"One important caveat to the psychological work on the above topics is that it has for the most part been based on limited samples of the human population (white, middle-class, American, male, students).
This lack of representativeness means that the theories and research findings may not be generalisable to to other populations.
Or even generalisable to females – more than 50 per cent of the population.
The issue of small sample size would eventually be recognised as a problem in neuroscience – as we reported here, most fMRI studies were completely worthless, despite the dramatic and colourful pictures of "your brain on..." (whatever a Harvard undergrad's brain was on that day).
But we rarely saw that caveat at the height of the mania for neuroscience-inspired pop psych, when the media couldn't give us enough fMRI-derived pseudoscience. Even as our chatterati was heralding the behavioural woo as "new discoveries in psychology", they either didn't know that the "brain science" on which it is founded was being conducted on tiny (typically fewer than 20 participants) studies."
In his book "Thinking, Fast and Slow", Chapter 10, Daniel Kahneman rightly warns of using statistics based on small samples. Sadly he subsequently bases much o fhis book on experiments conducted on few and unrepresentative samples. One of the experiments about how many people would rush to help someone in distress was conducted on only 15 people (3 times, each on 5 people. Nonetheless he claims that 27% of people would rush to help someone. (The experiment as described is also flawed, but that's another story.)
Which is unfortunate because Kahneman does have some useful insights about people and how we make decisions, as, I expect the 'spooks' at GCHQ probably do.