence program. This program was founded in 1997 specifically to cater to the state’s homeschool students. IDEA’s design was soon copied by other schools, and numerous such programs sprang up across the country. Some of the smaller correspondence schools are local to their own district and cater to their own students who may need some form of distance learning, but the larger ones are predominantly statewide in their reach and focus very specifically on enrolling conventional homeschool students.
In order to collect the test scores for these correspondence schools, I went through every school in the state’s Correspondence Program Directory. I looked each one up here to determine its district and then pulled up its state report card for 2011-2012. There were a total of 32 correspondence schools in the directory. Four of them did not submit reports for 2011-2012. I contacted these programs using the contact information provided in the directory and was told that these programs did not have any students during the 2011-2012 school year. In addition, four more of the correspondence programs, enrolling a total of 30 students, did not turn in reports for 2011-2012—or at least, these reports are not available through the state website. I have not contacted these schools, but it is possible that they simply did not have any students in the grades with required testing (grades 3 through 10). Either way, 30 students out of a total of 11007 is not large enough to make a difference. You can view my summary of the schools and their total enrollment here.
We will start first with the reading scores. You can see my compilation of the correspondence schools’ reading scores here. In the chart below, I have compared the correspondence school reading scores with the average reading scores for the entire state.
As you can see, the correspondence school students score better in reading than the state average. This reflects the large amount of data that has shown a homeschool advantage in reading comparative to other areas.
Next we turn to the correspondence school writing scores. You can see my compilation of the correspondence schools’ writing scores here. In the chart that follows, I have compared the correspondence school writing scores with the average writing scores for the entire state.
As you can see, the correspondence school students do right about just as well as the state average. They’re more likely to be proficient, and slightly less likely to be either advanced or below.
Finally we turn to the math scores. You can see my compilation of the correspondence schools’ math scores here. In the chart that follows, I have compared the correspondence school math scores with the average math scores for the entire state.
As you can see, the correspondence school students underperform the state average in math by a decent margin. This again is not surprising, actually, because it reflects wider data that suggests that homeschooled students have less of an advantage in math than in other subject areas.
Now I’m going to sum up all of this data in one chart. To do this I will combine “advanced” with “proficient” and “below” with “far below.” This allows us to compare the correspondence school scores with the state average scores in the categories “proficient and above” and “below proficient.” Here is what we find:
So what we find is that the correspondence school students have an edge over the state average in reading, but that the state average has an edge over the correspondence school students in math. Again, this appears to confirm something scholars have long suggested—that homeschooled students tend to comparatively overperform in reading and comparatively underperform in math. This makes sense, if you think about it—it’s not so hard to teach a child to read and get them started on a pile of books, especially when they have more time for reading than public school students may. It’s a lot harder to teach them a subject like math, which really necessitates an actual teacher and an understanding of mathematic principles and good pedagogy for teaching the subject.
But what this analysis leaves out is background factors. If the homeschool students enrolled in these correspondence schools are from families that are better educated or wealthier, they should be compared not with the state average but rather with other students from families that are better educated or wealthier. Similarly, if the students in these correspondence schools are from families that are less educated and not as well off, they should be compared with other students from families that are less educated and not as well off. Serendipitously, each report for each correspondence school breaks the test scores down into “economically disadvantaged” and “not economically disadvantaged,” which we can compare to data on the state average for these groups.
Do Background Factors Matter?
We have seen that that Alaska students enrolled in correspondence programs and educated at home—a total of 11,000 students—score above than the state average in reading and below the state average in math. Now we turn to a simple question: When it comes to Alaskan children homeschooled through correspondence schools, do background factors matter? When we break down these students into economically disadvantaged and not economically disadvantaged, we find that the answer is a most emphatic yes.
Because there are limits on the amount of time I have to play with this data, I have chosen to combine four of six largest correspondence programs and leave aside the remaining programs. I intended to use the six programs with over 500 students each, but I found that two of the six did not appear to have kept economic data on their students. This left me with four. These four schools together enrolled 6862 students, or well over half of the 11,000 total correspondence school students. However, testing is only required in grades 3 through 10. A total of 86.3% of the students in these grades, or 3750, were tested. That 3750 is our sample size. Of these 3750 students, 1430 (or 38%) were economically disadvantaged while 2320 (or 62%) were not economically disadvantaged.
