Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
No, I didn't miss these points. It's not really possible to comment on them without reading the original work, but it doesn't address what I posted.
But my claim is not that the best students aren't in the best districts.
Both of these are actually equivalent (that is, you can formulate an ANOVA in terms of a multi linear regression).
But it doesn't really say what you want it to say or think that it says. For example, if you chose a "better" measure than SES (for example, maybe it turns out that parental education is a better measure), you will get different results, because your model for the "parent" factor is better. It might turn out to be the case that something less tangible, such as parental interest in their childs education is a major factor, and teaching ability might be a more reliable proxy for that than parent SES (remember, according to you, teaching ability is correlated with school district, and the better school districts tend to select parents who are interested in quality education for their child).
1. Maybe I was not clear in separating my opinion from research findings but I am extrapolating that teachers are of better quality in better districts since I have no research that studied that directly.
2. Most of the quantitative measures that effect SES were run in the models Darling used but with any study of course there is the potential for other factors not measured. But of the two, SES and teacher quality, teacher quality had a larger effect in the studies presented. I would further argue that since all of the things you gave examples for would also positively correlate with SES its not possible for them to have a larger effect individual than the sum of all the variables that have a positive correlations with SES.
Quote:
The problem I was getting at is related but slightly different. Suppose you have two variables that each by themselves explain 90% of the variance (which implies they will be heavily correlated). Throw them into an ANOVA and one will come out on top, but that doesn't really tell you which one
is the "more important" factor, it will basically tell you that one of them had a slightly better correlation. If you were to for example replace one of the variables with its square or logarithm, you would find that the correlation went down but the transformation didn't make that variable any less predictive, it only made the linear model less effective because the true relationship is no longer linear.
Actually, an ANOVA will not tell you which one "comes out on top" , you would have to do a post-hoc test to know which has the largest p-value.
Not really sure why you are going into linear vs. non-parametric regression, if you do regression analysis all of those are covered. Obviously the variables are finite in any model so I am not sure why you think the model built from this data set is less reliable than any other model.
Quote:
Basically, there are two factors: "parents" and "teachers". The problem is that they are sufficiently correlated that it's hard to reliably determine that your "teachers" factors aren't partially proxies for "parents" factors (because more responsible parents will tend to insist on good schools with good teachers)
Random assignment would largely get you out of this mess because you can be sure that you were holding all other variables constant when you manipulated one of them. You can't re-assign parents, but it's possible to assign teachers. For example, you get to change the teacher without moving any of the correlated variables. It would be possible to do something like this, but it would probably be difficult to do so in a way that allowed a sufficiently variable sample of teachers.
I think you have a vastly different understanding of research than I do. It is not possible to put children into a sociological experiment on a scale large enough to get enough degrees of freedom to get any sort of significance. All sociological research has this limitation and even more so for education. This is why we have to look at other factors such as SES and teacher quality (defined by quantitative measures). It is also somewhat silly to reject their findings out of hand despite the fact that all of their finding are statistically sound. Do you reject all models regardless of subject made through regression for the same reasons?
Quote:
That's fine, I think it's the measure they should use for the purpose of an academic study. However, the general public tend to judge schools by the SAT scores of the outgoing class. By this measure, you could put the worlds best teachers in front of Irvington high students, and it would still have a reputation as a "bad" school.
Really? Even if you switched them at kindergarten and had them their entire school careers at the "better school"?
Do you really believe that if you put the Millburn kids in a "bad" school starting in kindergarten they would get the same SAT scores?
Since this could NEVER be tested we are just going to have to disagree. As a teacher, I firmly believe that for measures such as the SAT SES is far less important than the school/teacher effects. I have no idea how you would expect SES to overcome the lack of quality teacher and resources in a student's ability to learn say Algebra and Calculus but you are welcome to your opinion.
I had a field placement in Hasbrouck Heights and then substitute taught in Piscataway. Out of the two, Piscataway had better teachers who really cared about the students and better quality education overall. Piscataway also paid its subs at a higher rate than HH or Manalapan as well. And generally, from what I have heard, the best paying districts in the state are the Abbotts for the sole reason that the teachers have to put up with more rowdy kids. The Abbotts have more money since they receive special state funding.
