Professional and Academic Faux Pas

[DarwinAG]

There seems to be a plethora of unwritten rules in academia that is often difficult to navigate for new researchers.

I encountered one today, and I am still not sure if I understand the situation properly.

A professor was giving a job talk presentation and undergraduate researchers were encouraged to attend. In one study, she described that she collected saliva samples from 50 males and 50 females to determine if there are any changes in diurnal cortisol patterns as a function of perceived discrimination. Her data did not yield significant results. Another professor pointed out that the menstrual cycle in women may affect cortisol levels and that no data was provided regarding the last time the females had their period, which is a potential confound. I thought it was a valid point and proceeded with the questionthat "If you do not have the data from their last period, and it can affect cortisol levels, why not drop the women from analysis since data may add noise?" My question had mixed results. Some laughed (although I now think they were laughing at the idiocy of my question) and some professors had an adverse reaction that I would ask something like that. I suggested this to address the methodological flaw in the study, but it appeared that I brought up "other" issues in terms of women in research and women participating in research. I was completely unaware that my offhand comment on how to address a methodological flaw in the study would be perceived as antagonistic.

I am still not sure if I understand what happened, and I'd love to hear people's insight. I obviously do not want to offend anyone with my questions, so I will be more careful next time.

Edited December 13, 2012 by DarwinAG

[DarwinAG]

I'd also like to point out that I prefaced my comment with "I am not sure if you will still have enough power to run your analysis after dropping half the sample."

[Tolman's Rat]

Do you believe that the reaction would have been different had there been a male-specific confound that rendered their data invalid? To me, this seems like a very valid suggestion and not at all provocative. It is not as though you are prohibiting women from participating in a study, but rather recognizing the lack of adequate control, and adjusting accordingly. It's unfortunate that your colleagues replied in such an immature manner, and I have never encountered something like this. It is a design issue, not an ethical issue by any means.

[DarwinAG]

From what I can gather, the issue of polarizing "gender differences" came up. As I was explaining myself to one female peer, she immediately replied with a defensive "I have cortisol too." Additionally, she proceeded to explain to me how difficult it is for women because they are "perceived" differently their menstrual cycle and how this is sort of a "sensitive" topic to be brought up, even in the context of a scientific talk.

Edited December 13, 2012 by DarwinAG

[DarwinAG]

I don't think the reaction would have been the same if there was a male-specific confound. But that's just my personal speculation. It really was just a suggestion on the statistical analysis side. The rooms reaction was very unexpected. Since "cortisol" was used as an index of stress, I thought they were under the impression I was suggesting the "confound" was symptoms of "premenstrual syndrome", which is something that did not really occur to me. I was really just making the comment from my cursory knowledge of endocronology.

[firstsight]

Well, dropping half of the entire sample doesn't seem like too wise of a choice, especially as you are limiting yourself to a gender. Don't get me started on how women are consistently underrepresented in medical studies and how men are the norm. That said, I probably would have laughed, just because-- well, from my opinion, a confound like that is an issue that you deal with on replication; say that it is a limitation and recommend that follow-up be conducted while controlling for where women are in their menstrual cycle. People forget that men also have monthly hormonal cycles (likely because they don't bleed for a week) - so saying "only one gender has wide variances in hormones over the course of X amount of time" is likely a falsehood. I'm not an expert in all that, so don't quote me or anything - but just my first glance at it.

Now, then. There IS a large issue of gender differences in psychology, and since more and more women are populating psychology, it is probably an element of politics that you want to be aware of. There are many feminists about - and as a sidebar, some feminists (like myself) would eat that little girl that you mention for breakfast. Sensitive topic my arse - it's a physiological function, not some sort of mystic female-only cult knowledge with appropriate restrictions. People like that will likely never make it in a predatory academic environment - or maybe I've just been around evolutionary psychologists too long. Anyhow, /tangent. That said, I highly recommend being aware of the fact that some feminists are sensitive to any inferred gender discrimination within psychology - or more specifically, a man seeming to suggest that somehow women are less than appropriate human subjects due to whatever reasons. It's a difficult subject to really give any depth to in a short period of time, other than to recommend to be cognizant that there may be people sensitive to (gender/socioeconomic class/race/sexual orientation) issues - and most of the time, those will be in a minority or classically oppressed category, responding negatively to someone in the majority. I'm rambling. But I think y'get my point.

