WEBVTT 00:00:00.000 --> 00:00:03.700 Welcome to The Debate. Today, we are walking 00:00:03.700 --> 00:00:06.580 into a room that, well, almost everyone listening 00:00:06.580 --> 00:00:09.519 has entered at some point. It's a room of high 00:00:09.519 --> 00:00:13.539 stakes, sweaty palms, and carefully rehearsed 00:00:13.539 --> 00:00:16.039 answers. I am speaking, of course, about the 00:00:16.039 --> 00:00:19.000 job interview. It's the ubiquitous gatekeeper 00:00:19.000 --> 00:00:22.140 of employment, a ritual practiced globally from 00:00:22.140 --> 00:00:25.059 a casual coffee shop chat to a grueling multi 00:00:25.059 --> 00:00:28.100 -stage corporate interrogation. Ritual is the 00:00:28.100 --> 00:00:30.750 perfect word for it. Yeah, it's something we 00:00:30.750 --> 00:00:33.109 do because we've always done it, not, you know, 00:00:33.130 --> 00:00:35.670 necessarily because it works. We just assume 00:00:35.670 --> 00:00:37.789 that sitting two people in a room for an hour 00:00:37.789 --> 00:00:40.810 somehow reveals who is best for a job. But I'm 00:00:40.810 --> 00:00:42.789 here to argue that what we're actually watching 00:00:42.789 --> 00:00:46.130 is not a scientific assessment. It's a piece 00:00:46.130 --> 00:00:49.189 of performative theater. And that is the central 00:00:49.189 --> 00:00:52.560 tension we're unpacking today. Is the job interview 00:00:52.560 --> 00:00:55.579 a reliable, valid tool for identifying the best 00:00:55.579 --> 00:00:58.939 candidate, a predictor of future success? Or 00:00:58.939 --> 00:01:02.119 is it, as you suggest, a flawed social game that 00:01:02.119 --> 00:01:04.719 rewards the best actor rather than the best worker? 00:01:05.000 --> 00:01:07.719 I think, you know, I'm leaning heavily toward 00:01:07.719 --> 00:01:10.599 social game. I mean, I'll argue that the interview 00:01:10.599 --> 00:01:13.219 process is fundamentally compromised by human 00:01:13.219 --> 00:01:16.379 bias, the candidate's ability to, you know, fake 00:01:16.379 --> 00:01:19.530 their way through. And a system that, quite disturbingly, 00:01:19.549 --> 00:01:22.450 inadvertently rewards some pretty dark personality 00:01:22.450 --> 00:01:25.769 traits. I see the flaws, certainly, but I'm going 00:01:25.769 --> 00:01:27.969 to take the side of the defense here. I'll argue 00:01:27.969 --> 00:01:30.849 that when we stop just chatting and start using 00:01:30.849 --> 00:01:34.129 a properly structured process, one grounded in 00:01:34.129 --> 00:01:36.609 industrial psychology, the interview becomes 00:01:36.609 --> 00:01:38.790 one of the most accurate predictors of employee 00:01:38.790 --> 00:01:41.810 success we have. It's not the tool that's broken, 00:01:41.989 --> 00:01:45.349 it's how most people are using it. Ah, the classic 00:01:45.349 --> 00:01:48.819 defense. It works in the lab. But let's see if 00:01:48.819 --> 00:01:51.239 it holds up in the real world. Why don't you 00:01:51.239 --> 00:01:53.780 lay out the case for the defense of the interview? 00:01:54.219 --> 00:01:57.819 Gladly. My position is rooted in the hard data 00:01:57.819 --> 00:02:01.000 of predictive validity. You are absolutely correct 00:02:01.000 --> 00:02:03.439 that unstructured chats where a manager just 00:02:03.439 --> 00:02:07.519 asks, so tell me about yourself, are, well, statistically 00:02:07.519 --> 00:02:11.319 useless. However, the structured interview is 00:02:11.319 --> 00:02:14.930 a scientifically validated instrument. When we 00:02:14.930 --> 00:02:17.750 look at meta -analytic results, these are studies 00:02:17.750 --> 00:02:19.909 of studies, we see that structured interviews 00:02:19.909 --> 00:02:22.629 have validity coefficients ranging from about 00:02:22.629 --> 00:02:27.990 0 .20 to 0 .57. Okay, hold on. For the listeners 00:02:27.990 --> 00:02:30.909 who aren't statisticians, can we translate that? 00:02:31.050 --> 00:02:34.729 What does a 0 .57 actually mean in this context? 