Altius FL #2 BS section #2

[akimhaneul]

Can anyone tell me how to interpret graphs with stars over every single bars? Does this mean every bar is statistically significantly different from each other?

Thanks!

---

Members don't see this ad.

 

[aldol16]

It's in the caption. It means they're all statistically significant with p < 0.001 compared to control. They're not being compared to each other.

 

[akimhaneul]

It's in the caption. It means they're all statistically significant with p < 0.001 compared to control. They're not being compared to each other.

I forgot to include this but they said in the passage that IHH is the normal liver cells, HEP3B and HEPG2 are liver cancer cells, and HELA is nonliver cells. The question is asking to compare the normal liver cells and liver cancer cells. So does this mean that the stars over the bars indicate statistically significant difference between the bars?

 

[aldol16]

The control apparently isn't shown in the figure. Which means it's a really ****ty question from a test prep company. But typical. You can compare them using not the stars but the error bars and see if any overlap.

 

[desperadoflakes]

Yeah, whenever you see this, it means statistical significance.
Specifically:

ns = not significant
* = p < 0.05
** = p < 0.01
*** = p < 0.001

In this question, the stars mean that the difference between each bar and the control is significant. The control is shown in the next figure - its the same lucerifase reporter plasmid also transfected into HELA cells but carrying its normal 3'UTR instead of the gene UTR. They called its activity 100% and set the other values to that

 

[Altius Premier Tutor]

With all due respect @aldol16, you have no idea what you are talking about.

The control is not in the figure because the entire figure is normalized to the control. Note the y-axis label that says "% of control." The fact that the control is not in the figure not only makes perfect sense in this case, but is NOT an example of a ****ty question from a test prep company. Typical arrogant student.

The graph shown in this passage is EXACTLY how it was shown in the published peer-reviewed article: https://www.google.com/url?sa=t&rct...nV51SUjAw3jwRcqvA&sig2=Yo1MqhFF0dSWGgsty92z7A

I think you had better call up those authors and tell them how ****ty their graphs are.

I highly recommend the strategy of trying to learn how to better analyze such questions, rather than just calling them ****ty because you missed the quesiton or did not understand it. I think that was the OPs original intent.

 

[aldol16]

With all due respect @aldol16, you have no idea what you are talking about.

The control is not in the figure because the entire figure is normalized to the control. Note the y-axis label that says "% of control." The fact that the control is not in the figure not only makes perfect sense in this case, but is NOT an example of a ****ty question from a test prep company. Typical arrogant student.

The graph shown in this passage is EXACTLY how it was shown in the published peer-reviewed article: https://www.google.com/url?sa=t&rct...nV51SUjAw3jwRcqvA&sig2=Yo1MqhFF0dSWGgsty92z7A

I think you had better call up those authors and tell them how ****ty their graphs are.

I highly recommend the strategy of trying to learn how to better analyze such questions, rather than just calling them ****ty because you missed the quesiton or did not understand it. I think that was the OPs original intent.

I would recommend you get off your soap box before you fall and damage your head further. I was a student - undergrad many years ago and recently a graduate student. Unfortunately, I'm no longer a student but rather a post-doc whose job is to gather, analyze, and present data just like this. You might want to figure out who you're talking to before telling a PhD chemist that he doesn't know what he's talking about.

If I or any of my peers were reviewing this article, I would send it back with the strong suggestion that they include specifically what control(s) they are normalizing to. Transfected Hela cells? Transfected Hep3B cell? Both? Neither? I'm sure that they mention specifically what controls they did in the paper and include that information in the experimental or SI. It's a ****ty question not because they presented the data in this way but rather that your company omitted that data in trying to construct a coherent passage.

Why don't you use your analytical reasoning abilities and tell me how to interpret the bar graph for Hep3B? They transfected Hep3B cells and obviously are measuring luciferase activity as a percentage of something. Based on what you've given in the passage, can you tell me exactly what that "something" is? My guess is that it's Hep3B cells transfected with luciferase gene without the UTR. But since they say "% control," "control" is singular, and your company failed to provide any further information on the "control" or controls, that implies they are comparing to a singular control. Which would make me question why they're comparing to the same control when they're transfecting different cell lines/types.

Edit: Looked at the paper and that's exactly what they're doing - they transfected each of those cell lines with plasmids either containing the UTR or not and the "control" is actually "controls." In other words, each of those bars are significant as compared to the control from their own cell line, not some universal "control." Knowing one's controls is always a good thing in science - hope you take that into your career in the future.