[kaister]

When I read your question in the context of this discussion it did not at all strike me as insensitive or naive. It is a valid question. If for some reason there were reasoning that would explain how your question was "wrong", then I would just assume faculty members would do their best to explain it. More specifically, address the question in a nonjudgmental, teaching manner. This is not an ethical issue, nor did you bring any of the sort up. The fact that "whomever" drew the discussion in that direction is extremely immature. I'm sorry that happened to you. I don't think you did anything wrong. Even if it were an ethical issue, it shouldn't be met in such defensive tones, especially in an academic setting. I've had experiences like that and it's incredibly frustrating. No one learns in that sort of environment.

[PsychGirl1]

Weird reaction from your female peer, sort of embarrassing for my gender! Tell her to stop being offended every time someone says the word "female" and to put on her science hat :-)

First off, I doubt if you write up a manuscript and say you excluded half your data based on gender because of a potential confound you could have controlled for, and that it wasn't significant until you excluded half of the people, that they would be okay with that. But I can see running the enalysis for exploratory purposes, and/or looking at differences by gender, and discussing the issue within your paper, especially in the scope of future research. That totally makes sense.

There is also the issue of what firstsight said- you do run into hormonal differences within males as well (or at least, theories of hormonal differences- I do not know if it has been proven/disproven) that are not as easily accounted for by just asking about their last menstural cycle. This may be one of the reasons why you see a good number of pilot neuroimaging studies only use females for their small sample size experiments, in addition to asking about menstrual cycles. I don't know much about cortisol differences by gender throughout the month, but you probably can't remove half of your sample due to hormonal changes without showing that the other half doesn't have hormonal changes.

To be honest, I would have probably laughed because I thought you were joking if you mentioned that at a talk. "Just exlude half of your sample size basd on gender- whatttt!" But it wouldn't have been a mean laugh, and you do raise some valid points, that maybe some exploratory analyses could be done that could clarify some of the directions future research should take.

Edited December 13, 2012 by PsychGirl1

[DarwinAG]

Haha It had a hint of a joke so I would have expected that sort of response. I think firstsight hits the nail on this one. I need to ne more cognizant of certain sensitive issues.

[lewin]

In an unrelated but very important point*, the questioner who preceded you misused the word 'confound', which should only be used to describe variables that co-vary with both the independent and dependent variables. It's possible that "date since last menstruation" might have introduced random error variance into the cortisol data (as you said correctly, "added noise") but that isn't a confound unless the questioner is suggesting that "time since menstruation" is also related to perceived discrimination in some complicated way that suppressed the relationship between variables.

...and if the questioner was suggesting that women experience more discrimination depending on their menstrual cycle, the audience probably should have had a problem with that

* by "important" I mean "pedantic".

[lewin]

Sorry for double post but I wanted to keep my facetious and serious points separate.

The ethical point that some audience members might have been making is that some groups have been inappropriately excluded from research and so they don't receive its benefits; a notable example is women and heart disease. Another critique is that in the past researchers have used mostly men as participants, yet claimed they were learning things about humans in general. This plays into the bias that 'white men' were seen as the way humans work and other groups represent deviations from them.

But that's no reason for profs to jump down your throat about it, an undergrad who was trying to be constructive. Maybe they were getting weird about the word "exclude" ? In the future, you could say something like, "If what so-and-so saying is true, then the results should be moderated by gender. Is there an effect just within men?"

[DarwinAG]

In an unrelated but very important point*, the questioner who preceded you misused the word 'confound', which should only be used to describe variables that co-vary with both the independent and dependent variables. It's possible that "date since last menstruation" might have introduced random error variance into the cortisol data (as you said correctly, "added noise") but that isn't a confound unless the questioner is suggesting that "time since menstruation" is also related to perceived discrimination in some complicated way that suppressed the relationship between variables.

...and if the questioner was suggesting that women experience more discrimination depending on their menstrual cycle, the audience probably should have had a problem with that

* by "important" I mean "pedantic".

haha thanks Lewin.

[DarwinAG]

But that's no reason for profs to jump down your throat about it, an undergrad who was trying to be constructive. Maybe they were getting weird about the word "exclude" ? In the future, you could say something like, "If what so-and-so saying is true, then the results should be moderated by gender. Is there an effect just within men?"

I think they got weird with the word "exclude" and "noise". Calling it "noise" had evaluative connotations that I was not really cognizant about when I made my comment. That degenerated into me saying that "menstrual cycle is noise" which is a complete misrepresentation of what I said.

[Angua]

In one study, she described that she collected saliva samples from 50 males and 50 females to determine if there are any changes in diurnal cortisol patterns as a function of perceived discrimination.

