Cecily is a professor who is confident that people can know the outcomes of future random events (a so-called 'psi' effect*).
She is keen to run an experiment to show that people can predict outcomes of coin tosses at a rate better than chance.
She sets up a graduate student to do a study for her. The graduate student, Laz E Bum, is a slack research assistant uninterested in doing the work. So, he simply makes up the data.
He generates random data for 100 people guessing a thousand coin tosses each. The results from the random-number generator "show" that 50% of the guesses (of the fake respondents) are right, and 50% are wrong+.
The graduate student then tweaks a few of the results so that overall, on average, people correctly guess 501 out of 1000.
Do the final results of this study provide evidence of a psychokinetic effect?
It seems that there are two possible answers.
#1: No, because the data were fabricated (by Laz E Bum, the research assistant).
#2: Yes, because Cecily's (the professor and lead researcher) expectations led to the research outcomes supporting her hypothesis (anticipation?) that people could anticipate the outcomes of future events!
* The study of 'psi' or the power of mind over matter, of consciousness over the physical world, has been explored at length in psychology (often within a sub-field known as para-psychology). One of most famous labs studying psi is PEAR: Princeton Engineering Anomalies Research.
+ This result is in line with reported size of the psi effect: "'The effects that the volunteers accomplish are very small, but amazing. The operators are roughly altering one bit in 1,000,' explains Michael Ibison, a British mathematical physicist who has come to work for a year at PEAR after stints at Siemens, IBM, and Agfa. 'That means if you had a coin toss, psychokinesis could affect one of those coin tosses if you tossed a thousand times.'" (Van Bakel 1994, Wired)