 Thank you, and welcome, Professor Coley and Baker. We are very pleased to hear from you on your summary of findings on the people waiting scores for education funding of the month. We're here today with the House Education Committee and the House Ways and Means Committee. No, other than just to say good morning to everybody and looking forward to hearing the findings and appreciate the fact that we're doing this with 22 House members. We will do our best to get the presentation out there and get people a chance to ask questions. And I know both committees are going to have some time with you later as well, so their follow-up questions will be able to do that. Thank you for coming. Thank you. Thank you to Representative Webb and to my guest. I'm here today with Dr. Ms. Baker and Ms. Chan Coley. Associate Professor of the University of Vermont in the Education Policy Program. Bruce and I are representing a larger team of people, which includes our colleagues, Dr. Atchison and Jesse Lemon at the American Institutes for Research. And as a team, we took on this task of evaluating the people waiting format that's included in the existing school funding so it's actually going to come off. Today's purpose of today is to provide a high-level overview of the waiting study. We want to highlight the questions that we asked, how we answered those questions, and key findings. I'm sure most of you have had a chance to start to read the report. There's lots in there, to say the least. And so our goal is to set a baseline with respect to what the high-level findings are. And again, as you indicated, I think there'll be other opportunities for us to go more in depth. But what we're hoping to do is by providing a high level today that we can establish sort of a general conversation around the report, and then we can go more specific later. The report is very comprehensive. It was our intent to look at the issues and the questions that the General Assembly posed in a very systematic and comprehensive way. And again, we'll be looking at findings in more depth as we go along. Bruce and I are going to share the presentation today, both because it's a treat to have Bruce here. Last night, I drove in this morning from New Jersey, from Rutgers University, and also to save my voice a bit. So you're going to hear it from both of us this morning. And so I'm going to start by introducing a logo back and forth that's the take-offs on. But feel free to do dark questions at either of us as we talk about it. The way it's studied is in response to a legislative request. And the Romani Agency of Education was directed in Act 173 to undertake a study to examine and evaluate three things. The current weights for economically disadvantaged students, BLL students, and secondary-level students, and whether those weights should be modified, whether or not there's a need for new cost factors and weights to be incorporated into equalized people calculation, and three, the extent to which the special education census law grant should be adjusted for differences in incidence and costs associated with students with disabilities across school districts. So those were the both guideposts for this. And those guideposts were put out in the legislative request. Our study contained, our study, we approached the study with six key objectives, right? So we had the legislative request. We then sort of honed that around six key objectives. And that was, and the intent with these objectives was contextualize sort of the formal policies that we have now. It led to the best thinking on how to design effective state-level school-financing systems. To catch the insights and voices from the field in Vermont, as we contemplate potential next steps for Vermont policy-making. And to leverage national, regional, and state data to evaluate Vermont's current policies. And so with that backdrop, we had these six objectives. The first one was develop a national profile of cost factors and funding mechanisms used in state education funding formula. Two, to obtain stakeholder perceptions and experiences with existing funding formula. Three, identify aspects of student need and local educational context that account for differences in the cost of educating students to common standards. And here's what I'm talking about today. Educating to common standards. Four, to empirically derive weights for a select set of cost factors that can be included in Vermont's school funding formula. We're also going to hear that language quite a bit today about empirically deriving as opposed to just sort of coming up with the number. Five, assessing further adjustments to the Census-based Special Education Block Grant whether those are needed and if they are needed what they might like. And six, developing simulations that can be used to predict the effect of various changes to the funding formula. And now oftentimes when we talk particularly around school funding, we talk about, what happens if we do extra Y? Part of our task, as we saw it, was to do our best to simulate what potential changes might work like on a town-by-town district by district basis. So our goal with all of this is to provide a synthetic accounting of the findings, and notice that the report doesn't sort of go six objectives, we don't talk about it that way. Our goal is to be very synthetic and say sort of what does all of this tell us about what the status of the state of the field looks like. And with the goal of providing actionable evidence that you can use in decision making. So our focus in all of our work was on what is it that we thought that we can provide to you that's actionable for decision making. And so I'm going to stand off to Bruce now. And Bruce is going to talk a little bit about, is going to help set a common understanding for the purpose and design of how we think about cost differences in the school funding formula and why we think about it that way. And the intent behind that is to set a baseline level of understanding about what the broad range of policy options might look like, and thinking through potential work forms from us. So when we, in this kind of weird field of studying school finance policy and the design of these state school finance systems to provide resources to the schools, districts and the students that they serve, we look at these systems in terms of trying to provide for equal educational opportunity to all children. When we try to kind of operationalize the definition of equal educational opportunity, we do that in terms of opportunities to achieve given outcome goals. And a lot of this comes out of kind of the emergent, last several decades of education policy where we move toward measuring outcomes, holding schools and districts accountable for different outcome measures and ideally getting students to certain outcome goals through the K-12 system so that they can persist in complete and higher ed, have equal opportunity beyond the K-12 system. So we frame the analyses, we frame our policy goals in terms of ensuring equal educational opportunities for all students. And then we tend to operationalize that in terms of equal opportunity to achieve certain measured outcome objectives, measured outcome objectives that we perceive to be important and useful for kind of predicting later life outcomes for all students. Students come to school, as the slide said, students come to school with different backgrounds, different learning needs, and some of these operate at the individual student level, some of these are about the broader social context of schooling, which we'll hit on the next slide. And these different backgrounds require different levels and different mixes of educational resources to get these kids equal opportunity to achieve the same outcomes from one setting to another. It also happens to be the case that the costs associated with bringing certain students to certain outcome goals may differ from one setting to the next. And that those things may actually interact that a high poverty student population in a small remote rural setting may have different costs associated with achieving the common set of outcome goals than a similarly high poverty concentration of students in a larger city or town. So we try to look at how all these factors interact so that we can figure out how much more or less would be needed here versus here for this child versus that child to give them equal opportunity to get to whatever the outcome goal may be. So we can take that. So we kind of broke these out into boxes to say that there are specific individual student factors. We talk about these as risk factors leading into the empirical analysis because these are the factors that if a student comes to school, for example, with limited English speaking skills, there is a greater likelihood that that student, at least for that time, will do less well on the measured outcomes that we're looking at. It is a risk of doing less well on the measured outcome. Students with disabilities at varying degrees of severity are likely to not perform as well on the measured outcomes. And in that sense, it's a risk factor, not to cast it as negative, but in the sense of it poses a greater risk of a lower outcome on the measure of interest and therefore requires that we look at some approach to mitigating that risk. And the approach we're looking at here is what levels of funding are needed to provide the resources, programs, and services that might help to bring those kids to the same outcomes. Some of them are specific educational needs like disability status or English language learners where it's, we know which kids have which needs and what types of services they would need to mitigate the risk. And most often that translates to additional staff with additional qualifications, smaller ratios that lead to greater costs. There are also broader factors, which we have kind of social context factors. When we talk about the concentration of poverty in schools, it's not that any one child who qualifies for a free or reduced lunch under the National School Lunch Program has a specific identifiable educational need as with ELL students or special education students, but that a school that's serving 60, 60, 70% children who are from low-income families has generally different needs. They need to provide smaller classes, additional support services than a school that has five, 10, or 15% low-income children. If we're striving to achieve that, there's a class-side broad general strategies to mitigate the risk of this higher poverty social context. And we know that we've had dramatic increases in child poverty and certain parts of the state over the last several decades. Some of the major factors that are school and district structural factors, should have gotten beautiful. No, I'm sorry, I'm very concerned about that. Well, it kind of is this, one of the biggest drivers of differences in costs from the research of times and economies of scale. Very small schools and districts operate at much higher costs of achieving the same outcome goals. And there's a substantial body of research on that. It also interacts with population sparsity, which is to an extent a primary determinant of things like transportation costs as part of the operating budget. So district and school enrollment size and population sparsity are significant factors that may also interact in ways that we don't understand as well with kind of measures of morality, which are different, I'd say, for study the Vermont context as opposed to looking at West Texas. We also have a number of other geographic factors, like geographic variation in competitiveness for employee wages. What does it take to retain a teacher with the same qualifications in one location versus another in Chimney County versus the Northeast Kingdom versus Rutley County? Are there differences in what it would take in a teacher wage to get a capably qualified fifth or sixth grade math teacher in each of those locations? And we can use other kind of external labor market data to try to get an age on that. But that serves as the frame than both for the empirical analysis that we conduct and for the approaches that we take in school finance systems to adjust for those. In our report, we're trying to set up an empirical analysis that can directly inform how you would adjust in the funding formula. But these types of adjustments for these different factors for economy scale, for differences in competitive wages, for low income children, ELLs, and children with disabilities have existed in financial formulas across the country for a number of years because we've known that they do affect costs. But in most states, the adjustments to account for these factors aren't actually based on any kind of empirical analysis of the costs associated with achieving gotten out of gold. So I think this is where I turn it back to Tammy to talk about what's going on in state school funding policies. So nationally, one of the things we did is we did a national scan and said, well, what do other states do? And rather than saying, well, this is how the formula worked, you know, one of the questions that the general secretary posed to us was, what kinds of cost factors do other state formula, funding formula, take into account? And that's relevant because actually, as we'll talk about in a minute, our state funding formula doesn't take into account many cost factors, right? We actually really narrow set of cost factors. And so in summary, what we found was, is that if you break out cost factors again by the typology that Dr. Breaker suggested, in terms of student need, we find that all states include some sort of adjustment for the differences in costs associated with students with disabilities. 