Quick answer - It is Candidate #4.
Some back story on that conclusion...
Recently researchers at Old Dominion University published a study called 'How quickly do interviewers reach decisions? An examination of interviewers' decision-making time across applicants' in the Journal of Occupational and Organizational Psychology. They found that hiring manager decision-making takes closer to five minutes for the first interviewee, and reaches closer to eight minutes by the fourth applicant. After this, however, the time hiring managers take to reach a decision begins to decrease with each additional interview.
Here's a chart from the study:
From the researcher's conclusions on this data:
Interviewers tend to take longer to evaluate applicants near the beginning of their interview schedule and take less time to evaluate applicants near the end of their schedule. This may prevent applicants who appear later in the schedule from having a full opportunity to perform. Organizations may benefit from limiting the number of interviews an interviewer conducts in immediate succession to around four, which may decrease reliance on more automatic information processing strategies.
What conclusions can we draw from this data, and what changes might we need to consider to make sure we are not falling into the 'Candidate #4' trap?
Well, the first step is just being aware of this potential tendency. If you have to set up an interviewer or a hiring manager for a day-long set of candidate interviews, make sure you schedule some breaks such that they are not seeing a dozen people in a three-hour block. Chances are everyone after Candidate #4 are not getting a fair look, and we are wasting hiring manager time as well.
Next, if you are brining in a smaller set of short listed candidates for a second round of interviews, don't slate them in the same order with every interviewer they have to meet. Mix up the order across the interviewing team to try and reduce the effects of 'interview fatigue' adversely impacting any single candidate.
And last, keeping this data in mind should make us be more careful about tracking more data around interviewing and interviewers - how much time they spend per candidate, how much does the 'Candidate #4' efffect exist in the organization, and how can we use data on these processes to get better.
Data is our friend. Use wisely.