Homeschool advocates often champion studies they claim show that homeschooled students score thirty percentile points above average as proof of the superiority of homeschooling. Unfortunately, these studies have some serious flaws—they do not use random samples and they do not correct for background factors. For an overview of what we do and do not know about homeschooling and academics, see our introduction here. Given the flaws that characterize most studies of homeschooling, I was fascinated when I stumbled upon some data from Alaska.

There is a lot more here than what I will examine in this post. I am not formally trained in statistics and I will not even touch on what what the data says about race or gender. There are also limits to this data. We only have testing data from those Alaskan homeschoolers who participated in Alaska’s popular and innovative correspondence school programs. While this covers the majority of homeschoolers in Alaska, it means we don’t have data on the performance of the remainder of Alaska’s homeschoolers. Still, this data avoids some of the flaws of other studies of homeschooling and academics and is vastly more interesting than anything else I have seen.

Alaska’s Correspondence Schools

Alaskan parents 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. However, the majority of homeschool parents in Alaska choose to homeschool under one of Alaska’s many correspondence programs.

These correspondence programs do not use videos or mail-in workbooks, and they do not replicate the public school curriculum at home. They are designed with conventional homeschoolers in mind and allow parents to choose their own curriculum and plan out their own school years. They require a yearly education plan for each student, quarterly progress reports, and annual testing. In exchange, each parent receives around $2000 per child per year for use on things like textbooks, classes, and tutors.

In the 2012-2013 school year, almost 11,000 homeschooled students participated in 28 different correspondence programs across the state. 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 poverty level, and are released to the public.

### The Basic Scores

Throughout the remainder of this article, I will compare the scores of students homeschooled through Alaska’s correspondence schools with Alaska’s public school average. Students’ scores for reading, writing, and math are broken down into “advanced,” “proficient,” “below,” and “far below.”

In the following chart, I have combined “advanced” with “proficient” and “below” with “far below”:

Students homeschooled through Alaska’s popular correspondence schools do better in reading than the state average but worse in math. This difference in performance further confirms studies that have found a homeschool advantage in reading and a homeschool disadvantage in math.

### Background Factors

Very few studies of homeschooled students’ academic performance have accounted for background factors. We know that public school students whose parents have college degrees tend to do better academically than those whose parents do not have college degrees. When looking at the academic scores of homeschooled students, we need to ask how background factors like parental education affect children’s performance. Otherwise we cannot locate the effect of homeschooling from the effect of various background factors.

By breaking student test scores down by whether they are “economically disadvantaged” or “not economically disadvantaged,” we can examine how parental income affects homeschooled students’ test scores. Roughly one third of the students were economically disadvantaged and roughly two-thirds were not economically disadvantaged.

There is a clear academic difference between economically students homeschooled through Alaska’s correspondence schools who are and are not economically disadvantaged. The difference between these two groups in each subject is smaller than the difference between public school students who are and are not economically disadvantaged (you can view that difference here). Unlike for public schooled students, however, the difference in scores varies by subject—it is most significant in math and least significant in reading.

### Economically Disadvantaged Homeschoolers

In this section we will compare the scores of economically disadvantaged students homeschooled through Alaska’s correspondence schools with the state average scores for economically disadvantaged students.

Here are the scores are summarized, with “advanced” and “proficient” combined and “below” and “far below” combined.

When we look only at economically disadvantaged students, we find that those homeschooled through Alaska’s correspondence schools outperform the state average in writing and, especially, in reading, but slightly underperform the state average in math.

### NON Economically Disadvantaged Homeschoolers

In this section we will compare the scores of non-economically disadvantaged students homeschooled through Alaska’s correspondence schools with the state average scores for non-economically disadvantaged students. (The dividing line here is 200% of poverty.)

Here are the scores are summarized, with “advanced” and “proficient” combined and “below” and “far below” combined.

When we look only at students who are above 200% of poverty the reading difference disappears, we find that those homeschooled through Alaska’s correspondence schools perform slightly worse in reading, worse in writing, and significantly worse in math than the state average.

### Analysis

Data is all well and good, you say, but what does it all mean?

**Conclusion 1: Background factors matter, even in homeschooling.** This data confirms something that should not be surprising: Students homeschooled by wealthier parents do better academically than those homeschooled by poorer parents. It is likely that other background factors, like parental income, race, and family stability also affect homeschooled students’ scores. This conclusion is important because some homeschool advocates mislead prospective homeschool parents by telling them that factors like household income do not affect homeschooled students’ academic performance.

**Conclusion 2: The idea that homeschooling results in higher test scores is a myth.** Overall, students homeschooled through Alaska’s correspondence schools outperformed in reading and underperformed in math. In other words, homeschooling may change students’ academic performance, resulting in higher reading scores and lower math scores, it does not raise their overall scores. Further, homeschooled students above 200% of poverty actually underperformed their peers in every subject and students below 200% of poverty still scored below their peers in math.