47 states adjust for differences in costs associated with various levels of economic disadvantage for at risk students. 48 states adjust for differences in costs associated with educating English language learners to common outcomes. 35 states adjust for differences in costs for gifted intelligence students. And 32 states include some sort of grade level adjustment, and that is they acknowledge that the costs of providing education at different grade levels may be different. On your scale in sparsity, and I know this is of particular interest to this group, is 33 states in the country right now recognize that small districts and schools, and those located in sparsely populated areas face higher approved people costs. 11 states identify districts and schools solely based on size. One state identifies districts solely based on population density. And 21 states condition aid for small schools on being geographically necessary. That is they're both small and geographically isolated. So it's a two part test. And then third, 11 states incorporate some sort of regional cost adjustment. And frankly, these are larger states, and states that have very different teacher labor markets, for example, Maryland, and what they're doing is primarily adjusting for differences in teacher wages. States rely on them, and thinking about adjusting for cost differences, all states use different mechanisms of stuff. We're gonna talk a lot about waiting today, but I think it's important to understand that there's a toolkit that policy measures can use. It's not just about people waiting. There are different ways you can do this. We can think about single student waits, which is essentially what we have right now in Vermont, or some sort of per capita stipend amount, which is a dollar amount per student with a particular characteristics. That's one policy option. Another policy option is some sort of radiation of weight within a category that says, based on, for example, certain concentrations of poverty in a district, what if there was a different weight depending on the concentration of poverty? We have resource based allocations, which allocate funding, additional funding based on staffing profiles, and assumptions, thank you so much. You're welcome. Assumptions about staffing? I'm worried I'm going to dump that right on the computer. We've all had that moment, right? Rather than allocating on weights or some sort of dollars, they say, well, we're gonna allocate additional resources based on some ideal staffing profile, but we think that might cost. We've cost reimbursement, which is what historic we've used in Vermont for special education. And then we have catapult grant programs, which are dedicated targeted specific dollars for specific purposes, right? That operate oftentimes outside of funding formula. So we offer this up as sort of, again, a framework for thinking about larger questions around school finance policy before we launch into the weighting study specifically in our findings, because the goal is to help policymakers think about what are the tools in the toolkit if you have any. We want to talk a little bit about Vermont's policy. Now, there's lots of school, there's Vermont's school funding policy is complicated and there are lots of moving pieces. There are a couple of things that we wanted to highlight before in this conversation. And specifically just reminding ourselves what the program decision asked us to do. And that was to ensure substantial equal levels of tax effort, equal levels of school spending. And also that the wealth of the state should pay for educational spending, not individual towns. So those are kinds of garbills, right? The program decision provides. With respect to Vermont, Vermont's policy though for how we make adjustments for cost differences, particularly in the context of our weighting system, operates fundamentally differently than most other states. Most states use something called a foundation formula where there's a base amount that the state sets and then they make assumptions about the additional cost level above that base amount for differences in student need within a particular school or district. That's not how we work, right? And so when we talk, when you look at other states and you think about the weights used in other state formulas, those weights operate very differently than the weights operate here in this state. And we really want to stress that at the outset because it makes some of these state-by-state comparisons really tricky. And one of the things that undoubtedly will happen in a discussion around this report is people will try to make these comparisons. And so we want to make sure that we caution everybody that we have apples and oranges comparison here when we start to think about those states. And what the weights to you know, Vermont's formula, right? Well, we'll be back up. Vermont's formula is fundamentally different in that it is the local school, right? The local towns that establishes spending. And the assumption is that local schools and local towns and districts are going to make appropriate adjustments in their budgets for differences in educational costs, right? So think about that, right? The assumption in the formula is through local control that a district is going to look at their population. Think about the needs of those populations, right? Think about what resources need to be in place in that district that can ensure all students reach common outcomes. And then they set a budget. And then that budget is funded right through our statewide property tax. And the perks that they pay for that, those additional costs that they might have, right? Are discounted through the weighting, right? To our equalized pupil calculation. And then it packs tax capacity. It does not generate additional revenue, state revenue. And that's the other thing I think that's really important for all of us to keep in the back of our heads. And that's different than other states. The weights in our formula do not create additional revenue per se. What they're doing is they're creating additional tax capacity at the local level to pay for differences in costs. And those differences in costs are presumed to be reflected in local school district budgets. So when a school principal pulls the budget together, the assumption is that school principal has thought carefully about what all the package of resources need to be to make sure that all kids reach its own outcome. And that the cost of those resources are in the budget. And then we equalize tax effort, right? Which we scarred to paying for that across all districts in the state to the weights. That is really a different, very different from what other states do. So we need to keep that in the back of our mind. But the second thing we have to recognize is that that's our formula. But we need to step back and think about what is Vermont's school funding policy? And our policy in and of itself is comprised of many different parts. And the two parts we want to highlight is not just the formula and the weights, but also the categorical programs. Because we have categorical programs that provide specific targeted aid to school districts that are intended to offset cost differences. So some of our cost adjustments, right? And we talk about what we do in the framework. Some of our cost adjustments are through the weights that are in the formula. Some of them come in to play in the context of these categorical grants. And together, this is our funding policy, right? We have to think of all the pieces. That one is not just the weights, but also the categorical grant programs. And so in our formula, in our funding policy, our categorical grants provide supplemental funding for specific programs or services for certain kinds of students, for certain kinds of activities. Those dollars, right? Come off the top, if they were a district and spending number is generated, right? Then we also have the weights that a district, which then we also weight a district's average daily membership for cost factors. And use these district weighted, use district weighted membership to equalize for people spending for the purposes of political tax rates. That's our funding policy with regard to cost differences. It's not just the weights. The primary categorical grant programs, the non-operative categorical grant, the special education grant, the transportation grant and small school grant. Special education finance program and ministers to state special education funding laws. It's changing grant under Act 273. We're transitioning how we allocate dollars. But that is a categorical grant that's outside of the other general education funding formula. Transportation aid is available to reversal for half of the school district's expenditures to transport students to and from school. The exact reimbursement percentage, right, varies year to year a little bit. But, and the total amount is limited by appropriation. But that's also part of our package because that's intended to offset the cost of operating world schools or schools that have higher transportation costs. And then we also have the small school grant that awards categorical grant to small schools where the two-year average enrollment is less than 100 students. So these programs provide explicit and additional state aid to offset direct expenditures. Then we have our weighting, which, right, where the weights are used to calculate the number of equalized people. And we can think about the equalized people as the average student in terms of educational costs at a school district. Okay, that's, I think that's a helpful way that you might, that what we're doing is just trying to create average costs. And specifically the weights implicitly, whereas the categorical grant explicitly adjusts the costs, the weights implicitly adjusts the costs. And we're spending differences by equalizing for people spending across districts and the impact local tax code. And again, they don't generate additional revenue, they impact local tax capacity to generate education-related revenue. And I think one way you can think about this, and I think this is useful as we start to talk about the change in potential changing weights, right, is assuming, let's assume for a minute, that we had the school budget that was devoted in your town or your district last year, right? And we had an equalized number of equalized peoples, right? That determined the tax rate. If that budget was unchanged, right, everything stays the same. And the number of equalized peoples in the district goes up, the cost per pupil for the district goes down, and that lowers the education-related tax rates. If, again, that budget stays the same and the number of equalized peoples, right, through weighting goes down, the cost per pupil in that district goes up, and your education-related tax rates go up. That's how the weighting mechanism works. And again, I want to be clear, that's holding spending constant, right? This gets far more complicated when we have movements, not just in spending in a particular district, everybody's different, because we have a statewide system. And so one of the things we'll talk about when we get to our simulation, it's one of the reasons we need to be careful with those simulation estimates. They're helpful, but they're not arbitrary in that that's exactly right. We have to make a lot of assumptions around those. We, in our system right now, have four existing weights. That's a four. One is for economically disadvantaged students. The value of which is 0.25 for 1.25. English-language learners of 1.2 out. Secondary students, and secondary students are defined as grades seven through 12. This is in high school, it's seven through 12, and that's 1.13. We have a