tag:blogger.com,1999:blog-5880610.post2552944838849208535..comments2024-01-13T18:57:18.243-05:00Comments on Information Processing: Value-Added College RankingsSteve Hsuhttp://www.blogger.com/profile/02428333897272913660noreply@blogger.comBlogger16125tag:blogger.com,1999:blog-5880610.post-39226544461980463962015-05-07T22:36:11.142-04:002015-05-07T22:36:11.142-04:00For example, Reed College has a reputation as a go...For example, Reed College has a reputation as a good undergrad college for students who want to become college professors. Is this true? If so, is it a selection effect or a nurture effect?stevesailernoreply@blogger.comtag:blogger.com,1999:blog-5880610.post-66661423886975458792015-05-07T22:35:04.023-04:002015-05-07T22:35:04.023-04:00I'd like to see more focused studies that do m...I'd like to see more focused studies that do more apples to apples comparison. For example, if I'm a high school senior and I want to go to an undergrad school that will give me a leg up in getting accepted at a Top 14 law school, are there any colleges that overperform relative to their degree of exclusivity? In other words, I already know that getting into Harvard College is a good stevesailernoreply@blogger.comtag:blogger.com,1999:blog-5880610.post-64153650530596266222015-05-02T04:51:18.614-04:002015-05-02T04:51:18.614-04:00They seem to put proportion of STEM graduates into...They seem to put proportion of STEM graduates into the "actual outcomes" rather than "predicted outcomes" set of variables. This would bias their study in favor of engineering schools. The proportion of STEM should be considered a student characteristic and used to predict outcomes. A history major earning $80,000/yr is hitting it out of the park compared to an engineer Corneliusnoreply@blogger.comtag:blogger.com,1999:blog-5880610.post-63683555806389240542015-04-30T14:44:31.011-04:002015-04-30T14:44:31.011-04:00Steve, this is a bit off topic, but I'm assumi...Steve, this is a bit off topic, but I'm assuming you're against affirmative action. If you were dictator and had total control over elite college admissions, how would you change it? Would you replace the holistic admission process with a national entrance exam, which is common pretty much everywhere else? The Ivy League schools seem to believe that if there are too many Asians, their Davidnoreply@blogger.comtag:blogger.com,1999:blog-5880610.post-40734650379529168122015-04-30T14:15:51.651-04:002015-04-30T14:15:51.651-04:00In market economy, value of any thing is determine...In market economy, value of any thing is determined by supply/demand ratio. High demand/low supply ratio creates high value.<br />If ivy leagues offered universal education for all, then their value would be no difference from high school education.AGnoreply@blogger.comtag:blogger.com,1999:blog-5880610.post-51200241611655098622015-04-30T13:02:35.611-04:002015-04-30T13:02:35.611-04:00This.
If salary data is coming from PayScale.com,...This.<br /><br />If salary data is coming from PayScale.com, how representative can it be? How do they try to account for selection bias?HomoSapiensnoreply@blogger.comtag:blogger.com,1999:blog-5880610.post-88845584783503714492015-04-30T00:55:42.905-04:002015-04-30T00:55:42.905-04:00The problem of these rankings (prestige, value add...The problem of these rankings (prestige, value added, etc.) is that they always refer to the past, i.e. reflect the situation of ten or fifteen years ago and have uncertain predictive worth.disqus_ZsHlaWpOsFnoreply@blogger.comtag:blogger.com,1999:blog-5880610.post-74046139146212205452015-04-29T15:45:43.760-04:002015-04-29T15:45:43.760-04:00The self-reported PayScale data is suspect. PaySca...The self-reported PayScale data is suspect. PayScale and Glassdoor and such ilk have value but their error range is hard to quantify and I would guess a strong bias to the high end. People lie and the type of people looking for this info is not a random sample. Outside of guys like Steve with their earnings published so we can all know it real info is hard to get.Bobdisqusnoreply@blogger.comtag:blogger.com,1999:blog-5880610.post-62979199670156012102015-04-29T15:33:44.577-04:002015-04-29T15:33:44.577-04:00Ok, you graduate, can you get a job?...If you are ...Ok, you graduate, can you get a job?...If you are an American citizen, maybe not. "A restructuring and H-1B use affect the Magic Kingdom’s IT operations"yulvanoreply@blogger.comtag:blogger.com,1999:blog-5880610.post-78255510976452293302015-04-29T13:02:45.812-04:002015-04-29T13:02:45.812-04:00I'm disappointed they did not have enough data...I'm disappointed they did not have enough data to make predictions for the service academies. I have a feeling they would top the added value list. 3 of the top 16 median salaries (Harvard is 17) with mid/high math scores (~2 vs. ~3.5 for MIT and Caltech). It looks like the data they lack is that concerned with student loans (which makes sense). It would be interesting to rerun the Richard Seiternoreply@blogger.comtag:blogger.com,1999:blog-5880610.post-78154503630398286252015-04-29T12:56:39.029-04:002015-04-29T12:56:39.029-04:00Agreed.
The question I was trying to get at with...Agreed.<br /><br /><br />The question I was trying to get at with the $5k comment was what in the respective college profiles caused that difference in predictions?<br /><br /><br />It was interesting to sort the data by highest predicted salary (only ~1140/7370 colleges have that though). The Ivy League is well represented at the top of that list.Richard Seiternoreply@blogger.comtag:blogger.com,1999:blog-5880610.post-42836505958439984362015-04-29T12:19:02.464-04:002015-04-29T12:19:02.464-04:00They regress against student test scores and famil...They regress against student test scores and family income and use them in the predictor model.<br /><br /><br />I'm thinking that there may be "unobservables" that differ between students going to different schools.JayManhttp://jaymans.wordpress.com/noreply@blogger.comtag:blogger.com,1999:blog-5880610.post-13622627022388551242015-04-29T11:56:11.109-04:002015-04-29T11:56:11.109-04:00To be honest differences between the top few schoo...To be honest differences between the top few schools in this ranking are likely noise dominated. Do we really know mid-career median salary to within $5k for MIT vs Caltech? ...steve hsuhttp://plus.google.com/u/0/+StephenHsu/aboutnoreply@blogger.comtag:blogger.com,1999:blog-5880610.post-81532743559165280902015-04-29T11:33:00.396-04:002015-04-29T11:33:00.396-04:00The regression coefficients are available in the f...The regression coefficients are available in the full report (52 pages) Appendix Table 4 (pp. 36-37). They also give some model information (e.g. adjusted R^2).<br /><br />The only (direct, not demographic) student quality metric I see is math admission scores as Z scores. Caltech shows as 3.59 and MIT as 3.50 which indicates a little resolving power at that level. My understanding is the Richard Seiternoreply@blogger.comtag:blogger.com,1999:blog-5880610.post-60982039608991907212015-04-29T11:12:31.353-04:002015-04-29T11:12:31.353-04:00Ivy League: ouch!Ivy League: ouch!ZZZnoreply@blogger.comtag:blogger.com,1999:blog-5880610.post-36375256578141383812015-04-29T09:33:53.103-04:002015-04-29T09:33:53.103-04:00a college’s value-added measures the difference be...a college’s value-added measures the difference between actual alumni <br />outcomes (like salaries) and predicted outcomes for institutions <b>with <br />similar characteristics and students.</b><br /><br />I'd like to know how they establish that.<br /><br />Note Caltech grads are also much more likely<br /> to win Nobel Prizes and be elected to the National Academy of Science <br />or JayManhttp://jaymans.wordpress.com/noreply@blogger.com