In every area—reading, writing, and math—the economically disadvantaged correspondence school students had lower scores than those who were not economically disadvantaged. This is to be expected, given that economically disadvantaged public school students have lower scores than public school students who are not economically disadvantaged. Still, this finding contradicts many homeschool advocates’ assertions that things like parental education or economic status have little to no impact on homeschooled students’ academic success.
The following chart summarizes these findings.
As you can see, the difference in scores between correspondence school students who are economically disadvantaged and those who are not economically disadvantaged is most pronounced in mathematics and least pronounced in reading.
What’s interesting is that there is a smaller difference between these two groups than there is between the state averages for those two groups (you can view a chart for this later comparison here). While there is a 6% difference in the number of correspondence school students proficient or above in reading depending on economic background, there is a 21% difference between the state average for those two groups. While there is a 10% difference in the number of correspondence school students proficient or above in writing depending on economic background, there is a 23% difference in the state average for these two groups. Finally, while there is a 16% difference in the number of correspondence school students proficient or above in math depending on academic background, there is a 23% difference in the state average for these two groups. This difference, not surprisingly, is most pronounced in reading and least pronounced in math.
Why is the difference between the scores of economically disadvantaged correspondence school students and those of correspondence school students who are not economically disadvantaged smaller than the difference between the state average scores? As we shall see, it is because economically disadvantaged correspondence school students overperform as compared to their peers while those who are not economically disadvantaged actually underperform as compared to their peers.
Economically Disadvantaged Homeschoolers
Now we turn directly to economically disadvantaged homeschool students. The data we have available allows us to compare the scores of economically disadvantaged correspondence school students with the state average scores for economically disadvantaged students.
We start with reading scores.
Economically disadvantaged correspondence school students outperform the state average in reading for economically disadvantaged students by a large margin.
Next we turn to writing scores.
Economically disadvantaged students also score better on writing than the state average for economically disadvantaged students, though the difference is not quite as large as for writing.
Finally we turn to math.
In contrast to both reading and writing, economically disadvantaged correspondence school students underperform the state average for economically disadvantaged students, though the difference is not large.
Here are the scores are summarized, with the advanced and proficient scores combined and the below and far below scores combined.
What do these findings tell us? There is, of course, the obvious: These findings tell us that economically disadvantaged students who are homeschooled through correspondence schools in Alaska outperform the state average for economically disadvantaged students in reading and writing but not in math. (This finding appears, once again, to confirm the research suggesting that homeschoolers comparatively overperform on reading and comparatively underperform on math.) Of course, while these students do quite well in reading (and to a slightly lesser extent in writing) when compared to their economic bracket, they do still underscore the total state average.
Who are these economically disadvantaged parents who take their students out of public school and homeschool them through enrollment in correspondence schools? Is there a difference between economically disadvantaged homeschool students who are enrolled in correspondence schools and those who are not? How did these students perform in school before they were removed to be homeschooled? Or, if they were homeschooled from kindergarten, how would they have scored had they attended public school? In other words, the very fact that their parents chose to homeschool them and enroll them in a correspondence program makes economically disadvantaged correspondence school students different from, say, economically disadvantaged public school students. What is that difference? Are their parents more involved and more dedicated to their education? Does this alone explain the difference in scores, or does homeschooling itself improve these students’ scores?
Two more things I want to point out. First, somewhere around 80% of the students in these correspondence schools are white as compared to 50% of students in the state’s public schools. This may help explain some of these numbers, though I cannot say for sure. Second, it is likely that the correspondence school students who are economically disadvantaged have higher incomes than the average for economically disadvantaged students—i.e., that they are on average closer to the line between economically disadvantaged students and not economically disadvantaged students than are other economically disadvantaged students. This would make sense, and is actually rather intuitive—in order to homeschool, parents need to be able to afford for a parent to stay at home. It could be, too, that economically disadvantaged correspondence school students’ parents are better educated than the average. If couples who would not otherwise be economically disadvantaged are giving up a second income to homeschool, and thus becoming economically disadvantaged when they do not have to be, this could make sense.
NON Economically Disadvantaged Homeschoolers
We now know that economically disadvantaged correspondence school students overperform as compared to the state average for economically disadvantaged students in every area except math. But what about those students who are not economically disadvantaged? We now turn to the scores of non economically disadvantaged students, comparing correspondence school students with the state average. What we find is that the results are quite different from what we find for economically disadvantaged students.