This also coincidentally makes Altius' answer choice wrong. The question the p-test answers is: what is the probability that the observed difference between the UTR-containing cells of a particular cell line and the non-UTR containing controls in that same cell line arises randomly? Thus, one can only say that the probability of that difference between UTR and non-UTR cells in the same cell line being randomly this different is p<0.001 or whatever value they use for triple star. One cannot say whether the difference between cell lines is statistically significant. They simply did not do that statistical test and we do not have the data to answer the question.

I would be very surprised if they asserted in the paper that the Hep3B cells were different from the IHH - they simply do not have the numbers to say that and the reviewers would/should have caught that easily. It's a novice mistake.

 

[aldol16]

The graph shown in this passage is EXACTLY how it was shown in the published peer-reviewed article: https://www.google.com/url?sa=t&rct...nV51SUjAw3jwRcqvA&sig2=Yo1MqhFF0dSWGgsty92z7A

By the way, thanks for posting the actual article so now I can go explain the idiotic way Fig. 4 was presented in another thread on this same passage: Altius FL #2 BS section #3

 

[Altius Premier Tutor]

Your responses are case in point as to your arrogance.

But in fact, you did not know what you were talking about in the post to which I responded. Nor would I have used the tone I did in my response had you not started with the expletive insults. You called it a ***ty question because the control was not included in Figure 3. You seemed to suggest it should have another bar? I simply said the original did not. More importantly, the MCAT does NOT specify the identity of the control in every case; it's probably 50/50. In many cases, the AAMC authors will list only "control." Sometimes they'll then ask the examinee to identify the likely control in a question.

I believe the intent of the question was probably to see if students recognized that those asterisks were NOT actually for the differences between bars. The solution makes the point that one must use the error bars in this case, which it would not have if the asterisk-labeled p-values were the relevant statistical test.

I am no PhD, but then I've never claimed to be. Just a stupid hick who can answer 99% of MCAT questions correctly and has proven adept at helping others do the same, including ones like this that seem obvious to me but are often overcomplicated by "minds immeasurably superior" to my own...

By the way, this passage/question set was assembled by a woman who is the department chair at a major university, has been a PhD molecular biologist for some 4 decades, is very widely published, has taught for decades, and has big-time $$$ in research funding. I'm not saying that anyone, regardless of experience, might not author an MCAT passage that could benefit from some clarification, but your bloviation about your recent progression from grad student to post-bac is so ironically distasteful.

 

[akimhaneul]

Your responses are case in point as to your arrogance.

But in fact, you did not know what you were talking about in the post to which I responded. Nor would I have used the tone I did in my response had you not started with the expletive insults. You called it a ***ty question because the control was not included in Figure 3. You seemed to suggest it should have another bar? I simply said the original did not. More importantly, the MCAT does NOT specify the identity of the control in every case; it's probably 50/50. In many cases, the AAMC authors will list only "control." Sometimes they'll then ask the examinee to identify the likely control in a question.

I believe the intent of the question was probably to see if students recognized that those asterisks were NOT actually for the differences between bars. The solution makes the point that one must use the error bars in this case, which it would not have if the asterisk-labeled p-values were the relevant statistical test.

I am no PhD, but then I've never claimed to be. Just a stupid hick who can answer 99% of MCAT questions correctly and has proven adept at helping others do the same, including ones like this that seem obvious to me but are often overcomplicated by "minds immeasurably superior" to my own...

By the way, this passage/question set was assembled by a woman who is the department chair at a major university, has been a PhD molecular biologist for some 4 decades, is very widely published, has taught for decades, and has big-time $$$ in research funding. I'm not saying that anyone, regardless of experience, might not author an MCAT passage that could benefit from some clarification, but your bloviation about your recent progression from grad student to post-bac is so ironically distasteful.

Ok so how do you use error bars to see the difference?
Cause I looked up error bars and learned that if they don't overlap, difference may be statistically significant but you can't be sure. None of the error bars overlap here.


So just because the p value is very small, which mean the results are not due to chance, and error bars do not overlap, we can conclude that the bars are different from each other?

Sent from my iPhone using SDN mobile

 

[desperadoflakes]

Your responses are case in point as to your arrogance.

But in fact, you did not know what you were talking about in the post to which I responded. Nor would I have used the tone I did in my response had you not started with the expletive insults. You called it a ***ty question because the control was not included in Figure 3. You seemed to suggest it should have another bar? I simply said the original did not. More importantly, the MCAT does NOT specify the identity of the control in every case; it's probably 50/50. In many cases, the AAMC authors will list only "control." Sometimes they'll then ask the examinee to identify the likely control in a question.

I believe the intent of the question was probably to see if students recognized that those asterisks were NOT actually for the differences between bars. The solution makes the point that one must use the error bars in this case, which it would not have if the asterisk-labeled p-values were the relevant statistical test.