While kids in low income areas may be rowdy and disorderly, dealing with students in ultra-rich areas, especially new money areas, won't be any easier, especially if those rich kids are spoiled brats who, along with their parents, will tell you their parents have better degrees than you do and generally act condescending toward the teacher and do all they could to get their little angels out of trouble each time.
Someone here once mentioned about interviewing for a job in Marlboro's district. They told him/her that the students will tell teachers their parents have better degrees and that when there is an issue between a parent and a teacher, administration will side with the parents. I'd rather be in a district where administration supports and sticks up for its teachers.
What surprises me even more is how districts still manage to rank good even after teachers get bogged down by spoiled brats and their arrogant parents with the whole "my son, my daughter, my way" attitude. I abandoned this profession for other reasons.
not surprising....the lack of respect for teachers shown, especially lately (just take a look at this board), is appaling. we complain about how kids lack respect ..... yet the parents show the teaching profession less respect than your average welfare queen. and we're surprised kids act the way they do?
Livingston is a pathetic school district. Why would they make the unofficial official newsweek list? Take it down clown. Its about opinions.
Yeah, I get it. My post wasn't really about the unofficial official Newsweek list. I don't really give a clown down about that either. I was griping more about Livingston, specifically. I know our school district is far from pathetic and I shouldn't complain, blah blah blah, but my point is this:
Every year when it comes to vote on the school budget, the PTA writes letters that basically say: the property value of your home is directly correlated to the quality of the school system, and the quality of the school system depends on your passing the budget! So vote YES! Let's spend! Open your wallets and bend over.
And where does that money go? It doesn't all go toward bettering the futures and molding the minds of our kids. Far from it, I think. Most of it goes to the top heavy administration and their extravagant pet projects. You don't have to turn the high school into the Taj Mahal in order to garner a good reputation, offer a good education, and even yes, make it onto these unofficial official rankings lists. Like I already mentioned, Livingston just spent millions of dollars renovating and expanding the high school even though there are less kids attending the high school now than in decades past, and even then, the classrooms were never filled to capacity.
In general, too much emphasis is placed on the "quality" of the school system, I think. With a few obvious exceptions, a NJ public school education is already "top notch" compared to most other states. School districts like Livingston spend millions of extra additional dollars on items and agendas that really only benefit a handful of students (and some things like a giant expansion make no sense). In every graduating LHS (or any other high school) class, the majority of students end up going on to state or middle-of-the-road colleges and universities (and some go on to votech education...and there ain't nothing wrong with that...I know some mechanics and contractors, and a chef, that are making more money than other friends with masters degrees). If you're ultimately gonna end up at Rutgers (and I say that as a proud Rutgers grad), you can get there just fine from "lesser" (and I say "lesser" with an eye roll) school districts that don't flush tax dollars down the toilet.
2. Most of the quantitative measures that effect SES were run in the models Darling used but with any study of course there is the potential for other factors not measured. But of the two, SES and teacher quality, teacher quality had a larger effect in the studies presented.
No, it had a higher correlation. It's not really an effect unless you manipulate the variable in question (a "treatment") as opposed to just observing it. I think you're making an unwarranted assumption of causality.
Quote:
I would further argue that since all of the things you gave examples for would also positively correlate with SES its not possible for them to have a larger effect individual than the sum of all the variables that have a positive correlations with SES.
Maybe, probably. But my point still stands. You're observing correlations in the inputs. It's a fallacy to assume that the variable that has the higher correlation score is "more important" in some meaningful way. All it tells you is that it's more highly correlated with the dependent variable (they're all dependent variables really, but I mean the one that is being treated as a dependent variable for the purpose of the study).
Quote:
Actually, an ANOVA will not tell you which one "comes out on top" , you would have to do a post-hoc test to know which has the largest p-value.
No, higher p-value doesn't tell you that one variable is "more important" than another either. It tells you that your model fits better. The ANOVA gives you a break down of the variance without p values. p values tell you whether or not a given factor is significant at a certain level of confidence. Lower p-value doesn't mean that a factor is "more important", and it certainly doesn't imply a causal link from that factor to the dependent variable.
Quote:
It is also somewhat silly to reject their findings out of hand despite the fact that all of their finding are statistically sound.