**Conclusion 3: Homeschooled students’ reading advantage may be explained by background factors. **Studies have consistently found that homeschooled students are strongest in reading. Some scholars have suggested that homeschooling may offer a reading advantage. Given that homeschooled students generally have more time on their own for reading than other students, this makes intuitive sense. However, when we looked only at Alaskan students above 200% of poverty the reading difference disappeared. The homeschooled students scored no better, and in fact scored marginally worse, than their public schooled peers. When looking at data that suggests a homeschooling math advantage, we need to question whether this is a result of socio-economic factors rather than a result of homeschooling.

**Conclusion 4: Homeschooling Has a Math Problem. **Many studies have found that homeschooled students perform less well in math as compared to other subjects such as reading. What has been less clear, however, is whether they actually underperform public schooled students in math. Based on this data, the answer appears to be yes. Whether we look at the average scores, the scores of students above 200% of poverty, or the scores of students below 200% of poverty, homeschooled students perform worse in math than the public school average. While a full 43% of public school students above 200% of poverty are advanced in math, this is true for only 29% of their homeschooled peers.

**Conclusion 5: Something else is going on here, but we don’t know what. **While homeschooled students above 200% of poverty outperformed those below 200% of poverty, there was a large difference when each demographic was compared to its public school counterpart. When looking only at students above 200% of poverty, those homeschooled through Alaska’s correspondence schools underperformed the state average in every subject. In contrast, when looking only at students below 200% of poverty, those homeschooled through Alaska’s correspondence schools outperformed the state average in reading and writing and underperformed the state average only in math. Taken at face value, this suggests that homeschooling has better results for economically disadvantaged students than for those who are not economically disadvantaged.

There are a variety of factors that may help explain why this odd finding. The most likely explanation is that economically disadvantaged homeschoolers differ from economically disadvantaged public school students in additional factors such as race or family stability. Because homeschooling families more frequently exist on one income and have a large number of children, it may be that there is a discrepancy between income level and perceived socio-economic status for many homeschool families below 200% of poverty. In other words, parents may be better educated and have a higher level of access to social capital than where they fall relative to poverty level may suggest. It may also be that homeschooled students below 200% of poverty tend to be clustered up around 150% of poverty while public school students below 200% of poverty tend to be more spread out.

Any explanation that credits economically disadvantaged homeschooled students’ elevated scores to the benefits of homeschooling as an educational method must explain why homeschooling does not provide that same benefit for students who are not economically disadvantaged. It could be that the reason is that economically disadvantaged students are more likely to be in school districts with fewer resources while students who are not economically disadvantaged students are more likely to be in school districts with more resources. There are a variety of possible explanations, but this is a finding that really does demand an explanation, not a slogan or pithy response.

And there you have it. The Alaska data.

## The Alaska Data and Homeschool Academics

Homeschool advocates often champion studies they claim show that homeschooled students score thirty percentile points above average as proof of the superiority of homeschooling. Unfortunately, these studies have some serious flaws—they do not use random samples and they do not correct for background factors. For an overview of what we do and do not know about homeschooling and academics, see our introduction here. Given the flaws that characterize most studies of homeschooling, I was fascinated when I stumbled upon some data from Alaska.

There is a lot more here than what I will examine in this post. I am not formally trained in statistics and I will not even touch on what what the data says about race or gender. There are also limits to this data. We only have testing data from those Alaskan homeschoolers who participated in Alaska’s popular and innovative correspondence school programs. While this covers the majority of homeschoolers in Alaska, it means we don’t have data on the performance of the remainder of Alaska’s homeschoolers. Still, this data avoids some of the flaws of other studies of homeschooling and academics and is vastly more interesting than anything else I have seen.

Alaska’s Correspondence Schools

Alaskan parents 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. However, the majority of homeschool parents in Alaska choose to homeschool under one of Alaska’s many correspondence programs.

These correspondence programs do not use videos or mail-in workbooks, and they do not replicate the public school curriculum at home. They are designed with conventional homeschoolers in mind and allow parents to choose their own curriculum and plan out their own school years. They require a yearly education plan for each student, quarterly progress reports, and annual testing. In exchange, each parent receives around $2000 per child per year for use on things like textbooks, classes, and tutors.

In the 2012-2013 school year, almost 11,000 homeschooled students participated in 28 different correspondence programs across the state. 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 poverty level, and are released to the public.

## The Basic Scores

Throughout the remainder of this article, I will compare the scores of students homeschooled through Alaska’s correspondence schools with Alaska’s public school average. Students’ scores for reading, writing, and math are broken down into “advanced,” “proficient,” “below,” and “far below.”

In the following chart, I have combined “advanced” with “proficient” and “below” with “far below”:

Students homeschooled through Alaska’s popular correspondence schools do better in reading than the state average but worse in math. This difference in performance further confirms studies that have found a homeschool advantage in reading and a homeschool disadvantage in math.

## Background Factors

Very few studies of homeschooled students’ academic performance have accounted for background factors. We know that public school students whose parents have college degrees tend to do better academically than those whose parents do not have college degrees. When looking at the academic scores of homeschooled students, we need to ask how background factors like parental education affect children’s performance. Otherwise we cannot locate the effect of homeschooling from the effect of various background factors.