00:02:35.110 --> 00:02:37.629 Right, good question. It's a correlation between 00:02:37.629 --> 00:02:40.090 the interview score and actual job performance 00:02:40.090 --> 00:02:44.289 later on. And to put 0 .57 in perspective, that 00:02:44.289 --> 00:02:46.990 is comparable to cognitive ability tests, IQ 00:02:46.990 --> 00:02:49.629 tests, which are traditionally seen as the gold 00:02:49.629 --> 00:02:52.550 standard in hiring. It's significantly higher 00:02:52.550 --> 00:02:54.969 than, say, reference checks or just years of 00:02:54.969 --> 00:02:57.550 experience. But the key is what we are measuring. 00:02:57.770 --> 00:03:00.590 We aren't asking, what kind of animal would you 00:03:00.590 --> 00:03:03.710 be? We're assessing person job fit. We are measuring 00:03:03.710 --> 00:03:08.750 specific KSAOs. KSAOs? Knowledge, skills, abilities, 00:03:08.909 --> 00:03:11.830 and other attributes. We break the job down into 00:03:11.830 --> 00:03:14.569 these core elements. We aren't guessing, we're 00:03:14.569 --> 00:03:17.409 using critical incidents. We look at past employees 00:03:17.409 --> 00:03:19.949 in that role, identify specific behaviors that 00:03:19.949 --> 00:03:21.969 led to success or failure, and then we build 00:03:21.969 --> 00:03:24.550 questions around that. It turns a gut feeling 00:03:24.550 --> 00:03:27.090 into a data point. That all sounds incredibly 00:03:27.090 --> 00:03:29.849 tidy, but I look at the reality of the process 00:03:29.849 --> 00:03:33.389 and I see a minefield of measurement error. The 00:03:33.389 --> 00:03:36.270 vast, vast majority of interviews are not structured 00:03:36.270 --> 00:03:40.199 like that. Managers rely on intuition. And intuition 00:03:40.199 --> 00:03:44.120 is often just a fancy word for, I like this person 00:03:44.120 --> 00:03:47.680 because they remind me of myself. I will concede 00:03:47.680 --> 00:03:50.259 that the unstructured interview is the Wild West. 00:03:50.539 --> 00:03:53.539 It's worse than the Wild West. It's a hallucination 00:03:53.539 --> 00:03:56.719 of competence. Research shows that interviewers 00:03:56.719 --> 00:03:59.860 suffer from this massive overconfidence. They 00:03:59.860 --> 00:04:02.240 believe they're human lie detectors. They think 00:04:02.240 --> 00:04:04.979 they can judge character in 45 minutes. But the 00:04:04.979 --> 00:04:08.039 data shows they fail repeatedly to identify optimal 00:04:08.039 --> 00:04:11.020 candidates. What the interview primarily measures 00:04:11.020 --> 00:04:14.539 isn't job fit. It measures interviewee performance. 00:04:15.120 --> 00:04:17.759 We're testing social effectiveness skills, things 00:04:17.759 --> 00:04:19.779 like impression management, self -monitoring, 00:04:19.779 --> 00:04:22.399 relational control. We are hiring the person 00:04:22.399 --> 00:04:24.980 who controls the flow of conversation, not the 00:04:24.980 --> 00:04:26.800 person who could actually code the software or 00:04:26.800 --> 00:04:29.660 manage the P &L. But isn't social skill relevant? 00:04:29.899 --> 00:04:31.959 I mean, if you can't communicate in an interview, 00:04:32.100 --> 00:04:34.160 how can you communicate with a client? For a 00:04:34.160 --> 00:04:38.720 sales role? Sure, maybe. For a data analyst, 00:04:39.019 --> 00:04:42.160 probably not. But the failure of measurement 00:04:42.160 --> 00:04:45.740 goes even deeper. We have to talk about the self 00:04:45.740 --> 00:04:49.220 -fulfilling prophecy effect. An interviewer looks 00:04:49.220 --> 00:04:51.480 at a resume or a LinkedIn profile before the 00:04:51.480 --> 00:04:54.399 candidate even walks in. They form a snapshot 00:04:54.399 --> 00:04:57.439 judgment. If that impression is positive, they 00:04:57.439 --> 00:04:59.920 subconsciously treat the applicant with warmth. 