pre-kindergarten weight of 0.46. A couple things that are important to note here. First, these weights are historic artifacts. That's the best nicest terminology I can use for them, right? A pre-date at 60, in our investigation, we could find no evidence that they were empirically dropped. The 1.25 comes from some federal report around 1989 and 90 where some of it did average expenditures on kids in poverty. A lot of states adopted weights around the time of the international, and it was a national thing, and you know exactly where it came from, but it became embedded in policies in the number of states and here it is. And here it is. So when I say historical artifact, I mean 80s historical, flashback to the 80s, okay? So it's a historical artifact. None of these weights are empirically derived. Why am I stressing that? Because in order for weights, in order for this formula to work well, for the formula to do its job to adjust for cost differences, the assumption is that the weights are appropriately adjusting for cost differences. If those weights are not doing their job, then we are not appropriately adjusting for cost differences and spending differences across school districts in the state. And that impacts capacity, which then impacts the local decisions about what they spend, and then that just continues on, right? So there's no case where these weights are not empirically derived. They're historical artifacts. And frankly, they raise serious questions about the extent to which they are affected with respect to adjusting for cost differences across Vermont school districts. One of the key things we did in our study is we went out and we talked to stakeholders in the field about their perceptions, their experiences with the school funding formula. We thought that's really important because certainly empirically deriving weights is important, but also understanding what implementation looks like is also important. And so we talked to 35 individuals around the state, some of whom were in this room, both here and back here. The educational leaders, members of the general assembly, members of representative educational organizations and organization leadership and fiscal staff at AOE. And there was remarkable agreement across this group of individuals around six items. One, the cost factors. And those are those things that we adjust on, not the weights themselves, but just what we adjust on, do not reflect current educational circumstances. That they don't, that simply adjusting for poverty, ELL, and grade level, which is, is it really capturing very well what it counts for the real foods on the ground differences in educational costs around just across the district. Second, the values of the existing weights, ELL, poverty, or grade level weights, have weak ties with the actual differences in cost of educating students, which is not particularly surprising since they weren't empirically derived in the first place. Three, the state's small school grant program is problematic in its design and current operation. And I think we can safely say there was uniform frustration among every single stakeholder, among the stakeholders with the existing program. Four, there is a need for specific and targeted grant aid, categorical grant aid, to support schools struggling to meet different and increased levels of student need due to childhood trauma and mental health concerns. That bubbled up too. We also asked stakeholders about the special education block grant calculation. As you all know, right, there's been a lot of conversation around is the existing calculation fair? What are the potential implications of that calculation? And so we talked to stakeholders and said, what do you think about that? You know, stakeholder reactions were mixed. We need for potential adjustments at this time. And I'm going to be real clear about it at this time. At one of the continuum, we had one person say, the sky's not going to fall. At the other continuum, we had someone say, the correlation between poverty and disability is really strong, and we have to think about that. In some way, in the middle, we have people saying, again, these quotes sort of typify, it's too soon to tell whether the grant will be a problem. And to that last point, what we heard is, even among stakeholders who are concerned, whether or not the grant's going to be sufficient or equitable, there also was this recognition that the census grant and the concerns around the census grant were actually very closely tied to concerns around how student poverty is weighted in the general education formula. In that, in districts, in places where there were particular concerns around the census block grant, some of that concern actually was with respect to that they felt as if they were unable to spend enough on the general education side for students in poverty, and that those costs had been spilling over into the special education fund. And so there was a sense that if there were adjustments to, for example, the poverty rate, that that might stem some of the tide with regard to concerns on the census block grant. I think that's an important thing to consider as we go forward. And again, that's not to say there weren't concerns, but the concerns were mixed. And there was a general sense in, there was a general sense among all stakeholders is that concerns were most strongly related to the uncertainty with how the grant is going to be implemented. There's a lot of discussion. There's a lot of other problems though, right? The working group is still working on some of the implementation. And so there was a lot of questions like, well, we think it might be this, and if it was that, then I would be concerned, but I don't know if it's going to be that. And so that, I'm not sure if I'm concerned, right? So just to give you a fair appraisal of what the stakeholders said. I just said I've raised my members that doesn't have a question. Okay. Just a quick one, who actually leads and talking about the fact that it's a kind of historical. Yep. This slide's consistent with? Yes, it's consistent with. Just curious about the pre-K way to the 0.46 since that was more recently derived. I'm just wondering if that was computerized as well. That was not part of our study, right? It's just pre-K, it was not so we didn't evaluate that way to, in our... You wouldn't categorize it as the other ones as being quite so historical. Whether or not it's empirically derived, we didn't look into it. We did not look into that. But it's not... And it's much, right, and that way has been enacted much more recently. So the economic disadvantage, the wait for economically disadvantaged students and ELL students are our whole donors from pre-F-60. They haven't been changed since before F-60. Secondary students that wait was adjusted in 2007. It's actually decreased about half to 1.13. The Agency of Education evaluated that and by looking at sort of the ratio of elementary school spending versus secondary spending, they thought that 1.18 is what they saw there. So that's as close as any kind of empirical derivation that we have for that way. The point for a sixth. No, the secondary students. Oh, I'm talking to the... But the point for a sixth, that was not something we focused on in our discussion. Okay. Is that all? Yeah, that's very fine. Okay, thank you. Quickly, the stakeholder perspectives on small school grants, as I said, stakeholders are uniformly opposed to the continuum of small school grant program as it currently stands. One stakeholder, and I thought that would be particularly powerful. Everyone is looking for a better way forward. I see some nodding. Nearly all interview participants view the small schools grant program as fundamentally at odds with policy goals in particular in that 46. There was general agreement, however, that the state needs to support geographically necessary small schools, right? That while the existing grant program is at odds with that 46, there was one agreement among stakeholders that we have geographically necessary small schools and that we have an obligation as the state to support them. Just might not be through the small schools grant program as it currently is. And in general, stakeholders felt that incorporating weights for school size and morality in the equalized people calculation were the immediate concerns related to eliminating a small schools grant program. Is that in there? Sure. Oh, yep. We're shy. Thank you. Is there somewhere defined geographically necessary? So that's a good question. So I think that certainly is a normative concept at what threshold, and so one of the things we'll talk about in our empirical analysis is we started to look at that in terms of population density and we looked at what thresholds at which threshold for population density at which costs start to change. And so we use those frameworks. So there's a thing about that. Talk to me. At the point where population density no longer seems to have an effect on cost at the same time as size, that's where we saw the kind of phasing out. Now, the trick is, if you're doing this as empirical analysis, it's all on these nice kind of continuous curves. When we start trying to derive this in policy, we wanna draw right lines through it and you gotta figure out, but where are you gonna draw that line? Who's gonna fall just above and just below that line? So we have some guidance from the report. That's not to suggest that the cuts that we made in the data are what should absolutely be the policy, but that they were points where we saw that enrollment by sparsely related costs seemed to taper off to not necessarily being different than a non-sparsely, non-small school. And we have that in the economy as a scale or a term more broadly, which I think we know that school districts, at the district level, districts achieve kind of scale economies at anywhere from 1200 to 2000 students. The costs don't sharply go up until you get to districts with 300 or fewer students and even more sharply with 100 fewer students. So in those ranges where you're between 1200 and 2000 or even 1200 to 1500, it's kind of going up on a gradual curve at what point do you decide we need to step in with policy to account for this. Otherwise, it's simply unfair. And we re-interacted that with population sparsity so that we could couple those decisions. Well, this is a cluster of small schools that are in a non-sparse area and we want to be subsidizing that at a higher rate. It's a great question. I think it sort of foreshadows some of the discussion and debate that policy makers are going to have around the findings. And one of the things we try to do in the studies is we do some simulations, but we made some assumptions there and those are supposed to be illustrative but not sort of hard and fast. And we're looking forward to hearing the discussion but I want to just, let's hold questions. If you could just make a note on the page that you want to ask questions, we'll make sure that we'll get to you. Two other considerations identified stakeholders and bubble up that we wanted to know is there were frequently voiced concerns around the impact of Vermont's really college programs by industry's long-term membership. And the general consensus being that the way the ECP works right now is that when a student enrolls in early college, they come out of the count entirely. But what we heard from school districts is that school districts continue to serve these students. They continue to receive guidance services, they continue to participate in extracurriculars. They're not zero cost the district anymore. And that there were some concerns around that and there were suggestions that they're making opportunity to think of those students as a fraction of a full-time equivalent student. The second thing is a bigger issue and it really strikes sort of at the heart of how our funding system is set up. But we wanted to make sure that we highlighted because it came up quite a few times is that there's this underlying concern that any efforts to update the equalized people calculation. So we do all the hard policy, the political work updating weights. The goal of introducing more equity in the system may not actually translate to increased levels of spending. So the idea was we're trying to do this work with the weights so that the districts with higher needs students are now capable of sufficient resources. But remember when I started out, districts still get to decide what they spend, right? And so it could be that a district's equalized pupil count goes up over those arrows and their tax rate goes down. And instead of using that tax capacity to spend more on students to write equalized outcomes or close spending gaps, the local voters