I am no PhD, but then I've never claimed to be. Just a stupid hick who can answer 99% of MCAT questions correctly and has proven adept at helping others do the same, including ones like this that seem obvious to me but are often overcomplicated by "minds immeasurably superior" to my own...

By the way, this passage/question set was assembled by a woman who is the department chair at a major university, has been a PhD molecular biologist for some 4 decades, is very widely published, has taught for decades, and has big-time $$$ in research funding. I'm not saying that anyone, regardless of experience, might not author an MCAT passage that could benefit from some clarification, but your bloviation about your recent progression from grad student to post-bac is so ironically distasteful.

oh my god its lit

 

[aldol16]

Your responses are case in point as to your arrogance.

But in fact, you did not know what you were talking about in the post to which I responded. Nor would I have used the tone I did in my response had you not started with the expletive insults. You called it a ***ty question because the control was not included in Figure 3. You seemed to suggest it should have another bar? I simply said the original did not. More importantly, the MCAT does NOT specify the identity of the control in every case; it's probably 50/50. In many cases, the AAMC authors will list only "control." Sometimes they'll then ask the examinee to identify the likely control in a question.

I believe the intent of the question was probably to see if students recognized that those asterisks were NOT actually for the differences between bars. The solution makes the point that one must use the error bars in this case, which it would not have if the asterisk-labeled p-values were the relevant statistical test.

I am no PhD, but then I've never claimed to be. Just a stupid hick who can answer 99% of MCAT questions correctly and has proven adept at helping others do the same, including ones like this that seem obvious to me but are often overcomplicated by "minds immeasurably superior" to my own...

By the way, this passage/question set was assembled by a woman who is the department chair at a major university, has been a PhD molecular biologist for some 4 decades, is very widely published, has taught for decades, and has big-time $$$ in research funding. I'm not saying that anyone, regardless of experience, might not author an MCAT passage that could benefit from some clarification, but your bloviation about your recent progression from grad student to post-bac is so ironically distasteful.

Yes, I'm arrogant, you're hot ****, I think we've gotten that out of the way. I think at least one cycle of MCAT takers on this sub-forum can attest to my help on clarifying questions from companies like yours in the past year. I've also taught for four years as a graduate student and now as a post-doc (not sure where you get post-bac) so I have experience dealing with stubborn students who are convinced they are right when they're not. Now on to the substance.

You're right - the MCAT doesn't always tell you the control and asks you questions about the relevant controls. But your company gives the student here no information with which to work. On the MCAT, there will always be enough information with which to deduce the correct controls. For all the students know, all these cells react similarly to transfection and so the only relevant control could be empty vector-transfected Hela cells. The very fact that you keep saying "control" and "bar" speaks to how well you understand that there must be multiple controls here because there are multiple cell lines - a fact that would have made this problem more understandable because the students have no idea what comparison the p values are making. The way you show the data is misleading and scientifically mischievous. In the original paper, the authors make it clear in the figure legend what the controls are.

The error bars may be used to determine whether they are statistically significant differences. But in the answer explanation, Altius' explanation explicitly states: "but is confirmed as the authors rigorously demonstrated this with the p value, which is very small (p<0.001), meaning that it is highly unlikely that the differences observed are due to random variability." Clearly, Altius is stating that the p test is also a valid statistical test in this case (unless you want to dispute that statement). In fact, the p value says nothing about the observed differences between the cell lines as shown in the figure. The comparison using p values is not valid. This teaches students the absolute wrong thing about p values. In fact, if I were compiling this data, I would have shown the 100% bar with each cell line and made clear that the stars refer to the differences between UTR and non-UTR within one particular cell line. The fact that they didn't is a statement as to the strength of this paper. Which makes sense that it's published in an open-access journal with an impact factor of 5-6.

It's great that you have a faculty member with such a strong background writing these passages. Which makes it all the more hard to forgive when she gets a simple statistical fact wrong in the answer explanation. But you don't have to take my work for it since I'm arrogant and only peer review papers like this. Since you know who wrote the passage, why don't you ask that person about this and see what she says? Simple.

 

[aldol16]

Your responses are case in point as to your arrogance.

Also, since I'm such an arrogant prick, why are you still following me? Perhaps that was your predecessor? Genuinely curious.

 

[Altius Premier Tutor]

Let's put this to bed. I never said you were a "prick". I did call you arrogant, and probably should not have. Please accept my apology. Putting yourself in my shoes, it's not the most humble or friendly thing in the world either to lead out with calling the Altius passage ****ty, nor generalizing that such ****ty passages are "Typical," nor calling me/Altius the "idiot test prep company." If you find what you think is a problem with any Altius exam in the future, just PM me and I'll pass it up the line for further consideration.