00:04:59.980 --> 00:05:03.100 They smile, they nod, you know. That makes the 00:05:03.100 --> 00:05:06.060 applicant relax and perform better. If the impression 00:05:06.060 --> 00:05:09.000 is negative, the interviewer is cold, the applicant 00:05:09.000 --> 00:05:12.019 gets anxious, and they bomb. The interview doesn't 00:05:12.019 --> 00:05:14.639 measure the candidate, it validates the interviewer's 00:05:14.639 --> 00:05:17.500 pre -existing belief. You are touching on the 00:05:17.500 --> 00:05:20.980 heart of the debate here. The illusion of structure 00:05:20.980 --> 00:05:24.180 versus the reality of bias. Let's dig deeper 00:05:24.180 --> 00:05:27.379 into this. You claim that bias prevents accurate 00:05:27.379 --> 00:05:30.279 measurement. I'm not convinced by that. because 00:05:30.279 --> 00:05:32.579 the entire field of industrial organizational 00:05:32.579 --> 00:05:36.199 psychology has spent decades developing tools 00:05:36.199 --> 00:05:39.420 specifically to mitigate that bias through structure. 00:05:39.639 --> 00:05:42.360 You keep using that word structure, but what 00:05:42.360 --> 00:05:45.439 does it actually look like? Because in my experience, 00:05:45.660 --> 00:05:48.680 structure is often just a checklist that a hiring 00:05:48.680 --> 00:05:51.399 manager completely ignores the moment they feel 00:05:51.399 --> 00:05:54.379 a... spark with a candidate. Elements of structure 00:05:54.379 --> 00:05:57.160 identified in the research. It means asking the 00:05:57.160 --> 00:06:00.420 exact same questions of all applicants in the 00:06:00.420 --> 00:06:03.480 same order. It means no prompting. You don't 00:06:03.480 --> 00:06:05.819 help the candidate who is struggling. And crucially, 00:06:05.920 --> 00:06:08.740 it involves evaluation structure. You don't just 00:06:08.740 --> 00:06:11.800 say, I liked her at the end. You rate each answer 00:06:11.800 --> 00:06:15.139 individually using anchored rating scales, specifically 00:06:15.139 --> 00:06:18.850 bars. Bars? Behaviorally anchored rating scales. 00:06:19.029 --> 00:06:21.589 So instead of a 1 to 5 scale based on a feeling, 00:06:21.810 --> 00:06:24.930 a 5 is defined by a specific behavior. For example, 00:06:25.129 --> 00:06:27.490 candidate described a conflict resolution strategy 00:06:27.490 --> 00:06:30.149 involving active listening and compromise. A 00:06:30.149 --> 00:06:32.509 1 is candidate suggested avoiding the conflict. 00:06:32.769 --> 00:06:35.370 It forces the interviewer to grade the evidence, 00:06:35.569 --> 00:06:39.899 not the vibe. That is the ideal, I agree. But 00:06:39.899 --> 00:06:42.680 even with bars and panels, you're still introducing 00:06:42.680 --> 00:06:45.639 human raters, and humans are just plagued by 00:06:45.639 --> 00:06:48.439 job -irrelevant interviewer biases. Let's just 00:06:48.439 --> 00:06:51.399 talk about the attractiveness bias. It is well 00:06:51.399 --> 00:06:53.639 documented that physically attractive people 00:06:53.639 --> 00:06:56.600 are judged more favorably on job -related factors, 00:06:56.819 --> 00:06:59.240 even when their looks have absolutely nothing 00:06:59.240 --> 00:07:01.879 to do with the job. We just assume good -looking 00:07:01.879 --> 00:07:04.319 people are smarter, healthier, and more competent. 