decided to take that as a tax cut and continue to spend it at the same level. We heard that a lot. It's a thorny issue. I'll just say that. But we felt that since we heard it so many times that it was important to share that with policymakers that that is a concern. That around this report and the reports fundings. So we're going to turn it back to Bruce to talk about the weights. We really took, the approach that we took in this project really kind of hits a convert end. On one hand, we went out and talked to people to figure out how do they feel? What do they think are the problems with the formula? And then we flew back up to at least 10,000 feet to gather a whole bunch of data and run a whole bunch of messy complicated models to try to figure out what should a weighting structure look like to provide equal opportunity to achieve the outcome measures that we have. And we, but we still had to kind of get back and fit that to the concept behind the Vermont formula, which is to use this to generate some kind of a weighted equalized people count, which would then be used to in effect, provide for the increased capacity to raise the tax dollars to get to that money that would be needed to serve the needier population where you to at least stay at the same tax rate or even increase toward it, the competitive tax rate with those around you, which is, I mean, there are a whole lot of kind of economic theory about what the behavior of town would be in the shifting of the tax rates under these circumstances. We didn't get there in the report. So the approach that we took on the next slide is one of the really important things that we often forget to do in this type of cost modeling is just to actually assess, well, what are all the different measures that we have in a state that capture differences in child economic disadvantage? What are the different measures we might have on students who are English language learners? Do we have any greater level of kind of precision in the data on the degrees of severity and prevalence of disabilities? Now, what are the different types of measures that we might have? We had a number of different types of measures to capture the shares of low income children across schools in the state that might yield different weights, depending on how we estimate the model, and we wanted to start with this risk analysis to figure out, well, is it the percentage of children who qualify for subsidized lunch, free plus-reduced lunch would be the children who are from families that are below the 185% income threshold for poverty, as opposed to the district kind of child poverty and the measure that we have in Vermont, which is at a somewhat lower threshold, but not at the census poverty threshold. And then even using different sources of data can give you different kind of degrees of precision. We use data from the Center for Education Statistics over a multi-year period, and then use those data to kind of predict in the empty spaces from their data set for free-reduced lunch at the school level, as well as using the state's own ALEE collected free-reduced lunch level. It was kind of a weird thing for me to find in our risk analysis that the stronger correlation between the low income measures and the various kind of test scores, student outcome measures, was with the National Center for Education Statistics measure and student outcomes, as opposed to the state measure. In most states, we find much greater kind of precision on the state-collected measures, but I think it was the multiple years of data in the way that we kind of predicted values to fill in the blanks, and the fact that we might have some kind of ceiling effect with community eligibility reporting. So one of the really important things at that stage of the analysis, though, is to find the measures to pick up the variation across districts, right? Find that sometimes in a state that is generally very high in poverty concentration, using a measure that's at a lower income threshold can pick up the variation across districts better. More than half the kids in the state of Texas, just because of the income distribution of taxes, fall below that free-reduced price line. So when you use free-reduced lunch across schools in Texas, you have a whole lot of schools that are simply 80 to 100% free-reduced lunch, yet you know that there's real variation in the economic circumstances of kids across those schools and you want to pick that up and weight system. And when you use a lower income threshold, when you use a more stringent measure of poverty, that's also likely, when we move to the next step of looking at cost, lead to a bigger weight on more severe poverty, right? So all of these steps have to be kind of taken together. We want to do this risk analysis to come up with measures that predict differences in outcomes, figure out what measure is working best. And we did this ultimately with school-level data that were largely Vermont-sourced data, district-level data that were Vermont-sourced data. And we went into this with a concern, the second stage of this is to estimate these kind of excessively complicated cost models. And to get a cost model to kind of meet all of the statistical requirements it has to be, sometimes you have to make sure, every time, you have to make sure you have enough variation in the data, that there's variation across the settings in schools and districts and in the outcomes that they achieve. So we built into the analysis, doing a separate model based on a new study that the first version of which I produced about a year ago, the new version is coming out in February, where I had estimated a national education cost model with this new data set out of Stanford, the Stanford Education Data Archive, where they kind of equated all of the district outcome measures for every district in the country that took state assessment data, did some statistical tricks to make it comparable, norming it against made scores. There are some issues with those data, for sure, but it's better than having nothing. And if I may interject, I wanna really stress something that Bruce is saying here. Right, so what Bruce did, along with AIR and the team, is instead of just looking at Vermont, right, in Vermont's quirky, we hear about Vermont, you think this is Vermont, but when we approached this study, we wanted to make sure that the cost factors and the measures certainly worked in Vermont, but that we can validate them with national data, and one of the strengths of this study is that we were able to triangulate all of these findings, not just in Vermont, and verify them with regional and national data. So what you're seeing in our findings are not some idiosyncratic Vermont phenomena, right? We end up using the Vermont models and Vermont findings, but we triangulated those findings to verify them with regional and national data. And we're able to do that because Bruce has these unique datasets, and it allows us to do that. And I think that's really important to know, and I wanna just point that out. What we do is we set up, and I didn't wanna do it off the national model, because there's just a whole lot of variation nationally that might not be as directly accountable to Vermont. So I set up a model that used data from Western, Massachusetts, except for the Boston Metro, so Western Mass, New Hampshire, and Maine. I had upstate New York in it, but there were problems with the New York State data. There are a lot of interesting kind of features to the kind of rural population decline and other things about upstate New York, Western Mass, and Northern New Hampshire that are much more similar to Vermont than elsewhere. What I was most concerned about in fitting the Vermont-only models with things like picking up a stable, kind of viable way for English language learners, because we have so few pockets of concentrated English language learners in Vermont by adding in Western Mass, picking up places like Holyoke and Springfield, and other towns, kind of through the mill towns in New Hampshire that have seen dramatic influx of English language learners. I was able to capture a little more variation to, I think, get a kind of a more stable and valid estimate for English language learners that we go on and use later in the recommendations. So it was nice to be able to do a school-level model with Vermont data. Actually, one of the things you worry about in trying to estimate at this second stage is statistical model where we're relating existing spending data, outcome data, with consideration for all these different cost factors based on the measures we selected in the first step. So we get this statistical model that then tells us how much more for each unit change in low-income kids is the expenditure associated with achieving this outcome goal? How much more, with respect to ELL kids, is the expenditure associated with achieving this outcome goal? And we actually are also throwing corrections for what might be differences in the efficiency of that expenditure. And then even how do those things interact in a model where we're also considering how much more does it cost to be in a school with 100 or with 50 units in an area that has 300 residents per square mile versus 100 versus 50 versus 10? So we build that all into this model which becomes rather complicated, all that the tables of that are in the report where you can see what are the differences, differences in costs associated with moving from zero to 100% low-income, from zero to 100% ELL or zero to 100% on any other, as well as moving across different size categories of school. Because we know that model itself becomes complex just to meet the statistical requirement. There's certain kind of tests a model has to pass to meet the requirements of being a good and valid statistical model. We then want to be able to kind of boil that down into something that's reasonable for policy. You could conceivably take the predicted cost values for the model and back out from that just a general equalized pupil count, that this is how much more it costs in Rutland than in Rutland town or in Manuski than in Burlington to achieve common outcome goals. You could take it with a global measure and boil it down to an equalized pupil count and back out the tax rates from there. But we know that there's a whole lot of kind of black box aspect to that. So we then took it the next step to fitting kind of simplified models in this weight estimation where we take, but what are the factors that we would actually just be using in the formula? What's the low-income pupil count we'd use in the formula, the ELL count we'd use in the formula and some district size groupings, as far as the groupings, and use what we would have in a formula to predict the cost predictions from the cost model. And we can predict those with over 90% accuracy with a simplified model. So we thought, okay, and with something then that's actually usable and interpretable as policy. So that's this next step. We estimate this complicated model if you come up with cost predictions for each school or each district in the state or even in the region. And then we move to just looking at the specific measures that might be used in weighting the formula and predicting the cost predictions from the cost model to come up with, well, how much more in dollar terms and then in weight terms does it cost to get common outcomes across the range of poverty for math schools and districts across the range of ELL students in the math schools and districts. And then the next step is to kind of work that all backwards into a tax rate calculating simulation which was done largely by Drew Atchison with American Institutes for Research. So, yeah, I think I've checked it. We'll just skip ahead. I think he covered all of that. So let's move on to slide 23. So out of all that, what did we find? Well, we found the things that, you know, we'd be known to affect cost, but we were able to generate some useful insights as to how much, how we could construct a weight for economically disadvantaged students based on the measures currently used that can get district level in Vermont, even though our best cost model had used school level measures, we were able to create a bridge between the two and use an economically disadvantaged measure based on the existing district measure. We had more variation in what we saw in our cost models on the estimates related to the cost associated with English language learners, but I think the greatest confidence in those areas where we had the greatest variation, which was in the regional model, almost everything else that we use in our policy guidance is from the Vermont school level model, but we also were able to pick up some of the differences. We