By breaking student test scores down by whether they are “economically disadvantaged” or “not economically disadvantaged,” we can examine how parental income affects homeschooled students’ test scores. Roughly one third of the students were economically disadvantaged and roughly two-thirds were not economically disadvantaged.

There is a clear academic difference between economically students homeschooled through Alaska’s correspondence schools who are and are not economically disadvantaged. The difference between these two groups in each subject is smaller than the difference between public school students who are and are not economically disadvantaged (you can view that difference here). Unlike for public schooled students, however, the difference in scores varies by subject—it is most significant in math and least significant in reading.

## Economically Disadvantaged Homeschoolers

In this section we will compare the scores of economically disadvantaged students homeschooled through Alaska’s correspondence schools with the state average scores for economically disadvantaged students.

Here are the scores are summarized, with “advanced” and “proficient” combined and “below” and “far below” combined.

When we look only at economically disadvantaged students, we find that those homeschooled through Alaska’s correspondence schools outperform the state average in writing and, especially, in reading, but slightly underperform the state average in math.

## NON Economically Disadvantaged Homeschoolers

In this section we will compare the scores of non-economically disadvantaged students homeschooled through Alaska’s correspondence schools with the state average scores for non-economically disadvantaged students. (The dividing line here is 200% of poverty.)

Here are the scores are summarized, with “advanced” and “proficient” combined and “below” and “far below” combined.

When we look only at students who are above 200% of poverty the reading difference disappears, we find that those homeschooled through Alaska’s correspondence schools perform slightly worse in reading, worse in writing, and significantly worse in math than the state average.

## Analysis

Data is all well and good, you say, but what does it all mean?

Conclusion 1: Background factors matter, even in homeschooling.This data confirms something that should not be surprising: Students homeschooled by wealthier parents do better academically than those homeschooled by poorer parents. It is likely that other background factors, like parental income, race, and family stability also affect homeschooled students’ scores. This conclusion is important because some homeschool advocates mislead prospective homeschool parents by telling them that factors like household income do not affect homeschooled students’ academic performance.Conclusion 2: The idea that homeschooling results in higher test scores is a myth.Overall, students homeschooled through Alaska’s correspondence schools outperformed in reading and underperformed in math. In other words, homeschooling may change students’ academic performance, resulting in higher reading scores and lower math scores, it does not raise their overall scores. Further, homeschooled students above 200% of poverty actually underperformed their peers in every subject and students below 200% of poverty still scored below their peers in math.Conclusion 3: Homeschooled students’ reading advantage may be explained by background factors.Studies have consistently found that homeschooled students are strongest in reading. Some scholars have suggested that homeschooling may offer a reading advantage. Given that homeschooled students generally have more time on their own for reading than other students, this makes intuitive sense. However, when we looked only at Alaskan students above 200% of poverty the reading difference disappeared. The homeschooled students scored no better, and in fact scored marginally worse, than their public schooled peers. When looking at data that suggests a homeschooling math advantage, we need to question whether this is a result of socio-economic factors rather than a result of homeschooling.Conclusion 4: Homeschooling Has a Math Problem.Many studies have found that homeschooled students perform less well in math as compared to other subjects such as reading. What has been less clear, however, is whether they actually underperform public schooled students in math. Based on this data, the answer appears to be yes. Whether we look at the average scores, the scores of students above 200% of poverty, or the scores of students below 200% of poverty, homeschooled students perform worse in math than the public school average. While a full 43% of public school students above 200% of poverty are advanced in math, this is true for only 29% of their homeschooled peers.Conclusion 5: Something else is going on here, but we don’t know what.While homeschooled students above 200% of poverty outperformed those below 200% of poverty, there was a large difference when each demographic was compared to its public school counterpart. When looking only at students above 200% of poverty, those homeschooled through Alaska’s correspondence schools underperformed the state average in every subject. In contrast, when looking only at students below 200% of poverty, those homeschooled through Alaska’s correspondence schools outperformed the state average in reading and writing and underperformed the state average only in math. Taken at face value, this suggests that homeschooling has better results for economically disadvantaged students than for those who are not economically disadvantaged.There are a variety of factors that may help explain why this odd finding. The most likely explanation is that economically disadvantaged homeschoolers differ from economically disadvantaged public school students in additional factors such as race or family stability. Because homeschooling families more frequently exist on one income and have a large number of children, it may be that there is a discrepancy between income level and perceived socio-economic status for many homeschool families below 200% of poverty. In other words, parents may be better educated and have a higher level of access to social capital than where they fall relative to poverty level may suggest. It may also be that homeschooled students below 200% of poverty tend to be clustered up around 150% of poverty while public school students below 200% of poverty tend to be more spread out.

Any explanation that credits economically disadvantaged homeschooled students’ elevated scores to the benefits of homeschooling as an educational method must explain why homeschooling does not provide that same benefit for students who are not economically disadvantaged. It could be that the reason is that economically disadvantaged students are more likely to be in school districts with fewer resources while students who are not economically disadvantaged students are more likely to be in school districts with more resources. There are a variety of possible explanations, but this is a finding that really does demand an explanation, not a slogan or pithy response.

And there you have it. The Alaska data.