Rahul Kumar Singh
Every year, after the UPSC results are declared, the country celebrates hard work, resilience and merit. Yet in parallel, another story begins to gather force: viral posts, angry threads and familiar comparisons with Dronacharya, all pointing to one question, whether candidates from marginalised communities are being given lower interview marks despite performing as well as or better than socially advantaged groups in the written examination.
Since such claims are often dismissed as reactions to a select few cases, I decided to investigate: does the pattern hold in the data? I analysed UPSC results of selected candidates from 2020 to 2025 to see whether interview scores systematically differ for marginalised candidates once written marks are taken into account.
After all, the interview is where merit stops being assessed through standardised written performance and begins to be filtered through subjective judgement. The interview is meant to evaluate communication, confidence and personality, among other things. That makes it important, but also harder to assess fairly. In a high-stakes examination, even a small difference in interview marks can change final rank, service allocation and career trajectory.
Study and outcome
My results show that the concern cannot be dismissed as a handful of isolated cases. The raw differences in interview marks are visible in the data and the gap becomes larger once I compare candidates with similar written-test performance.
On controlling for written marks, year effects and disability status, I find that candidates from reserved categories receive lower interview scores than comparable General Category candidates. The estimated gap is about 12 marks for candidates from Other Backward Classes (OBC), 13 marks for Economically Weaker Section or EWS candidates, 20 marks for Scheduled Caste candidates and 21 marks for Scheduled Tribe candidates.
These are significant differences in an examination where a single mark can strike one off the merit list. At the same time, this evidence should be read with caution. The analysis is based only on selected candidates, not the full pool of those who appeared for interviews. Because selection depends on written and interview marks together, conditioning on the final list mechanically induces a negative association between the two within my sample.
That is, a candidate with relatively low marks in the written paper might still have scored well enough in the interview to make it to the final cut. The opposite also holds: Very high marks in a written paper can offset relatively lower interview scores for many candidates.
For this reason, higher written scores appear to predict slightly lower interview marks here. This pattern need not hold among all interviewees. For the same reason, I cannot fully rule out selection effects or establish discrimination conclusively for the entire pool of interview candidates.
The pattern becomes even more meaningful when I look beyond marks and examine rank movement. When final rank is compared with a new ranking, based only on marks in the written test, reserved category candidates lose position after the interview stage is added.
Relative to General candidates, OBC candidates lose about 48 places, EWS candidates about 52 places, Scheduled Caste candidates about 70 places and Scheduled Tribe candidates about 73 places on average. In simple terms, the interview stage does not merely confirm written performance. It actively rearranges the ordering of candidates and it appears to do so in socially patterned ways.
There is another troubling result. The return to written performance in interview marks is weaker for Scheduled Caste and Scheduled Tribe candidates. That is, if two candidates improve their written score by the same amount, the increase in interview marks is smaller for Scheduled Caste/Scheduled Tribe candidates than for General candidates.
Put simply, stronger written performance does not convert into interview marks equally across social groups. This does not by itself prove discrimination. But it does raise a legitimate question: what exactly is being rewarded in the interview room?
How to interpret these findings
There are two broad ways. The first is the possibility of social bias. Interview boards may not intend to discriminate, but subjective evaluation can still reproduce hierarchy. Certain accents, styles of speaking, forms of confidence or ways of framing answers may appear more suitable simply because they are socially familiar or their views more acceptable to evaluators. In that situation, the interview may begin to reward cultural ease rather than merit.
The second explanation is structural, and in many ways, it may be more important. Candidates from socially advantaged backgrounds often enter the interview process with forms of preparation that are difficult to see in official data. They may come from households or social circles where people have already appeared for civil services interviews or similar examinations. They may also be better placed to access expensive coaching, mock interviews and one-to-one mentoring that are unaffordable for candidates from marginalised backgrounds.
They may receive advice on how to sit, how to answer, how to handle stress questions, how to project confidence and how to sound mature in a formal setting. These are forms of social capital that not all have access to equally.
None of this implies that disadvantage is hidden. Inter-generational poverty, malnutrition and exclusion are extensively documented. What is harder to see in official data is the other side of the ledger: how advantage quietly translates into ease and familiarity inside the interview room. Candidates from marginalised communities often do not possess the same support system.
Many are first-generation entrants into spaces like the UPSC interview room. They may have the same written ability, the same seriousness and the same determination, but not the same familiarity with elite institutional norms. In such cases, the interview may be measuring inherited exposure as much as it is measuring suitability for public service. This is why the current debate of whether individual board members are biased.
The deeper issue is whether the interview stage – even if it happens unintentionally – converts social background into evaluative advantage. That is not a private complaint. It is a public institutional question. It goes to the heart of how the Indian state defines and rewards merit.
The UPSC’s credibility rests on the belief that it identifies merit fairly. That is why assessing the effects, if any, of subjective evaluation will only strengthen it.
What should be done?
First, the UPSC should consider greater transparency in interview evaluation. This does not require compromising confidentiality. The commission can release more information on score distributions, variation across years and broad evaluation criteria. Greater transparency would reduce speculation and allow evidence-based discussion.
Second, the interview process itself deserves review. More structured interviews, clearer rubrics and greater standardisation in evaluation may reduce the role of impressionistic judgements. The purpose of the interview should be to identify judgement and suitability in public work, not to privilege social performance.
Third, if the problem lies partly outside the interview board, then the response must also go beyond the board. Universities, public institutions and state agencies should develop and strengthen mentoring and interview-support systems for first-generation and marginalised aspirants. If some candidates enter the room with years of social preparation behind them, fairness requires that others are not left to bridge that gap alone.
The UPSC result season always brings celebration, disappointment and debate. This year, it should also bring reflection about whether it is evaluating public potential fairly across India’s deeply unequal social landscape. Until that question is taken seriously, the debate over interview marks will keep returning, and rightly so.
Also Read: UPSC Cracks Down With AI: 569 Civil Services Applications Rejected Before Prelims https://www.vibesofindia.com/upsc-uses-ai-to-reject-569-ineligible-civil-services-applications-before-prelims-2026/