are not, when we look at these kind of differences in middle school and secondary grades, the way we had looked at it in our models, knowing that there are various configurations of schools out there is to look by the proportion of students in certain grade ranges. That's something that Drew and Jesse and I have come to after doing enough years of this and trying to come up with variables that identify the K5 schools, the K6 school, the 3A school, the four step, and all the different configurations, it was easier to try to come up with a more uniform measure, and one that can still be used directly in policy. What proportion of kids are in this grade range? What proportion in this grade range? But these differences that we see by grade range are not still in our mind necessarily cost factors. They are, they pop up in these models as differences in expenditure based on the grade distribution of students. But to a large extent, we don't know, for example, if we were to have invested more in the early grade, we'd be getting better outcomes in the upper grades. There may be some differences, and we tend to see those grade range adjustments more often than not reflect historical practice, which is a little different than when, because they're not as tightly linked with the outcome measures directly as poverty or ELL status. So the grade range thing is a little fuzzier, but the geographically, the population density and school size factors were highly consistent with other analyses in the economies of scale and populations of course. These are the, when we boiled it down to weights, through that second stage of the analysis to then be used in our simulations, we came down to a set of weights that were significantly larger for child poverty. But we had already known that. Even back, a lot of states, one of the most common ways that states go about adopting their poverty weights is what our other states do. This state just did that, this state just, that's how we came up with these 20, 25% weights. By the mid to late 90s, there started to be emerging evidence that, well, it was anywhere between 50 and 150% based on different analyses that were being done in academic work. So a number of states started moving to like 40% weights. You don't quite have enough money to get to the 50, and we're certainly not going to the 100s or we're going to the 40. Few states have jumped to actually directly using kind of cost modeling analysis. Kansas has used cost modeling analysis to directly inform its policy. And Texas had used cost modeling analysis, but it didn't result in substantial policy change. They were going about that around the year 2000. The Kansas, when they did a study in 2006, which informed policy reforms, which were then undermined by the recession. And they just went back and did another study in 2018 using cost modeling, which is likely to guide the changes down the line. And that second one comes out with weightings that are more in line with even what we've come out with here. So we take two approaches here. Tanya's going to address a little bit more because we know that special education is ultimately treated separately in policy. But our first set of models, we actually run two sets of models, one that includes high instruments, low severity, and low incidence, high severity special ed students in the cause model. And when you run that model, because there's a relationship between disability concentration and poverty, you get significant weight on disability concentration, which then eats up some of the weight on poverty. And when we pull that out, if you wanna have a kind of a consistent overall school finance policy, you'd wanna make sure that you're capturing the full collective effect. So if you pull out the special ed and don't separately weight it for the poverty concentration or the other weightings that might interact with it, then those have to go back in the other or vice versa. So knowing that these were separate policy considerations that they were likely to remain that way, we then ran the model with and without special ed students in the cost model and in the weighting model so that we could move those kind of conflated adjustments around moving into special ed, taking it out of the general, or move it back into the general, taking out of the special ed. And I think I'll tip it over to Damia. It creates a policy choice, right? And so just to recap, Bruce just said, so when you look at these cost factors, these cost factors were impurely derived, right? We didn't say, well, these are the cost factors. We think there are, these were impurely derived from these models to say these are the things that account for differences in cost across districts. But we also know that one of the things that cost would account for differences across, difference in cost across districts is special education students, right? But we have a categorical program for special education students. And so if we include them in the weights, we double count, right? So we have to think about that. So there are two, there's a policy decision, right? And the policy decision is with respect to any potential adjustment in differences in cost associated with students with disabilities. We could do that through the census block grant, right? Adjust the grant amount. Or we could adopt a different set of weights that implicitly adjust for differences in the incidence of special education students, right? Across districts, right? And they're different lines of length, I'm gonna say. The models are, the models help us do that. But that's really the fork in the road, right? Is that you have a choice. Pulse makers have a choice. If you're interested in adjusting it, right? And addressing potential differences, cost differences. We could either do this implicitly through the equalized people calculation by choosing a set of weights that incorporates some of that variation in the weights you select. Or you could choose not to adjust at all. Or adjust specifically over on the census grant, right? And so the, what was that? Classification. Yeah, I'm just gonna go there. So. Because it's a little confusing. Yeah, so, so, so. Column number two, here's the cheat sheet. Column number two is the column of weights that if you want to implicitly adjust with the equalized people calculation for differences in student aid associated with the incidence of students with disabilities, okay? But if you do that, you should not make an adjustment in the census block grant for that because you've double counted, okay? If, right? If the policy decision is to either do nothing, right? Or weight, right? On making any adjustments to the census block grant right now, or to adjust the census block grant. It's column number three, okay? Because that model controls your students with disabilities and that's high regression numbers, okay? Go ahead, Peter. Just a very quick clarification. Like let's just say line one, the existing weight is 0.25. It's really, it's one plus 0.25. So it is one plus 0.25. And if you do that, then you have to add one to those two other numbers, right? These numbers are, yes. Yes, yeah. They're like that to be comparable, okay? So and the reason the 0.25 is listed is if you look in statute, it says 0.25, right? But it's multiplied by something. And so what we wanted to do to eliminate the potential for confusion around what would actually go in statute, the number, if you were to adopt column three, right? Those are the numbers that would go in statute that are comparable with the numbers that you have in statute now with the calculation. And if you look at the report, there's a detailed table that walks through step by step by step to equalize people calculation that we have now and exactly how these weights could be incorporated in their values, okay? So that's why that's like that. Does that make sense? Yeah, so if you were to adopt two of line one column three, you would count a student who fell into that category really as 3.97, not as 2.97. That's correct, but the number that would go in statute is 2.97, that would be the multiplier, okay? So a couple of other things I wanted to, so have a look at the right, right? For deciding column two or column three with regard to weights. A couple of things to point out here is as Bruce mentioned, we ran models using district data, school data, regional data. All the models triangulated, right? But they're all little different. In our evaluation, each of them created a different set of weights, right? In our professional judgment, the weights that are derived from the school level models which is what you see here are the recommended weights. And when I say recommended, I mean recommended across the different model configurations, okay? I'm being very careful. We're not making a recommended policy recognition, but if you look at the different models, these are the weights that we feel are the strongest from a new pair of weights. The other thing you have to keep in mind with these weights is that you have to think about the weights in terms of columns. What do I mean by that? That means because the weights are derived from a regression model where you have all these other things in them, the minute you start, all the weights are conditioned on each other. So you can't just say, oh, I like 2.97 and 0.2. Oh, you can't do that. You have to think of these as packages. And that makes sense, right? Because what you're trying to do is create comprehensive or cohesive policy, right? And the minute you start picking and choosing specific weights, you unwind that because the models weren't developed that way. And nor should policy, right? We need to think about this as a system and we have to think about this comprehensively. So when you start to think about this table or other tables that might be generated, you have to think about columns. And those columns are not weight values, but also the rows. And the rows are the things that you include as cost factors. And so all of those weights are conditioned on the fact that those cost factors are also included in the model. Does that make sense? Okay, so what we can do, why don't we just flag this as one that we might need to come back to? Yeah, that's fine. I just want to make sure people understand. There will, as they said, there will be lots of questions that come up, but those are some big guideposts. As you start maybe to go back and see, read things, those are some important guideposts to help work through that. What slide number is that we don't have in the room? Okay. That is slide number 24. It's also the same table that's in the executive summer, okay? About a fork in the road, right? So you can think about, policymakers are inclined to adjust for differences in special education. You can do it implicitly through the weights, or you can write weight and see, or if you decide to do something now, you also can adjust the census grant calculation. There are two ways you can adjust that calculation. If you think about the calculation, it's some fixed dollar amount times the number of students in a district. So you've got two policy levels you can pull here. I can adjust the dollar amount per student. I can adjust the student count. One of the things that we talked about in the report that predated some preceded Act 173 is we actually talked about maybe a way of adjusting the dollar amount. There's a lot of consternation around that. And I'll refer to Representative Beck because I think in the committee meeting when we talked about that, you made a really important point, which was you said clips, right? That just gets really complicated. You lose the predictability and transparency in the formula because every year we're trying to figure out what the dollar amount is. And one of the strengths even among stakeholders who may not be strong supporters of the grant program is that the predictability and transparency is actually really valued. And so our recommendation is if you decide to adjust the grant amount, that you don't tinker with the dollars, you think about how can we adjust the pupil count. And there are two options that we propose. Again, these were based on stakeholder input as well as empirical analysis. Option one is to take that flat grant amount, also known as the unified base amount in statute, and multiply it by the number of equalized pupils in a district, right? It's pretty straightforward. It's predictable for school districts. It ties, right? It presumes that the weights used in the equalized pupil calculation are capturing the variation in student aid, right? Option two, excuse me, is to multiply the unified base amount by the poverty weighted pupil count. Yeah. This one derives from the fact that we have a figure in the report