I don't "reject" their findings. I'm rejecting your interpretation of their findings.
Quote:
Do you reject all models regardless of subject made through regression for the same reasons?
This is not just a regression model, they are observing correlations in related variables (as opposed to an intervention where they change one of the variables). This creates several problems with interpreting the resulting statistics. In particular, you can't draw causal links.
Quote:
Really? Even if you switched them at kindergarten and had them their entire school careers at the "better school"?
Do you really believe that if you put the Millburn kids in a "bad" school starting in kindergarten they would get the same SAT scores?
I don't think they'd ever have a truly comparable educational experience, because at some point the Millburn parents would step up to the plate and make sure that their child received adequate education whether through school, private tutor, or whatever. You know, take responsibility for their children instead of expecting the state to do the same, something that is perhaps difficult for hard core liberals to understand.
I think you'd need to give those Millburn kids a really bad educational experience to offset the effect of having the right parents.
Quote:
Since this could NEVER be tested we are just going to have to disagree. As a teacher, I firmly believe that for measures such as the SAT SES is far less important than the school/teacher effects. I have no idea how you would expect SES to overcome the lack of quality teacher and resources in a student's ability to learn say Algebra and Calculus but you are welcome to your opinion.
Maybe having the student actually attempt the assigned homework problems instead of standing on the corner dealing drugs has some effect on their performance ? I'm not suggesting that they will do just fine with absolutely no instruction (a hypothetical, since again, part of the parent factor is that the parents will ensure the child receives adequate instruction).
I'm suggesting that factors which the student (and therefore the responsible parent) have some control over, play a greater role than what the teacher does. This whole line of argument is lost on hard core liberals, because the notions of personal and parental responsibility are completely foreign to them. Instead, everything is the responsibility of the state.
No, it had a higher correlation. It's not really an effect unless you manipulate the variable in question (a "treatment") as opposed to just observing it. I think you're making an unwarranted assumption of causality.
Maybe, probably. But my point still stands. You're observing correlations in the inputs. It's a fallacy to assume that the variable that has the higher correlation score is "more important" in some meaningful way. All it tells you is that it's more highly correlated with the dependent variable (they're all dependent variables really, but I mean the one that is being treated as a dependent variable for the purpose of the study).
No, higher p-value doesn't tell you that one variable is "more important" than another either. It tells you that your model fits better. The ANOVA gives you a break down of the variance without p values. p values tell you whether or not a given factor is significant at a certain level of confidence. Lower p-value doesn't mean that a factor is "more important", and it certainly doesn't imply a causal link from that factor to the dependent variable.
I don't "reject" their findings. I'm rejecting your interpretation of their findings.
This is not just a regression model, they are observing correlations in related variables (as opposed to an intervention where they change one of the variables). This creates several problems with interpreting the resulting statistics. In particular, you can't draw causal links.
I don't think they'd ever have a truly comparable educational experience, because at some point the Millburn parents would step up to the plate and make sure that their child received adequate education whether through school, private tutor, or whatever. You know, take responsibility for their children instead of expecting the state to do the same, something that is perhaps difficult for hard core liberals to understand.
I think you'd need to give those Millburn kids a really bad educational experience to offset the effect of having the right parents.
Maybe having the student actually attempt the assigned homework problems instead of standing on the corner dealing drugs has some effect on their performance ? I'm not suggesting that they will do just fine with absolutely no instruction (a hypothetical, since again, part of the parent factor is that the parents will ensure the child receives adequate instruction).
I'm suggesting that factors which the student (and therefore the responsible parent) have some control over, play a greater role than what the teacher does. This whole line of argument is lost on hard core liberals, because the notions of personal and parental responsibility are completely foreign to them. Instead, everything is the responsibility of the state.
1. You seem to be dancing around "correlation is not causation". Agreed. Would you be more comfortable if I had said the single largest factor predicting student achievement would be teacher quality?
2. Are you suggesting that stronger correlates are not more suggestive of a casual relationship than weaker correlates? It seems to be that you are saying that despite a statistically stronger r that the causal factor is the weaker correlate.