We begin with the reading scores.
This is actually a really surprising finding. Studies have consistently indicated that homeschooled students overperform in reading when compared to the national average, and indeed, we saw this hold true when we looked at data for economically disadvantaged correspondence school students. But when we compare correspondence school students who are not economically disadvantaged with the state average for students who are not economically disadvantaged (rather than merely with the state average), we find that the correspondence school students actually slightly underperform in reading. This is an incredibly surprising finding, and it suggests that much if not all of homeschooled students’ seeming overperformance in reading may be due not to homeschooling but rather to their socioeconomic background.
Now we move on to writing scores.
Here the difference is slightly more dramatic: Correspondence school students who are not economically disadvantaged, when compared with the average for all students who are not economically disadvantaged, underperform in writing.
Finally we turn to the math scores.
Here the difference is most dramatic. When compared with their economic peers, non economically disadvantaged correspondence school students underperform in math by a serious margin. There is a 14% difference in the number of students who are advanced in math—and by that I mean that while a full 43% of the state average are advanced in math, only 29% of correspondence school students are advanced in math.
Here are the scores are summarized, with the advanced and proficient scores combined and the below and far below scores combined.
This data challenges homeschool advocates’ assertion that homeschooling by itself makes students succeed. It suggests that much of homeschooled students’ seeming advantage in test scores actually reflects socioeconomic status rather than the effects of homeschooling.
But once again, we are left with many questions. Why were these students removed from public school to be homeschooled? We know that many students are homeschooled from kindergarten on and that some students are homeschooled for reasons other than academics, but it is also the case that some number of homeschooled students were withdrawn from public school because school wasn’t working for them. Perhaps they were falling behind academically as a result of bullying or bad teachers. Perhaps some of these non economically disadvantaged students were homeschooled because they live in bad school districts, and would underperform the state average for their economic bracket if they were placed in school. These sorts of factors might help explain why these students underperformed as compared to their peers.
What of other states? Can we compare this data with data from elsewhere? Unfortunately, the points of comparison we have are limited. The only other state that currently collects and releases homeschool testing data is Arkansas. (For an example of such a report, see here.) The Arkansas data suggests that homeschooled students overperform in reading by a large margin and slightly underperform in math. Unfortunately, we do not have the data to correct for background factors. It is quite possible that when background factors are corrected for, we will see a similar pattern to that in Alaska, and find that the overperformance in math is more due to differences in economic background than it is to homeschooling. Given that the uncorrected Arkansas data looks very similar to the uncorrected Alaska data, this would not be a surprising finding. Of course, we might also find that homeschoolers in Arkansas score differently from homeschoolers in Alaska. Unfortunately, we simply have no way to know.
What findings can we take away from the Alaska data? First I think we should be clear on what we cannot take away from this data. It would be wrong to read this data and determine that homeschooling elevates economically disadvantaged correspondence school students’ reading and writing scores. It would also be wrong to read this data and conclude that homeschooling depresses non economically disadvantaged correspondence school students’ writing and math scores. The reason we cannot make these conclusions is that economically disadvantaged correspondence school students are not identical to economically disadvantaged public school students and non economically disadvantaged correspondence school students are not identical to non economically disadvantaged public school students. Why are they not identical? Because, quite simply, their parents removed them from public school to homeschool them through correspondence schools, and that makes them different.
However, there are a few things that we can take away from this data. First, this data appears to confirm the studies that have found that homeschooled students comparatively overperform in reading and comparatively underperform in math. In fact, this data suggests that homeschoolers may have a math problem. This is an important finding, as understanding the existence of a problem is the first step to fixing it. Homeschooling parents should know this finding so that they can be especially on guard when it comes to their children’s math education, ready to try something new or hire a tutor if needed. Second, this data confirms the incredible importance of background factors in affecting homeschool performance. It would be dishonest to look at a study of homeschooled students in upper middle income brackets and somehow assume that their test scores can be generalized to all homeschoolers across all income brackets. These background factors do not just affect public school students’ performance, they also affect homeschool students’ performance.
And there you have it. The Alaska data. Of course, what we really could use is a statistician’s analysis of this data. They might be able to tease out even more interesting nuance and information, and also to point to just how significant various differences in scores are.