00:07:05.000 --> 00:07:07.519 Can a rating scale really turn off that biological 00:07:07.519 --> 00:07:10.300 switch? It doesn't turn it off, but it dilutes 00:07:10.300 --> 00:07:12.959 it. Structured interviews with multiple raters, 00:07:12.959 --> 00:07:15.480 panel interviews, have been shown to reach inter 00:07:15.480 --> 00:07:18.199 -rater reliability levels of 0 .75 or above. 00:07:18.399 --> 00:07:21.139 When you have three people scoring based on strict 00:07:21.139 --> 00:07:24.120 criteria, the individual bias of one person tends 00:07:24.120 --> 00:07:26.500 to, you know, wash out. I'm sorry, but I just 00:07:26.500 --> 00:07:28.839 don't buy that structure eliminates these deep 00:07:28.839 --> 00:07:31.660 -seated signals. Consider vocal attractiveness. 00:07:32.569 --> 00:07:34.949 We judge people based on pitch, speech rate, 00:07:35.009 --> 00:07:38.990 and loudness. A certain mix of speech rate, loudness, 00:07:38.990 --> 00:07:41.829 and pitch predicts ratings. If someone has a 00:07:41.829 --> 00:07:45.430 good voice, deep, resonant, confident, we assume 00:07:45.430 --> 00:07:48.110 they're a leader. If they speak slowly or have 00:07:48.110 --> 00:07:51.029 a vocal fry, we dock them points. We're measuring 00:07:51.029 --> 00:07:54.370 acoustics, not aptitude. So you're arguing that 00:07:54.370 --> 00:07:57.389 we are biologically wired to hire the alpha? 00:07:57.769 --> 00:07:59.949 Essentially, yes. And it's not just biology, 00:08:00.250 --> 00:08:03.279 it's sociology. We have to discuss the Bertrand 00:08:03.279 --> 00:08:06.139 and Mullinathan study. This is the famous resume 00:08:06.139 --> 00:08:08.800 audit. They found that applicants with white 00:08:08.800 --> 00:08:11.360 -sounding names got 50 % more callbacks than 00:08:11.360 --> 00:08:13.240 those with African -American -sounding names 00:08:13.240 --> 00:08:17.360 with the exact same resume. This error starts 00:08:17.360 --> 00:08:20.120 before the interview even happens, but it absolutely 00:08:20.120 --> 00:08:23.019 bleeds into the room. If you walk in with a certain 00:08:23.019 --> 00:08:25.420 name or an accent, the interviewer's intuition 00:08:25.420 --> 00:08:28.470 is already primed against you. That study is 00:08:28.470 --> 00:08:30.589 a powerful indictment of the screening phase, 00:08:30.790 --> 00:08:33.570 absolutely. But strangely enough, it actually 00:08:33.570 --> 00:08:36.629 supports my argument for structure. The solution 00:08:36.629 --> 00:08:39.629 to name bias or accent bias isn't to get rid 00:08:39.629 --> 00:08:41.950 of interviews. It's to make them more robotic. 00:08:42.309 --> 00:08:44.929 Research shows that when interviewers are trained 00:08:44.929 --> 00:08:47.830 and forced to score specific competencies, the 00:08:47.830 --> 00:08:50.350 influence of race and gender diminishes. The 00:08:50.350 --> 00:08:52.870 problem isn't the tool. It's the misuse of the 00:08:52.870 --> 00:08:56.100 tool. But here is the irony. Interviewers hate 00:08:56.100 --> 00:08:58.879 structure. They resist it. They prefer unstructured 00:08:58.879 --> 00:09:01.879 formats because it allows them to feel autonomous. 00:09:02.100 --> 00:09:05.340 They want to trust their gut. So you have a tool 00:09:05.340 --> 00:09:08.139 that works, structure, but the users, the hiring 00:09:08.139 --> 00:09:10.919 managers, refuse to use it properly because it 00:09:10.919 --> 00:09:15.080 feels impersonal. They want the theater. That's 00:09:15.080 --> 00:09:17.500 a fair critique of human nature, but not of the 00:09:17.500 --> 00:09:20.580 science. If organizations want predictive power, 00:09:20.840 --> 00:09:23.629 they have to enforce the discipline. But I want 00:09:23.629 --> 00:09:26.009 to pivot to your earlier point about performative 00:09:26.009 --> 