 

I think some of the reaction may have stemmed from your suggestion in the specific context of this study.  Maybe I'm misunderstanding, but was the purpose of the study to examine biological correlates of social discrimination?  It sounds like that was the purpose, and it also sounds like they specifically balanced men versus women in the study, presumably to allow for gender comparisons?  If so, suggesting that they drop the women from the study may be problematic not only for the gender implications that have already been discussed, but also because it would undermine the purpose of the study.  If they wanted to compare men and women, dropping the women because would be highly counterproductive.  Even if no gender comparison was planned or intended, however, if the subject was perceptions of discrimination, leaving women out of the study altogether poses some rather fundamental validity questions.

[BeingThere]

It is too bad your comment was not addressed with a serious and scholarly reply, DarwinAG.  From reading the comments in this thread, it is obvious that would have been possible.  I am but a newbie in the world of psych academe, but I would not call your comment a faux pas.  I'm not even sure if I would consider it ill-worded.  As a woman and as a researcher, I find nothing offensive in your statement. 

 

On a methodological note -- and what could be a methodological faux pas? -- possibly dropping the women completely would not be the best course.  But couldn't one run a separate analysis using the men only, and report both findings? 

 

And further, would any variation in cortisol levels in women due to their menstrual cycle be greater than variation caused by other extraneous variables in any sub-group of the sample (the list of other extraneous variables is probably endless -- a biggie might be age, another might be race)?  Has that been studied?  In other words, is that really an extraneous variable worth taking into account?  One way to get an inkling would be to do a separate analysis on just the men. 

 

Yeah, as a student I would have been much more interested in sussing out the issues with the methodology than getting hung up on whether someone said something that I could take the wrong way if I was so inclined.

[DarwinAG]

I think some of the reaction may have stemmed from your suggestion in the specific context of this study.  Maybe I'm misunderstanding, but was the purpose of the study to examine biological correlates of social discrimination?  It sounds like that was the purpose, and it also sounds like they specifically balanced men versus women in the study, presumably to allow for gender comparisons?  If so, suggesting that they drop the women from the study may be problematic not only for the gender implications that have already been discussed, but also because it would undermine the purpose of the study.  If they wanted to compare men and women, dropping the women because would be highly counterproductive.  Even if no gender comparison was planned or intended, however, if the subject was perceptions of discrimination, leaving women out of the study altogether poses some rather fundamental validity questions.

 

The purpose of the study was to show the effects of social discrimination on physiology, not to examine the biological correlates of social discrimation. Those are two very different things.

 

They did not specifically balanced men and women. That was an accidental feature of the the sampling method. They did not think there would be a gender difference that is why men and women were lumped together in the analysis. If they wanted to compare man and women, they could have easily done so. My suggestion was to drop the women from the overall cortison analysis because they did not control for onset of menstrual cycle which presumably has an effect on cortisol levels. If they controlled for it methodologically, then by all means conduct the analysis with both men and women. Of course, my suggestion assumed that the menstrual cycle does indeed significantly impact cortisol level. And to reiterate, my suggestion was based on a lack of methodological control on cortisol levels. I am not suggesting to not include women in studies altogether.

[Angua]

The purpose of the study was to show the effects of social discrimination on physiology, not to examine the biological correlates of social discrimation. Those are two very different things.

 

Perhaps, but it doesn't change my point: people may have reacted this way (I've actually been thinking about why people would laugh) partly because it seems that removing women from a study whose focus was social discrimination would cause problems.  Clearly this is not my area of expertise, and so I approach these questions very differently.  In the research I deal with, external social validity is a primary concern - I could never drop all of one gender from a sample. I know that there is often more flexibility in neuro/bio work, and so maybe it would be fine in this case.  There's no need to defend your suggestion to me - you asked for thoughts about why people reacted that way, and I was just trying to give you another perspective.

[DarwinAG]

Perhaps, but it doesn't change my point: people may have reacted this way (I've actually been thinking about why people would laugh) partly because it seems that removing women from a study whose focus was social discrimination would cause problems.  Clearly this is not my area of expertise, and so I approach these questions very differently.  In the research I deal with, external social validity is a primary concern - I could never drop all of one gender from a sample. I know that there is often more flexibility in neuro/bio work, and so maybe it would be fine in this case.  There's no need to defend your suggestion to me - you asked for thoughts about why people reacted that way, and I was just trying to give you another perspective.