on our number, where we show, and I think you've shown this on the report, that there is in particular a significant relationship between disability concentration and concentration across Vermont districts. So that probably the one area of great overlap, that if you were to only adjust by the poverty weighted count and not the fully equalized pupil count, you're missing the possibility that speciality cost might also be influenced by scales, varsity, and overwhelming with ELL, for example. So there's certainly a stronger relationship with poverty, but taking that particular approach assumes that the special education costs don't vary with respect to the other factors. Varsity equals people count multiplying by that would be a more generous adjustment than the poverty related adjustment. You'll notice that the unified base amount changes for the poverty, that's just because in order to stick with legislative intent, which is to stick within a specific appropriation because the poverty weighted counts now equalize back to the number of students in the state, the same grade, you have to do that, and by doing that, because it's not, you have more pupils. And so you just have a bigger denominator, and so it's not that that reduces the, it's not that that reduces your overall appropriation, it just, we have to write the size of that because we have a higher pupil count, that's all. So you have two options, right? Well, you have three options, four options, right? You have, you can do it explicitly, right? You can do nothing. If you do something, here are some ideas about how you can do that, and we simulate these options in the back of the report, so you can kind of get a flavor for what this might look like, okay? Any other question or question? When you're talking about using the weights, equalize pupil count on a poverty weighted pupil count, are you talking about the weights you just put in that other child? Great question. Or the ones we have now, whether it really matters. Both. How can you do both? So what we do, let me tell you, it's like against United Commission. Not at the same time. Not at the same time. You only have one we only have the ones. What we did in the simulations in the back, right? What we do in the simulations in the back, and I flip forward, is we offer simulations that do all those different weights. Yeah, at least it does have an understanding of it. You know what, I'm gonna pull back. I don't have a question, but we'll do this in committee. I'll do it, yeah. I know, but I can't understand the rest of what she's gonna say without asking how she's doing both at once. So I'll leave right now. Well, we're not doing both at once, right? No, what I said was, is that we calculate, we do the simulations in a number of different ways, right? What's in that chart? When it says equalize pupil count, is that what we have now? Yeah. Or is that option one on the other side? Option one is when we do the simulations, it's broken out into option 1.1, option 2, and option 1.3, right? So we use it the existing first year equalize pupil count for option 1.1, option 1.2 uses the weights that we proposed, option 1.3 uses the weights we proposed with the substitution region of the yellow line. Okay, okay. So, it is, we do this. But that doesn't show in the chart, no, no, no, we don't intentionally, I didn't do that here. These are distinct options, and then we calculate the different ways that we're focusing the simulation. Okay, so we've talked about the simulations a bunch already, right? We're not going to go into specific simulations in the back, I think that's for committee, and in the end, but what we do, didn't ever just do here is familiarize your, with how you can look at those simulations, right? And the simulations are intended to show how the cost factors and the weights derived from our analysis might be integrated in formal system school funding formula, you have two scenarios, right? Column two, column three that you just saw in terms of the weights. And this table summarizes that. We manipulate one thing, and Dr. Bates mentioned this, is given that we have a relatively small ELL population of the state, we do change out and use in some of the simulations through regional ELL, and that is the column, that's the row highlighted in yellow. You can see how we use that. But each one of these simulations shows you the weights that we used, and you can look at it, each corresponds to the appendix of the report, that show you exactly the impact of those changes on the equalized people count for the districts. We also simulate the tax rates, however, in looking at that particularly, the changes in tax rates are going to be really clear. That presumes fiscal year 18 spending, both within that district and statewide, the base slash yield that the state established for fiscal year 18, right? So it's a retrospective application to existing data. So the reason I say that is I don't want, I would be, I get concerned that people are gonna sort of get ourselves tied up in conversation about microchanges in tax rates, because these are really illustrative, and they are conditioned on these sort of assumptions about what happened in the year past. We all know that that's not going to be what happens this year, not in town meeting days. So again, it gives you a sense. I think when you look at those simulations, what is more powerful is to understand what does that give you to the equalized people count in my town, your district, and the percentage change. And it's also important to note that they are conditioned on the choices of specific measures that we've used in determining those equalized people counts. Right. That all of that is subject to some shift in the actual kind of policy choices in determining what is the exact version of this measure to be used, the timing of its collection, and so on, and all of that will lead to subtle changes. It presumes that these weights and these cost factors are used and applied retrospectively. Okay. And again, the simulations are still a powerful tool for decision-making and understanding, but what I don't want to caution people is if you go back to your constituents, when you're here from your constituents that this is the precise number and this is what, that's not exactly the correct interpretation. Right. That's to give you, it's really supposed to illustrate what it will look like under a very specific set of conditions. And we were having too much fun just giving one or two options. So we had to throw a few. We asked for whistles. By the way, the R.P.A. should buy in and we asked for six. Here, down at four. So hopefully that's okay. It makes it a little less confusing that way. Okay. We talked about the specialization of candid law grants. So how do you put this all together? Right. So we give you a lot of information, a lot of pieces, reports, tourniquet, a lot about making for a chart that sort of lays this out in terms of decision-making. It might be useful and also as a way to wrap up. Is there's an initial question, right? Do policy makers want to incorporate new cost factors and weights in equalized people calculations? If the answer's no, I think it's time to recess for a long issue, right? If the answer's yes, your next question is, well, what weights slash cost factors do we use? We've offered you, in our report, a set of recommended weights of out-of-our models, right? In that table that first talked you through. And you have a choice of column two or column three. And column two specifically adjusts for differences in special ed. The other one says we're gonna do that around the spread. So if you follow our report project, your next choice would be column two or column three. If you take the weights that implicitly adjust for differences in students with disabilities across districts, then there's no need to move on to the additional decision, which is do we adjust the census block grant? If you take column three, right, which has where you have implicitly, your next decision is, okay, do we adjust the grant? If it's yes, you have options one and two, right, which are conditioned on how you calculate equalized people, right? And your decision there is, well, do we use an equalized, do we change the count of students that we use in our census grant calculation of an equalized people or are poverty weighted versus the existing ADM, okay? So the idea is to sort of lay it out. Now there will be variations on this, but hopefully that gives you sort of a roadmap for taking these where are the key decisions and what is the sequence of those decisions. So, some key, just to recap, some key conclusions. From my approach to adjusting the differences in educational costs across school districts has remained relatively unchanged for 20 years. Stagnation in the state's education funding policies is a source of, has been, is a source of concern. Existing policies are viewed as widely used as outdated and falling short of equalizing educational costs across school districts and by extension opportunities to learn for students across the state. The manner in which the state currently calculates the number of equalized people in the school district has been criticized for being out of step with contemporary educational industries. And existing funding programs also fail to recognize significant shifts in states' educational policies and practices. For example, flexible pathways for early college pose new challenges. Findings from the study suggest that it's time to incorporate new cost factors as the things we weigh in on and weights for those things in Vermont's education funding formula. Findings suggest that the existing weights for economically disadvantaged ALL students fall short or appropriately adjusting the cost of educating these students to standards. That we need new cost factors for school size of population density and that these could replace the existing small schools grant program. And that refining the secondary school rate to include a middle level and a secondary level adjustment as opposed to resulting everything together of seven to 12 might better align weights with educational policy practice. Finally, that modified equalized people calculation may not translate into increased levels of spending in districts with higher need. The additional tax capacity generated by a higher of people as people felt may be seen as an opportunity to reduce taxes rather than close spending and opportunity gaps. And then there's a need for new categorical status of student mental health and trauma-based instruction. Thank you. I know we're going to be hearing from the secretary as well after correct you. I wasn't planning on it. Oh, okay. I think our presentation is much more comprehensive. Okay. So then what's our time limit for the departure? You're okay? Yeah, we're good. Okay. I think we will open it up for questions. I know, Representative Austin, you had a question? Yeah, just two questions. I'm wondering if there's any data that shows that addressing poverty would address the needs of children with reduced lunch as opposed to addressing them in school and addressing poverty? Certainly there is. In terms of data on housing policy, it's really interesting. There's so much of it goes on. So you're trying to translate the research on housing policy vouchers in Manhattan, so Vermont. There's a large body of research in broader kind of public policy management. It focuses on housing policy, transportation policy, food security, all of these other factors that could be addressed in a more comprehensive overall policy strategy that we often find ourselves in our narrow field trying to then compensate for through waiting for the school funding formula. So the weights that we come up with are weights that are necessary in the context of the existing broader social policy in Vermont. Could you change, maybe a fun geeky academic study for me, if you change those contextual policy structures and we go back and we run these models, do we see that the weights have gone down for poverty because we've better addressed it elsewhere? And then it's also a virtue of this strategy is that updating the analyses and data given policy changes is not incredibly difficult to do. So we can test whether policy changes result in different weight structure changes and recalibrate formulas. I think more easily than we can with other methods. If I might add something, I think it's also, your question also raises a question of tax policy too, right? So when we have health and human services programs, those are funded out of the general education fund. They're not funded out of the fund. And to the extent that those programs and policies and practices are underfunded or have had cuts, the schools