00:09:28.830 theater. You suggested that candidates are just 00:09:28.830 --> 00:09:32.269 acting. Exactly. And we need to talk about faking 00:09:32.269 --> 00:09:35.870 and the darker side of personality. Well, I acknowledge 00:09:35.870 --> 00:09:39.789 that impression management, or IM, occurs. But 00:09:39.789 --> 00:09:42.309 we need to distinguish between honest IM, which 00:09:42.309 --> 00:09:44.990 is self -promotion of true skills, and deceptive 00:09:44.990 --> 00:09:47.850 IM. If I tell you I'm a hard worker and I actually 00:09:47.850 --> 00:09:50.330 am, that's impression management, but it's also 00:09:50.330 --> 00:09:53.000 the truth. That's a valid distinction for a normal 00:09:53.000 --> 00:09:55.679 candidate. But have you considered that the interview 00:09:55.679 --> 00:09:58.279 format is practically designed to be exploited 00:09:58.279 --> 00:10:01.860 by the dark triad of personality? I'm talking 00:10:01.860 --> 00:10:05.419 about narcissism, machiavellianism, and psychopathy. 00:10:05.639 --> 00:10:08.200 You think they have a structural advantage in 00:10:08.200 --> 00:10:11.720 an interview? A massive one. Let's look at narcissism. 00:10:12.179 --> 00:10:14.620 Narcissists are skilled at glorying. They're 00:10:14.620 --> 00:10:17.460 comfortable taking credit, boasting, dominating 00:10:17.460 --> 00:10:20.440 the room. In a short, 30 -minute interaction, 00:10:20.799 --> 00:10:23.259 a narcissist looks like a high -potential leader. 00:10:23.440 --> 00:10:26.379 They create a fantastic first impression because 00:10:26.379 --> 00:10:28.460 they're often better dressed, more charming, 00:10:28.600 --> 00:10:31.500 and they exhibit these self -assured body movements. 00:10:31.639 --> 00:10:34.840 They mask their inability to perform long -term 00:10:34.840 --> 00:10:38.700 with short -term charisma. That is a risk, but 00:10:38.700 --> 00:10:41.139 that's why we use past behavioral questions. 00:10:41.360 --> 00:10:44.879 If I ask a narcissist, tell me about a time you 00:10:44.879 --> 00:10:48.549 failed. Their inability to accept fault usually 00:10:48.549 --> 00:10:52.230 reveals itself. The STAR technique, situation, 00:10:52.769 --> 00:10:57.029 task, action, result, forces candidates to be 00:10:57.029 --> 00:11:01.049 specific. It is much harder to fake a detailed 00:11:01.049 --> 00:11:03.909 past event than it is to fake a hypothetical 00:11:03.909 --> 00:11:06.970 situational question like, what would you do 00:11:06.970 --> 00:11:10.730 if? Harder, perhaps, but not impossible for a 00:11:10.730 --> 00:11:13.169 Machiavellian. I mean, these individuals are 00:11:13.169 --> 00:11:16.019 manipulative by nature. They can fabricate a 00:11:16.019 --> 00:11:19.539 star response that sounds perfect. And then there 00:11:19.539 --> 00:11:22.240 are the corporate psychopaths. They display this 00:11:22.240 --> 00:11:25.320 superficial charm and high energy. They're thrill 00:11:25.320 --> 00:11:27.779 -seeking. And in an interview, that thrill -seeking 00:11:27.779 --> 00:11:29.980 is often mistaken for entrepreneurial spirit 00:11:29.980 --> 00:11:33.220 or high motivation. The interviewer thinks, wow, 00:11:33.379 --> 00:11:35.580 this person is a go -getter, when in reality, 00:11:35.679 --> 00:11:38.059 they're just impulsive and irresponsible. That 00:11:38.059 --> 00:11:41.139 is a terrifying prospect, I admit. You're saying 00:11:41.139 --> 00:11:43.820 we are selecting for toxicity because it looks 00:11:43.820 --> 00:11:47.299 like confidence. Precisely. One study found that 00:11:47.299 --> 00:11:49.860 80 % of participants lied about job -related 00:11:49.860 --> 00:11:53.340 skills in an interview. 