I don't know the author who compiled that passage personally, but Altius does have every passage marked w/ the authors and PhD reviewers so our internal curriculum team can go back and clarify issues as they arise. I'll share your comments with my boss and he can decide if he wants to review them with the passage author. My point w/ mentioning her was that you make it sound like Altius is a bunch of *****s, when people with credentials much greater than your own have worked on our materials extensively and continue scouring them every day for possible improvements, sometimes disagreeing vociferously w/ one another. I'll bet if you wrote a passage for us and we put it through our normal editing process, some other PhDs would have a few corrections for you, not to mention the hits you might take on the "internet-forum-osphere." Although, I do know a few of our PhD editors and I don't think anyone would call you an idiot, or call your work **it, no matter how glaring your mistake.

Have a good night.

 

[aldol16]

Let's put this to bed. I never said you were a "prick". I did call you arrogant, and probably should not have. Please accept my apology. Putting yourself in my shoes, it's not the most humble or friendly thing in the world either to lead out with calling the Altius passage ****ty, nor generalizing that such ****ty passages are "Typical," nor calling me/Altius the "idiot test prep company." If you find what you think is a problem with any Altius exam in the future, just PM me and I'll pass it up the line for further consideration.

I understand your defensiveness. But let me also tell you where I'm coming from. In these forums, I've seen many MCAT-style questions from companies - not only yours but also from TPR, TBR, NS, etc. - that are misleading at best and plainly scientifically invalid at worst. As an educator, in real life and on here, I consider it my responsibility to correct the scientific errors because students can't go into the MCAT believing one thing when it's simply untrue. I appreciate the difficulty associated with writing these questions, as there are layers upon layers of depth that you have to consider. But if a student posts on here about an Altius question that is scientifically wrong or provides a scientifically incorrect explanation (as of the p test in this case), then I have a duty to correct that not only for them but for others who might have the same question as well. I consider this the quickest response to an inaccuracy in any test prep companies' material and welcome any scientific challenge to my opinion. If I'm wrong, I'll admit it. In the future, I can also PM you and bring the error to your attention, but I cannot neglect to answer the question put forward by the student. That would be irresponsible on my part. I've said before and I'll say again that I believe any test prep company prepares a student adequately contentwise for the exam. That statement has not changed. I do believe, however, that every company, including Altius, can write better practice questions and when an error is pointed out, can work constructively towards improving that passage/question for future test-takers. I believe that's our duty as educators.

I don't know the author who compiled that passage personally, but Altius does have every passage marked w/ the authors and PhD reviewers so our internal curriculum team can go back and clarify issues as they arise. I'll share your comments with my boss and he can decide if he wants to review them with the passage author. My point w/ mentioning her was that you make it sound like Altius is a bunch of *****s, when people with credentials much greater than your own have worked on our materials extensively and continue scouring them every day for possible improvements, sometimes disagreeing vociferously w/ one another. I'll bet if you wrote a passage for us and we put it through our normal editing process, some other PhDs would have a few corrections for you, not to mention the hits you might take on the "internet-forum-osphere." Although, I do know a few of our PhD editors and I don't think anyone would call you an idiot, or call your work **it, no matter how glaring your mistake.

I would place my money on you. It's hard to write a passage. I know that. You wouldn't believe how many times I've written organic chemistry passage-style questions for undergraduate course exams and had things pointed out to me by students that I didn't even consider. But that feedback made my questions better for future classes. But that doesn't mean that when an error is pointed out, a company should double down on it and invoke the credentials of the writer as evidence as to its scientific validity. That's not how science works. In fact, it's the very disagreements - the scientists pointing out one another's errors - that drives science. Schrodinger and Einstein are two of the most influential names in physics. But one of them has to be dead wrong. When I review a paper, it doesn't matter if the PI is a Nobel Prize-winner in the field. If there is a scientific mistake in the paper, I am bound by duty and ethics to point it out. Even the leaders in the field are not perfect and make mistakes. What makes them leaders is that they are willing to review those mistakes and correct them if necessary. That's why every journal has an errata section.

And it's not only your company and other test prep companies. I've found several mistakes in the AAMC materials as well - mistakes that other students have been intelligent enough to point out. These are from the makers of the real thing who are expected to have extensive safeguards against these sorts of mistakes!

I should not have used the word "idiot" and I apologize for that. But please understand my exasperation after seeing question after question on here not be scientifically correct or include enough information to actually answer the question. I understand that since I don't see all Altius questions and only see the subset that students have the most trouble with, it's not a representative sample of Altius questions. So I shouldn't generalize from that.