still need to serve struggling students who come in, right? They're still responsible. And so what you see is the schools having to put in place washroom programs, mental health services, things that they might historically perceive to the community. And those are additional education costs. Well, the head fund pays for those which comes out of property taxes. What comes out, right? So I mean, this is actually a question, there's a taxation question that's embedded in your remark as well, right? Which is to the extent that the general fund is underfunding health and human services that are necessarily for strong families and communities and students, that doesn't mean that students don't have needs, right? If they can't access those services outside of school, that the schools, in order to ensure that the schools have safe place of learning, students can achieve common outcomes, they're having to spend more. And we've heard this over and over and over again in this study around mental health and trauma, right? Boom. Well, those increased costs in school budgets are funded in a fundamentally different way. They're paid for with different dollars, right? And so I think we also have to think about how are we paying for those services? Are we paying on the dental letter? Are we paying for a medical? The other question ties into where has it been determined that schools would pick up the cost of mental health and impact of trauma? I mean, that's to me a health issue. I mean, I feel like that. I don't know that it can still determine that. I think that what you see is carrying confidence in educators in the field are responding to an emergent and immediate student needs around safety, around student mental health. But those are directly related to educational outcomes too, right? And so I think this is a systemic question that goes far beyond waiting in this particular study. But we certainly in our stakeholder interviews heard these kinds of sentiments come up across not just from our educators in the field, but from the school policy, maybe. It follows up all the different purposes. The school is on the train to address them. Some of these questions I think we can answer in committee. Okay. Thank you. Representative Young. I mean, obviously there's a really significant difference between the poverty factor. I'm wondering what percentage of Vermont students are poverty that would be weighted at that level? On the measure that we used for the simulation guide, that's what I don't recall. If you can be maniacal, look at that. I just want to be precise. The other thing is to remember, is that difference by district? Yes, that's right. So they need to turn on average, how many are gonna generate that much weight? So it's very maniacal. Yeah, I was just wondering how would I make, just in terms of a question and answer, we'll be spending time in the next several weeks providing just the baseline information about this report in committees. So if you have further information, and Dr. Chloe's agreed to be available, and we'll be planning to do that as best we can to educate everyone on just what's in the report. Thank you. It's gonna come. Well, let me have one other question. There is a dead, for example, I went, as many of us probably did, when we looked at the impact on our own schools that we might represent, and I found, for example, in the tables covering ruralness and population density, none of the very rural, small schools that I represent. And I just think it's important to point out that the information really cuts to a point in time when some districts are unified, some are not, and that I assume that if we were to use that to adopt policy, we might need more updated information as to how that all shakes out. Absolutely, and what we've done is we'd want to come up with, when it comes to the sparse population, sparsely measure of the size, the size measure is really just to roll with people's, but you need the most recent, and you need a comprehensive version, and you need something that's at least a, directly comparable to what we've tried to model on if you're gonna use it as a cost value, but you're gonna have to figure out for legislation purposes, what are the exact data and measures that would be used that will cover all districts and will be, you know, updateable in a timely and not fashion to run the formula each year. So yeah, we're basically able here to provide illustrations with the best available data we had retrospectively through 2018, and we're already in 2020 now, so. And to represent the column's point, also just the population density measure that we're applying is the population density of the district in which the school resides. So it's not the population density of the school, and the reason for that is under Act 46 with unified governance, the idea is that it's a sparsely populated district and this is a school that's locally within that district. That also might be. Right, although we still have some schools that are, that can be single schools, single districts. That's right, that gets at the Act 46, but I wanna be clear that the population density measure is the population density of the district in which a school. So this is a more global assessment of population density than it is before micro one. And so my question is better than more specific. On the recommended weight tables, you had said, you know, you really need to adopt the full column as follows. So what if you were to pull a row out, like let's say we decided ELL should really be something that's supported categorically rather than with a student weight. How does that change the policy decision? I would argue that the basis for funding that separately would have to be derived from a calculation backed out of the weight. If we wanted to approach ELL funding separately from the outset, the strategy we could have taken empirically and the process would be analogous to what we did for special ed, can we estimate separate models, not accounting for ELL, to see how ELL and poverty are overlapping so that we can roll some of that back in the poverty way that the equalized people count and take the rest out in this separate ELL block. I think it's problematic to try to take something like ELL out because of the way these things do fit together. Arguably it can be a little problematic because of the overlay of special ed and poverty, but I think we came up with a fairly useful strategy for either keeping that effect in the equalized people count or adjusting the special ed block. Yeah, I guess with ELL it's essentially a... Peter. Elena. All right, thank you. It's all right. We'll take it to our committee. I represent it too. Thank you for a very nice presentation and a very interesting report. I want to say also thank you for the suggestion of the categorical aid for mental health and trauma. I think our schools have become much more trauma-informed over the last 10 years. And I think that they are eager to do more. And I think that that is such a huge need and whether you're funded, like, you know, we fund Medicaid to schools for other health stuff. I mean, it doesn't have to come out of the ed fund. It's just, you know, it just shouldn't be done. I was confused a little bit about your geographical necessity definition. I'm thinking you can't get from point A to point B in the winter because the mountain pass as well. You guys seem to be thinking about density of the population. So that's a really fair question. And certainly that's something that you sort of noticed before it was considering. So there are lots of different ways to think about morality and the additional costs of operating schools in sort of geographic challenge areas. And certainly other states have tried to do that. I think when we were looking through this, our goal was to identify, as the person was talking about, measures that are, first of all, captured variation well. But we also need reliable and valid measures that can be replicated over time easily, data are easily collected and maintained. The population density measure is something that the agency of education has created and is using in their work right now. And so it's something that's maintained in terms of the indicator. It also, in our models, captures variation really well. Does that not mean that another indicator could be used? Yes, that's true. But ours was a judgment call as to something that was transparent, something that was reliable, valid in terms of knowing that we were measuring what we were measuring. And we could do it in an consistent way over time. And something that could be generally by the agency in terms of perpetuity without sort of new resources. These generalizable available measures, which to fit the model and can't necessarily model anecdote, but that's not to say that policy shouldn't account for that. Actually, it's something I ran into in a study in Wyoming that that exact problem was, well, yes, but you can't cross the mountain pass between, you know, October and March. So those, obviously, they're gonna be policy considerations that go above and beyond the generalizable formula. And you just gotta figure out what are gonna be the decision bases on which those are gonna be made. I don't know how many schools, two weeks that would apply. That means we're talking about the small schools, great to be anything else. And then if I could ask you one last one, what outcomes are we trying to equalize? When you were aiming all of your things at equalizing outcomes, what sort of the outcomes you were trying to equalize? We had to, this is one of the really frustrating things on the statistical modeling side of this is you have to work with that because it's consistently measured across schools and districts and students. So our models are fit to achieving equal outcomes on state assessment scores. We wanted to be able to broaden that to graduation rate for, you know, add ideally a system where you're collecting data and persistence and completion into higher ed. That'd be great. It'd be really nice to estimate a model that had a richer set of outcomes or to estimate models across multiple outcomes. In those cases where I've had some opportunity to broaden the outcome measures, the patterns of variation across communities still tend to be relatively consistent where the outcomes are still within the scope of academic and graduation rate, persistence, completion, and so on. Average test scores in reading and math tend to be at least somewhat statistically predictive of those things. So given these kind of links, you know, when at that level of aggregation, and that's not to say that we can go to the level of any one child and say they got this test score and they're gonna make it or they're not. That's a totally different level of judgment call. Something I've written very critically about in a lot of my work. At that level of aggregation, at school and district level, we do tend to see that where poverty is higher and leads to a risk of lower test scores, the associated risk of lower graduation rate or lower college attendance is relatively similar. So if you say you line up, so it kind of works, but it's really unfortunate that we've spent so much of our time only measuring reading and math test scores for third through A creators for so many years, so. Thank you. I'll be quick. Thank you. I am looking at simulation P1, so I have a question about that. Related to the weights, my understanding is P1 corresponds with column three. So with control for the share of SWDs. As I look at that compared to the current weighting, so if, as Representative Conlow was saying, it says 0.25 in statute, we think of it as 1.25, right, status quo. So 3.97 would give you equivalent. So that is more than a three-fold increase in terms of the weighting for an economically disadvantaged student. Now, logically to me, that feels like it would increase the equalized pupil count in any district where you have students with economic disadvantage. Now, of course, if you want to see that's not the case, in fact, the town that I live in has a, we're between 30 and 40% of these students who would fit in that category and our equalized pupil estimate goes up by 1%. So what I'm trying to understand is what are the factors in that column that are taken away? Because I know that some of them are multiplicative, but there must be some factor that is reducing pupil counts. Remember that the equalized pupil count, right? So there are two different things going on here. So I think what you're, I think what I'm hearing you talk about is what your weighted pupil count would obviously go up, right? But remember that then everybody's weighted pupil count and then we equalize it down to the total population. And so it