80%. If the majority 00:11:53.340 --> 00:11:55.740 of people are lying and the people listening 00:11:55.740 --> 00:11:57.740 are overconfident in their ability to detect 00:11:57.740 --> 00:12:00.019 those lies, we aren't conducting an assessment. 00:12:00.159 --> 00:12:02.399 We are attending a fiction writing workshop. 00:12:02.759 --> 00:12:05.399 But we aren't just looking for skills or facts. 00:12:05.799 --> 00:12:09.100 We are trying to find fit. And this leads us 00:12:09.100 --> 00:12:11.820 to another crucial aspect, anxiety and technology. 00:12:12.299 --> 00:12:15.340 We are looking for person -organization fit. 00:12:15.679 --> 00:12:18.379 Fit. That is one of the most dangerous words 00:12:18.379 --> 00:12:20.980 in hiring. Why do you say that? It's standard 00:12:20.980 --> 00:12:23.240 practice to ensure someone aligns with the company's 00:12:23.240 --> 00:12:26.480 values. Because fit is often just a euphemism 00:12:26.480 --> 00:12:29.879 for ingratiation. Research shows that the more 00:12:29.879 --> 00:12:32.840 an applicant sucks up, compliments the organization, 00:12:33.360 --> 00:12:36.100 agrees with the interviewer, the higher their 00:12:36.100 --> 00:12:39.559 fit score. It rewards sycophancy, not alignment. 00:12:39.820 --> 00:12:42.059 It rewards people who know how to play the game. 00:12:42.279 --> 00:12:45.240 I would frame it differently. Fit measures whether 00:12:45.240 --> 00:12:47.940 a candidate shares the core values of the organization. 00:12:48.559 --> 00:12:50.740 Structured interviews allow for a dual assessment, 00:12:51.100 --> 00:12:54.279 determining if they have the KSAOs and if they 00:12:54.279 --> 00:12:56.720 align with the culture. And with modern technology, 00:12:56.919 --> 00:12:59.779 we can expand the pool. Media richness theory 00:12:59.779 --> 00:13:02.480 suggests face -to -face is best for data gathering, 00:13:02.659 --> 00:13:05.139 but video and phone interviews allow us to screen 00:13:05.139 --> 00:13:07.539 more diverse candidates from further away. But 00:13:07.539 --> 00:13:09.740 technology is actually making the bias problem 00:13:09.740 --> 00:13:12.440 worse, specifically when it comes to interview 00:13:12.440 --> 00:13:15.740 anxiety. Anxiety is a natural reaction, though. 00:13:15.820 --> 00:13:18.899 Everyone gets nervous. But it is a barrier that 00:13:18.899 --> 00:13:22.379 penalizes specific groups unfairly. Research 00:13:22.379 --> 00:13:25.519 identifies low assertiveness and slow speech 00:13:25.519 --> 00:13:29.200 rate as key indicators of anxiety. If a candidate 00:13:29.200 --> 00:13:31.799 speaks slowly because they're thoughtful or maybe 00:13:31.799 --> 00:13:34.379 because of a cultural difference, they are perceived 00:13:34.379 --> 00:13:38.019 as anxious and less competent. Anxiety does not 00:13:38.019 --> 00:13:40.179 correlate with job performance for many roles. 00:13:40.379 --> 00:13:42.539 Think of a software engineer or an accountant. 00:13:42.740 --> 00:13:45.559 Yet it absolutely destroys interview performance. 00:13:46.019 --> 00:13:48.649 That's true. If the job doesn't require public 00:13:48.649 --> 00:13:50.450 speaking, we shouldn't be testing for it. And 00:13:50.450 --> 00:13:53.710 now add video interviews to the mix. Video is 00:13:53.710 --> 00:13:57.129 perceived as less personable and less fair by 00:13:57.129 --> 00:13:59.850 applicants. If there is a lag, you know, poor 00:13:59.850 --> 00:14:03.169 latency in the connection, it causes misinterpretation 00:14:03.169 --> 00:14:06.190 of nonverbal cues. A pause might look like hesitation 00:14:06.190 --> 00:14:09.929 or stupidity when it's actually just bad Wi -Fi. 00:14:10.330 --> 00:14:12.610 We are literally judging people on their internet 