<p>Which factors affect graduation rates at national universities the most?
based on 202 universities</p>
<p>Here are some things I looked at and the correlations with graduation rate:</p>
<p>.83 SAT CR 25th percentile
.83 SAT math 25th percentile
.80 SAT CR 75th percentile
.82 SAT math 75th percentile</p>
<p>.83 Peer Assessment from US News
-.64 admissions percent
.48 public (=1) or private (=2) (privates have higher graduation)
.46 endowment per student FTE
.45 number of faculty members in National Academy of Sciences
.33 percent of expenditures for research
-.24 percent of expenditures for public service
.21 percent of students in engineering, math, science, technology
.16 percent of expenditures for instruction</p>
<p>the above are all statistically significant correlations</p>
<p>the non-significant factors included:
-.11 yield
.06 size of freshman class
.03 number of bachelors degrees awarded
.11 percent of expenditures for academic support</p>
<p>SATs indicate the quality of students
PA indicates the quality of academic programs</p>
<p>collegehelp, I think you need wealth of the students or their families in your data. How much students really want to go to college? Other possibililties for the students? Cost of the school? Health of the student or family? Married or single? White trash, or takes out the trash?</p>
<p>dstark-
I agree with you. I entered all of the above factors into a multiple regression and was able to account for 84% of the variability in graduation rates. That means 16% is not accounted for and is probably due to factors such as those you mentioned.</p>
<p>83% is rather high, but in case you care to explore further:</p>
<p>I wonder if the 25%ile would predict with yet greater confidence if a variable is added to reflect how it compares to the mean of the college as a whole, vs. as a predictor solely by itself. Meaning perhaps a 25%ile score of 400, where the median of that college is 700, predicts a higher dropout rate than a 25%ile score of 400, where the college’s median score is 600. So maybe two variables: one being the 25%ile by itself, another being 25%ile as % of median, or something similar with greater thought behind it, and see if the combo predicts beter than solo…</p>
<p>Then yes, some kind of economic indicator.</p>
<p>Then also it would be interesting to see of there was any discernable contribution due to location and/ or school size; would have to create dummy variables and decide where/ how to classify each type of location & size. Then one can explore whether students tend to dislike certain environments porportionately more than others, as a statistical matter, and therefore transfer out more.</p>
<p>Okay, I have a headache just looking at the numbers in the OP’s post. Could someone use the following school as an example and give guidance to me as to why the graduation rate is so low?</p>
<p>The school is University of Asheville. This is a small public U in NC. They are a liberal arts college, so you won’t find certain majors, so deciding to major in engineering might be a good reason to transfer out. The area I am told is nice. There are some atypical (older) students. I would really like to understand why so many do not graduate, or if this school is used as a stepping stone to UNC-CH, or elsewhere.</p>
<p>monydad-
I looked at school size and it was not a factor with correlations below .1.</p>
<p>Location could be a factor but how would I quantify that? I wish I had the school’s latitude. There might be a confounding factor since many of the best schools are in the northeast. It could lead us to conclude that students love cold and snow. :)</p>
<p>It would take a long time to code schools individually as urban-rural. I would have to look many of them up.</p>
<p>What kind of financial factor could I add? I think there could be a negative relationship between financial aid and graduation rate since much aid is need-based and poor students get more aid. I would have to know income levels, which I do not.</p>
<p>I was struck by how strong the correlation was between SAT and graduation rate, considering that some schools have stopped requiring it and some posters say SATs don’t predict much.</p>
<p>SATs and private vs public are a decent proxy for income. Income goes up as SAT scores goes up. You could try looking at % of students with need based aid. The larger the %, the less a family’s income will affect graduation rate. If more people get financial aid, then maybe a school is less likely to have students drop out for cost issues.</p>
<p>Are you using 4 year grad rate or 6?</p>
<p>
is this based on % of degrees awarded in those fields?</p>
<p>northeastmom-
UNC Ashville has a low graduation rate because of the student SATs, the percent of students in the top 10% of their HS class, and because of financial indicators. US News is able to predict their graduation rate almost exactly from their statistics. They underperform expectations by only 1%</p>
<p>Venkat89-
I am using a 6-year graduation rate.