becomes sort of this proportionate response. So what that suggests to me, and I haven't looked at your specific time, there was such a, is that although you went up, you didn't go up as much as others, right? Right, because of the equalization amongst, right? So if you go up to 300, right? But when you see it goes up, goes up even more, right? Remember that all gets sort of deflated back down to the actual AEM, and so it becomes what is your proportionate share of the eating, the average of the eating? Right, the poverty, that the much higher poverty way is going to, in relative position, move the higher poverty districts higher on the equalized pupil count. But in absolute position now, it's much higher because we deflate the whole thing back down to be a zero sum. And then that deflation and categorizing like a coefficient of some more, you know? It's in the simulation. It's actually one of the cells, if you scroll through the calculated columns, you'll see that the deflation factor, it might be the bottom cell in that upper left box by picturing it, right? It's like a 0.54 or something like that. There's this spreadsheet where you can actually enter these things, but you see the deflation factors varies by year. Right. And this is something that we're managing these calculates anyway. Like this is an issue every single year. I think we equalize, we create this long term weighted PK12 AEM, that when we get this number of students, that's far exceeds what we actually have. And then we've got to deflate it down proportionally using, as Bill Talbot language, the factor, right? The factor. Right. And so that factor changes every year, but it changes every year depending on what the total number, the total count is. And so that's why you see that. And it feels like, okay, well, we might not, so shouldn't we've gone up one? Well, it's all relative. Okay, thanks. Does that help? Yeah. Other questions? Yeah. Just this last exchange coming, thinking about something different. I understand that a lot of schools may be upset with their relative position adopted versus unadopted trauma in schoolboy. But what I think what you just said was, factors can change from, I shouldn't say more twice, the number, the calculation could change from year to year based on the factor. The factor may change. The relative position, if you were to adopt our weights as policy, they're gonna drive the relative position. I understand. Right? Now, your absolute position once we, when you make an abrupt change from current policy weights to our weights, your factor is gonna, it's gonna be harder to interpret what's going on when you're changing absolute positions. But your relative position is gonna be very heavily driven by the poverty weight. I understand. Let me go a little, what I think was my question. Okay. You gave a good answer. All right. So we changed from last year to next year, and that's a big change. There's an adjustment in check out. Under the, under the last year, current, school boards know from year to year that high school students are worth this much, you know, pre-K this much, that I think in running calculation, and they come up with a number that's, that using those numbers is predictable. Right? But if I understand it, if we collectively adopt your model, and we adopt it up, and poverty factors change within school districts from year to year, there's another thing that will move from, right? The weighted factor will change the outcome as poverty rates go up or down in different school districts. Well, there'll be a more abrupt shift than when you see that. There's gonna be an abrupt initiative from current policy to new policy. There will be, now, if you adopt as part of this an annualized updating that picks up the drift over time, there will be more subtle shifts from year to year after that, but they should not be so significant, and poverty does not change itself abruptly one year of that next. So those shifts should be subtle moving forward, but abrupt initially if you did it all at once. And that was the case with that one, I don't mind what I was trying to get at, subtle shifts over time after we do the change. That's right. Right, now the interest in twists is because it's all based on then the local decision about how to adjust tax rates. Yeah, yeah, I understand. Then it's, you know. Thank you. That's good. Then the other thing to keep in mind is as was the case with that one, 73, there can be a phase in the period. Right, like this isn't, right, this isn't, well, but my point, I think you raised my point, is that- She was a bad change this thing too. No. I think he's in the local response. Yeah, my point being is that you raised a good point. Is that if any kind of significant shift in the wait is going to be in your point too, it's going to create some significant shifts relative, right, and that's a relative policy. Thank you. Sorry about that. Other questions? Your attendance resignation. It's a national emergency. I just wondered, I wonder if you got the poverty info for Representative Young. I'm going to get back to that number. Oh, okay. I'm so excited. I wrote it down. So I, okay, I'm going to go back to you. I just- I'm floating through an answering question. I don't want to make a mistake on that. No problem. Because it's all applied by the factor at the end. The aggregate cost-effectiveness. But what I'll do is I will, when I respond, I wrote down that question and also represent Cullen's question. When I respond, I'll read the representative's web and Anselin, and she can distribute the responses. Cool. And I'm assuming from the simulations that you're assuming the same amount of money. You're not assuming new money goes into the system. That's exactly right. Right, so this is fiscal 2018 approved budgets. So this is zero rights. This is rearranging debt chips, right? Like, this is not, we're not assuming any new money, right? And part of the strength in doing the retroactively is you can see how those shifts would look like, right? But what happens when school budgets are part past this March? You know, the games start to all work out. And you made, I think the correct point, that the fact that school has more poverty and they have lower tax rates doesn't mean they're going to spend that money on better programs for economically disadvantaged kids. But the flip side of that is that we don't know what would happen in those districts that are going to have higher tax rates. We don't know if they're gonna reduce spending and do less or, you know, sort of what, I mean, there's an awful lot of guesswork and assumptions and so on about how this might play out even if it's based in a miscreant quits. So, yeah. Cutting off of that in some way. I mean, given that this is a zero sum game and what it really does is affects tax rates. Is there, if we looked at addressing some of these issues such as poverty by weighting, are we better off considering it more a vital argument? It's a policy decision. I know and you don't like to make those. That's not our, that's not our lead. It's interesting in a context of a formula that does push it back onto altering the local tax capacity as opposed to setting a target funding level. Could you get around that by funding something directly? I think, you know, one of the areas, some of the research in my area suggests that the more you allocate as a categorical grant, the less efficient expenditures become, right? And that's part of the basis for why California finally moved to a unified kind of weighted foundation a formula as opposed to whacking the array of categorical grants. Categorical grants force the expenditure on certain things and then the administrative structures to monitor the expenditure on certain things and ways that lead to unfortunate inefficiencies. But there is that weird deal here where, well, if we put it into the equalized pupil county it could just be taken as tax relief and might not be the same on the programs that are needed. So. The other thing to take into account, right? Talk straight off. When we looked at all those different sort of policy tools, right, categorical grants, sort of public finance literature and sort of fast practices says that you can use categorical grants for specific and targeted needs, right? So a categorical grant would be for a specific program policy or practice, right? And those grant funds are restricted to that program policy or practice. And so part of the decision there is, is sort of unpacking what we said or said well, is if you provide dollars for low income students and that's just general aid, then that's where it's sort of it's, that's at odds with sort of the best practices and sort of how we think about using categorical funding mechanisms wisely in public finance. If it's for a specific program or purpose or policy then there may be some arguments, but as Dr. Baker said, you know, whenever you segregate funds you introduce different kinds of bureaucratic infrastructure, sometimes those are appropriate, sometimes they're not. And they come right through. So especially as a perfect funding formula, right? And so these are the trade-offs you have to weigh. And of course my question does read back the fact that 173 had to do with moving away from categorical grant. I did. Peter Anthony. Peter Anthony. Maybe it's unfair to ask this, but I'm using the form for what, the opportunity presents. My community is among the low spenders. And the portrait that you've painted about, the education institutions, the school board putting out essentially here's the resources we need. Given all the things that you said, my frustration as a municipal person and as a debater in this forum, the difficulty is you get that figure, the voters see that figure. How do you get around the but it won't pass factor? And I understand the inefficiency of categorical, but the point is once you strip out some expenditures, then you can get it passed. But you can't get it passed if it's in the general. And I don't know if this is the way in which our election laws have said this is the way you must present the question to the voters because I'm not sure people understand ultimately what it costs them when they look at the warning. And that's not your problem. But I just, it's a frustration of being a low spender and forever being a low spender unless we can get over the it won't pass properly. Is there another question? I'm just wondering in your research, if you found a district and I don't want to know what district it is. But that was a painting very high outcome, like glaringly or not surprisingly high outcomes for the amount of poverty in Vermont. I didn't look, I have a forthcoming study where we're going to be looking at that nationally. And we're going to then be digging into things based on the national cost model. What are the kind of these districts at the boundary that, given their conditions, seem to be achieving more than expected, or less than expected under the circumstances that they face. And the other end of it, we're going to be doing kind of deep dive into a number of districts around the country. On that, I actually don't think in the most recent run of my national model that there are any Vermont districts that popped out on the edges of the national distribution. So that I can say that I intentionally didn't look at, because that's one of the really thorny questions, because when you start looking at that district, and you start looking at the other two. So, yeah. Thank you. And Secretary Friends, you are going to be your team is going to be talking with us about the MOOC scores. Yeah, absolutely. And you can just aggregate some of the data at least on that. Yeah, I just signed this last conversation. It's when we think about that flow chart and decision making one of my initial reactions to the report and sort of following up on the chairs, both chairs, their observations. And we think about where the policy might break off and as we have to understand the motivations of how folks make decisions at the local level. And I would argue that the Equalize Pupil function is brings a different reaction from a budgeting standpoint than the Catechorical Grant. So how districts make decisions around addressing Equalize or education spending overall as a function of trying to make cuts or what have you. And then how they use the Catechorical Grant to offset their liability on the Ed Fund. Those uttered dynamics around motivation or incentives and behavior are ones that should probably inform which fork of the road we go down during the policy. We will each have an opportunity. You're willing to come back as we sort of go into our report issue. Thank you so much. Thank you. You're welcome. You're part of the big work. Thank you.