14) Boy, you sure have been cranky lately. What is it with you and psychologists?
14) Boy, you sure have been cranky lately. What is it with you and psychologists?
To paraphrase Garrett Morriss, "psychology has been very, very good to me." So, I'm appreciative of how much I've learned and gained from psychologists and psychology.
However, there are some things about psychologists that drive me crazy.
Foremost is their fixation on their theories or rather the weak link between theory and their empirical work. Psychologists can write the most long-winded discussion of the most superficial set of analyses. More generously, the poverty of their empirical work contrasts with the richness of their theories. For example, developmental theories emphasize the dyadic nature of many relationships. However, most empirical work typically regresses a child outcome on a measure of parenting (perhaps lagged). This association is well established, and the next step is embracing the underlying simultaneity. Establishing causality is difficult (or maybe impossible), but other social sciences spend far more time tackling this problem.
A second example involves non-linearity. Many psychological theories emphasize the interaction between factors. For example, the theory of reasoned action emphasizes that behavioral intentions are a function of attitudes, norms and perceived control. Norms, however, matter only if individuals believe they have control. This type of relationship implies non-linearities above and beyond a simple interaction. When was the last time one saw a true non-linear model in an empirical paper in an empirical article in psychology?
All of this aggravates me because my papers have been getting rejected lately, largely for being "methods" papers, lacking substantive content. My own preference is to try to better measure these basic descriptive characteristics, and to spare the reader these long-winded expositions on theory barely supported (or not) by the data.
I also am a little aggravated by the norms for professional interaction in psychology. Psychologists as a group (especially the ones who have been clinicians) are very nice. However, when spending the taxpayers' money, a little directness and accountability is a good thing. My psychologist colleagues are often not forceful enough in demanding the best of each other. And they get really offended when I demand it of them.
Psychologists just don't love their data well enough or at least respect it. Some of them are just astounding data-dredgers. They really know the answer before they start their anlayses and continue until they find what they were looking for. It's about that simple. You can see it in their unwillingness to share their data. Generally, if you ask, they will indicate that "you don't know enough about my program to evaluate it". What the heck does that mean? That I'm not committed enough to finding an effect. It seems that way.
As I meet more and more psych methods folks, my experiences seem the norm, and more than a few of them have had to abandon projects because of highly dubious practices. And if you talk to the junior data analysts on these projects, you'll never meet a more cynical group of human beings.
And finally, many applied psychologists are just astoundingly ignorant of statistical methods in an enterprise that should be labelled quantitative social science. I guarantee you that 90% of psychologists could not invert this matrix:
a b
c d
Many would not even know what the question meant. I know multiple chair professors of psychology at major universities that couldn't pass an undergraduate econometrics course. And these are smart people--for example, one went to Stanford at age 16. Believe me, Stanford was not looking for me then (or subsequently).
And psychologists can tell me, with a straight face some of the silliest things.
Recently, psychologists have suggested to me that
1) All I need to do is to measure X before Y and we know X causes Y.
Sure, the rooster crows and that causes the sun to rise. Right.
2) One just told me, "we proved randomization worked in our study". Ah, yea, right--the benefit of randomization involves unobservables, and I guess this study doesn't have any. (I guess what the person doesn't understand is that you can prove randomization didn't work but you can never prove it did.)
to be continue
To paraphrase Garrett Morriss, "psychology has been very, very good to me." So, I'm appreciative of how much I've learned and gained from psychologists and psychology.
However, there are some things about psychologists that drive me crazy.
Foremost is their fixation on their theories or rather the weak link between theory and their empirical work. Psychologists can write the most long-winded discussion of the most superficial set of analyses. More generously, the poverty of their empirical work contrasts with the richness of their theories. For example, developmental theories emphasize the dyadic nature of many relationships. However, most empirical work typically regresses a child outcome on a measure of parenting (perhaps lagged). This association is well established, and the next step is embracing the underlying simultaneity. Establishing causality is difficult (or maybe impossible), but other social sciences spend far more time tackling this problem.
A second example involves non-linearity. Many psychological theories emphasize the interaction between factors. For example, the theory of reasoned action emphasizes that behavioral intentions are a function of attitudes, norms and perceived control. Norms, however, matter only if individuals believe they have control. This type of relationship implies non-linearities above and beyond a simple interaction. When was the last time one saw a true non-linear model in an empirical paper in an empirical article in psychology?
All of this aggravates me because my papers have been getting rejected lately, largely for being "methods" papers, lacking substantive content. My own preference is to try to better measure these basic descriptive characteristics, and to spare the reader these long-winded expositions on theory barely supported (or not) by the data.
I also am a little aggravated by the norms for professional interaction in psychology. Psychologists as a group (especially the ones who have been clinicians) are very nice. However, when spending the taxpayers' money, a little directness and accountability is a good thing. My psychologist colleagues are often not forceful enough in demanding the best of each other. And they get really offended when I demand it of them.
Psychologists just don't love their data well enough or at least respect it. Some of them are just astounding data-dredgers. They really know the answer before they start their anlayses and continue until they find what they were looking for. It's about that simple. You can see it in their unwillingness to share their data. Generally, if you ask, they will indicate that "you don't know enough about my program to evaluate it". What the heck does that mean? That I'm not committed enough to finding an effect. It seems that way.
As I meet more and more psych methods folks, my experiences seem the norm, and more than a few of them have had to abandon projects because of highly dubious practices. And if you talk to the junior data analysts on these projects, you'll never meet a more cynical group of human beings.
And finally, many applied psychologists are just astoundingly ignorant of statistical methods in an enterprise that should be labelled quantitative social science. I guarantee you that 90% of psychologists could not invert this matrix:
a b
c d
Many would not even know what the question meant. I know multiple chair professors of psychology at major universities that couldn't pass an undergraduate econometrics course. And these are smart people--for example, one went to Stanford at age 16. Believe me, Stanford was not looking for me then (or subsequently).
And psychologists can tell me, with a straight face some of the silliest things.
Recently, psychologists have suggested to me that
1) All I need to do is to measure X before Y and we know X causes Y.
Sure, the rooster crows and that causes the sun to rise. Right.
2) One just told me, "we proved randomization worked in our study". Ah, yea, right--the benefit of randomization involves unobservables, and I guess this study doesn't have any. (I guess what the person doesn't understand is that you can prove randomization didn't work but you can never prove it did.)
to be continue
posted by Michael at
6:42 PM
|
0 Comments