 

Two things. One, feedback and another perspective is always welcome and greatly appreciated, but I disagree with "there's no need to defend your suggestion". As scientist we are encouraged to be active and critical thinkers, and accepting another's position without even actively engaging with it I think goes contrary to that spirit. I think feedback should not just be passively accepted. I have thought about what you said and replied accordingly, and I would expect the same thing in turn. Agreeing with it would still entail the same critical thinking process. I think this is the best way to move a discussion and better understand an issue. With that said, I especially like hearing yours and other people's perspective as I think it enhances my own, but eliciting feedback doesnt make those feedback immune to a critical evaluation.

 

Two, I agree with you that women should be included in the study on social discrimination (although the focus of this study was  ethnic discrimination, which is a minor point I think), but that the researchers should have controlled for things that would significantly impact cortisol levels. It is not uncommon in physhophysiology research to control for variables that would signifincantly impact the physiological DV in question. For example, in some EEG studiess, the sample is often restricted to right-handed people because of relevant brain differences in dominant handedness. Additional examples are controlling for antidepression medication, caffeine intake, and sleep cycles. All of these can impact relevant physiological DVs. I go back to my point that women should be included but controlling for relevant variables that could impact cortisol levels, in this case onset of the menstrual cycle. It wasn't controlled for in this particular study.

 

I actually think she could have designed the study with more stringent controls and still retain important external social validity.

[lypiphera]

So, I don't think your question was offensive, per se, but the issue may have been in the wording.  Rather than suggest excluding women since we don't know if there is an issue with menstrual cycles, I might have suggested that you examine the genders separately, see if there is any significant difference, and then put the whole sample back together if not.  I guess its the wider concept of excluding women because of the possibility that a period could affect results.  Unless there's good evidence from your sample that the female data behaved differently from the male data in terms of your analysis, I would not exclude them offhand.  It may have been seen as you saying that its okay to exclude women from research because their periods interfere.  I don't think that is what you are saying at all, it sounds like just a statistical suggestion, but I could see how someone would misinterpret it.

[watson]

My guess is they were laughing at the suggestion of dropping half the sample after you ran the study just because you didn't get the results you want.  It's not the best way to do science and actually a hot topic right now about research analysis ethics.  Re-analyzing without the women to see if it warranted running a new study would be fine, but you'd have to then pretty much dump the first study to stay really above board.  Also, as bad as this is, people do tend to get annoyed when the undergrads ask questions/make comments (especially at job talks) and this can lead to less than polite reactions to their questions.

[watson]

I should also point out--in my program, we are constantly teasing each other and being lighthearted during talks, so there may not have been ill-will there necessarily.  It's usually a group of people who are all friends, cracking jokes at each others' expense with no offense taken--to an outsider getting laughed at will probably seem like people are being really mean, but it may actually be something that they didn't intend to be taken to heart.

[Tuza]

I read most of this thread, but just wanted to point out that removing women from a study will greatly reduce the external validity of the project.  Say they do find significant findings, they'll only be valid for men.  The presenter should have collected data on menstrual cycle as well as acute and chronic life stressor (as these also impact cortisol levels) and then run a multilevel model controlling for those parameters.  Your question seems a bit naive to me, but I always think it's better to ask questions rather than keep them in...believe me, I've heard some pretty crazy questions come out of some professors mouths where you just have to keep a straight face and not laugh at them.

[DarwinAG]

I read most of this thread, but just wanted to point out that removing women from a study will greatly reduce the external validity of the project.  Say they do find significant findings, they'll only be valid for men.  The presenter should have collected data on menstrual cycle as well as acute and chronic life stressor (as these also impact cortisol levels) and then run a multilevel model controlling for those parameters.  Your question seems a bit naive to me, but I always think it's better to ask questions rather than keep them in...believe me, I've heard some pretty crazy questions come out of some professors mouths where you just have to keep a straight face and not laugh at them.

I think you may have glossed over some critical points. It has been a long thread.

-I did not suggest to drop women from the study altogether (prior to data collection). There is a difference between suggesting to not collect data from women at all to drop potentially noisy data from analysis after is hast been collected.

-My suggestion to remove women from analysis was after "noisy" data has been collected already. Let's say you found statistical findings with potentially noisy data, then you really haven't illuminated anything.

-And the study was on acute and chronic stress from perceived ethnic stigmatization. Data was also collected on acute and chronic life stressors.

-I am of the philosophy that if you have a genuine question. You should ask. You don't want to compound ignorance with silence. I think an environment that facilitates this is better for learning than one where you have to by vigilant about not appearing naive or stupid.