00:14:12.610 --> 00:14:15.289 connection speed, not their competence. I see 00:14:15.289 --> 00:14:18.190 the validity in the technical concerns. Media 00:14:18.190 --> 00:14:20.629 richness theory does support the idea that face 00:14:20.629 --> 00:14:23.049 -to -face is superior because it carries more 00:14:23.049 --> 00:14:25.830 data. Verbal and nonverbal cues are read in the 00:14:25.830 --> 00:14:28.870 moment. However, we can't ignore that technology 00:14:28.870 --> 00:14:31.610 enables us to recruit from a much wider geographic 00:14:31.610 --> 00:14:34.269 area, which can actually increase diversity. 00:14:34.669 --> 00:14:37.370 It increases the pull, sure, but it degrades 00:14:37.370 --> 00:14:40.240 the quality of the interaction. You mentioned 00:14:40.240 --> 00:14:43.320 person -organization fit earlier. In the video 00:14:43.320 --> 00:14:45.700 interview, it is much harder for a candidate 00:14:45.700 --> 00:14:48.580 to assess the company culture. It becomes a one 00:14:48.580 --> 00:14:50.620 -way interrogation rather than a two -way exchange. 00:14:50.980 --> 00:14:53.990 So to summarize where we stand. I'm not denying 00:14:53.990 --> 00:14:57.029 that the human element introduces noise, but 00:14:57.029 --> 00:14:59.330 my position remains that the structured interview 00:14:59.330 --> 00:15:02.070 is the most practical tool we have for assessing 00:15:02.070 --> 00:15:05.830 complex human traits. You can't measure conscientiousness 00:15:05.830 --> 00:15:08.690 or agreeableness or communication skills with 00:15:08.690 --> 00:15:11.769 a paper test alone. When we use behavioral questions 00:15:11.769 --> 00:15:15.269 based on critical incidents, on real job requirements, 00:15:15.570 --> 00:15:18.509 we achieve high predictive validity. We just 00:15:18.509 --> 00:15:20.950 need to ensure interviewers are trained to ignore 00:15:20.950 --> 00:15:23.750 the noise of appearance and vocal pitch. And 00:15:23.750 --> 00:15:26.149 I maintain that the interview is fundamentally 00:15:26.149 --> 00:15:30.129 a social game, where the best actors, not necessarily 00:15:30.129 --> 00:15:33.610 the best workers, win. The process is plagued 00:15:33.610 --> 00:15:37.029 by construct bias and method bias, particularly 00:15:37.029 --> 00:15:40.629 across cultures. When we rely on intuition and 00:15:40.629 --> 00:15:44.129 face -to -face charm, we open the door to narcissists 00:15:44.129 --> 00:15:46.789 and corporate psychopaths. We're using a tool 00:15:46.789 --> 00:15:49.570 that feels good, but, you know, it often fails 00:15:49.570 --> 00:15:51.970 to deliver the truth. It seems we have a tension 00:15:51.970 --> 00:15:55.350 between the ideal of a standardized assessment 00:15:55.350 --> 00:15:58.269 and the reality of human social interaction. 00:15:59.049 --> 00:16:02.070 Precisely. The science says one thing, but the 00:16:02.070 --> 00:16:04.440 practice says another. And perhaps the answer 00:16:04.440 --> 00:16:07.159 lies in being aware of both, understanding the 00:16:07.159 --> 00:16:10.080 power of the tool, but also respecting its limitations. 00:16:10.379 --> 00:16:13.480 And maybe, just maybe, being a little more skeptical 00:16:13.480 --> 00:16:15.639 of the person who charms the pants off you in 00:16:15.639 --> 00:16:17.720 the first five minutes. That is always good advice. 00:16:18.059 --> 00:16:20.399 Thank you for listening to The Debate. Whether 00:16:20.399 --> 00:16:22.779 you are the one hiring or the one hoping to be 00:16:22.779 --> 00:16:25.759 hired, we hope this gives you a new lens on that 00:16:25.759 --> 00:16:28.240 time -honored ritual. Good luck out there. You're 00:16:28.240 --> 00:16:29.779 going to need it. Until next time.