Yes, based on percent of degrees awarded in those fields.</p>
<p>what do you think is the best way to interpret the % of students graduating from science and engineering? Does that mean tech schools are better at graduating students in 6 years (probably not 4)? Or does that mean schools with lower attrition rates in the most difficult majors tend to have less attrition in “easier” majors, as in fewer people dropping out because they couldn’t cut it?</p>
<p>Venkat89-
My hunch is that schools with a higher percent of students in sciences/eng/tech may also have higher SAT scores. The correlation between percent science/eng/tech is actually an indirect relationship through SAT. In fact, my intuition tells me that it is actually harder to graduate from science, engineering, and comp sci.</p>
<p>"Location could be a factor but how would I quantify that? " </p>
<p>I believe this can be done using the so-called “dummy variables” technique, one for rural, one for suburban, one for urban, with rough “eyeball classification” by you (if more than that is too much work). I’m pretty sure one can do this, but since my last econometrics class was in 1983 that’s all I’ve got.</p>
<p>My size+ location idea was based on the notion that maybe a small school in a remote location might tend to have more transfers, for a given SAT stratum, than a larger school in a non-urban location, due to kids at the small school having nothing to do and being more bored. So it was the two factors together I was thinking might have an impact.</p>
<p>So I guess my dummy variables would be one each for: urban + small, Urban +large, non-urban+small, and non-rban+large, suburban+large, suburban+small. Or could break down even further. See if any of them have a significant extra effect. Likewise one could do this for regions of country.</p>
<p>"What kind of financial factor could I add? "</p>
<p>I don’t know, but I bet mini, the czar of financial impacts, could craft an appropriate one for you, if so motivated. Possibly it will involve average amount of debt students graduate with, % pell grants, % on financial aid, other stuff I don’t know of and/or haven’t thought about.</p>
<p>I added some financial indicators to my data.</p>
<p>Here are the correlations with graduation percent:
.06 percent of students receiving a grant from the institution
.53 average amount of institutional grant
-.31 percent receiving any financial aid</p>
<p>When I added these to the multiple regression, it did not substantially increase the amount of variability in graduation percent being accounted for. The information in these financial indicators must already be embedded in some other variables.</p>
<p>The amount of institutional grant aid was a major factor (by itself). The percent receiving any aid had a negative (inverse) relationship with graduation rate. I wonder what is going on there?</p>
<p>Another reason that smaller schools such as UNC-Asheville and other smaller public universities have lower grad rates might be because they have a higher ratio of part-time students, for instance according to a recent fact book on UNC-A’s website there were over 500 part-time or non-degree undergrads and almost 2900 full-time undergrads. In comparison, according to the latest fact book on Harvard’s webste there were over 6600 full-time undergrads but only 7 part-time students (Harvard also has an extension program for adult/continuing ed/part-time students - less than 500 students pursuing undergrad degrees but my guess is that extension students are not included in calculating graduation rates. However many smaller schools do not have separate extension programs, and if part-time students are included in graduation rates, the rates would obviously be much lower at schools with higher ratios of these students than if they were not included… (of course, I am not sure how rates are calculated but there are lots of ways to play with statistics so they can often be somewhat deceiving ;))</p>
<p>"The percent receiving any aid had a negative (inverse) relationship with graduation rate. I wonder what is going on there? "</p>
<p>Maybe if your education has cost you(r parents) a fortune over the first couple years, you make an extra effort/ feel obligated to push through to the end, lest that huge sunk cost be deemed a waste;</p>
<p>but if it didn’t cost you/ them a small fortune/ they don’t have any significant sunk costs involved, then it doesn’t feel like as big a deal to bow out if things aren’t going so great?</p>
<p>BTW, maybe you’re already doing this, but suggest introduce each potential variable in one at a time (“step-wise”), rather than all together at once.</p>
<p>Well if you look at a school like Juniata College, with over double the 4 year graduation rate, how is that explained? UNC-A and Juniata have nearly the same average test scores. In fact, Juniata is test optional, so those students not submitting are not even factored in. Juniata is located in one of the poorest areas in PA, so many students are not from wealth families. </p>
<p>Scansmom, you are absolutely correct about the high number of part timers. The question is why they are part time students. Is it because the work load is too much for these students to go full time? Is it the money (very low instate tuition cost to attend btw)? Is it because there are unmotivated students? Is it because they cannot get the courses that they require to graduate in a timely manner? I think there are reasons behind the high number of part time students, but I don’t know what those reasons are. Perhaps a bit of all of the above. Also, I am sure that some transfer out ,using Asheville as a stepping stone to another UNC school like Chapel Hill, either for prestige or because the major is not offered at Asheville. Then again, one could say that about Juniata (another small school, but a private one), yet they do graduate a lot of students in 4 years.</p>
<p>Many non-poor students elect to pay their own way through college and these students are much more common at the state U’s. This seems especially true in the upper midwest where it is fairly common for middle-class kids to pay their own way in college.</p>
<p>Northeastmom-
Juniata admits 36% of their students from the top 10% of their HS class whareas UNC Ashville only admits 20%. UNC Ashville also has a higher acceptance rate by about 10%. Although the SATs are roughly the same, Juniata students have a better overall profile.</p>
<p>What’s more, Juniata graduation rate overperforms expectations by about 7% according to US News. That’s huge. My guess is that the faculty at Juniata are more engaging with students than at UNC Ashville.</p>
<p>collegehelp, thank you. If I had to guess, I’d say that Juniata’s graduation rate is high because of the classes that one may take are more flexible. You may design your own major if it doesn’t exist with approval. High interest in an area helps motivate! JMO. Avoiding a weak area and being able to substitute and tweak a program moves one closer to graduation in my opinion. I think that the student’s academic abilities are not all that different when they come in (Juniata vs. Asheville), but this is my opinion.</p>
<p>Oh yes, you could measure graduation rate against time spent in a frat/sorority, does a city or rural school have a higher graduation rate, % of professors with thick accents I know must play some role in the process. How about the willingness of the school to accept summer school community college credits or high school AP credits with just a 3? Does the thickness of a dorm wall to block out sound play a role? I suppose everything can contribute to graduation rates. Even weather might effect graduation rates. Yup, on and on…</p>