3. I think you are incredibly flawed in thinking that level of parental responsibility is a function of socioeconomic factor. Its bordering on preposterous to say that just because someone is from a lower SES they have a lower level of parental and personal responsibility than a family from the highest SES. Based on that reasoning the wealthiest families have a higher level or parental and personal responsibility than middle class ones. Are you really saying that or are you just saying that about poor people?
1. You seem to be dancing around "correlation is not causation". Agreed. Would you be more comfortable if I had said the single largest factor predicting student achievement would be teacher quality?
Yes, that was my point.
Quote:
2. Are you suggesting that stronger correlates are not more suggestive of a casual relationship than weaker correlates? It seems to be that you are saying that despite a statistically stronger r that the causal factor is the weaker correlate.
No, I'm suggesting that the strength of the correlation doesn't tell us what the causal dynamics are -- they don't tell us whether A causes B, B causes A, or some latent variable C causes both A and B.
Once you have a statistically significant correlation, it certainly suggests that there is some causal relationship, but the strength of the correlation doesn't shed any light onto which of those three is it.
Quote:
3. I think you are incredibly flawed in thinking that level of parental responsibility is a function of socioeconomic factor. Its bordering on preposterous to say that just because someone is from a lower SES they have a lower level of parental and personal responsibility than a family from the highest SES. Based on that reasoning the wealthiest families have a higher level or parental and personal responsibility than middle class ones. Are you really saying that or are you just saying that about poor people?
No, I'm not saying that parental responsibility and SES are the same thing. I wrote elsewhere along the lines that parental responsiblity is probably the key parental factor (e.g. I'm hypothesizing that i is one of the latent "C" causal variables), and that while this is correlated with SES, it isn't exactly the same thing. Moreover, it is probably also correlated with teacher effectiveness (because more responsible parents will look for better schools that have better teachers).
So I'm actually saying that SES could be used as a proxy for "parental responsibility" but it's a fairly noisy one and hence the goodness of fit will suffer, resulting in relatively low r scores.
No, I'm not saying that parental responsibility and SES are the same thing. I wrote elsewhere along the lines that parental responsiblity is probably the key parental factor (e.g. I'm hypothesizing that i is one of the latent "C" causal variables), and that while this is correlated with SES, it isn't exactly the same thing. Moreover, it is probably also correlated with teacher effectiveness (because more responsible parents will look for better schools that have better teachers).
So I'm actually saying that SES could be used as a proxy for "parental responsibility" but it's a fairly noisy one and hence the goodness of fit will suffer, resulting in relatively low r scores.
I am still not following this at all. The first paragraph above seems to be saying that SES and parental responsibility are positively correlated but you started out by saying "No".
Do you think that people with lower SES have a lower sense of parental responsibility?
I am still not following this at all. The first paragraph above seems to be saying that SES and parental responsibility are positively correlated but you started out by saying "No".
Do you think that people with lower SES have a lower sense of parental responsibility?
I think they are correlated, but I don't go so far as to assert that the rank order correlation is 1.0 (indeed, I believe it would be quite a lot lower than that) which is the position that you appear to have attributed to me.
I think they are correlated, but I don't go so far as to assert that the rank order correlation is 1.0 (indeed, I believe it would be quite a lot lower than that) which is the position that you appear to have attributed to me.
I think the entire premise is flawed because you are basing it on YOUR personal definition of parental responsibility. Working class families may have just a strong sense of parental responsibility and maybe even higher when you consider how much of their income is spent on their children as compared to wealthy families. But their roles as parents are going to be self-defined differently.
For example, wealthy families maybe likely concerned with ensuring their offspring make as much money as their parents do while working class families maybe more concerned with their children's behavior.
So unless you can come up with a universal definition for parental responsibility common to all SES and a way to quantify it, your whole contention that they are correlated is basically crap.
Last time NJ monthly did the survey in 2008, Haddonfield Memorial made #14. That was the only one though.
Are Haddonfield's schools going downhill?
No, I'm not sure about this list. It can be slanted. I have a hard time believing that no school from the Philly burbs made this list. It's tough to know what criteria they are using...
Please register to post and access all features of our very popular forum. It is free and quick. Over $68,000 in prizes has already been given out to active posters on our forum. Additional giveaways are planned.
Detailed information about all U.S. cities, counties, and zip codes on our site: City-data.com.
Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
