Public-opinion research is the slowest fast-business in India. The instruments are old. The frames are imperfect. The pressure to be first is greater than the pressure to be honest.
I came to this work from journalism. For more than two decades I reported and produced the news across national television and print before I began measuring the opinion behind it. That apprenticeship taught me one thing above all � the ground is more honest than the assumption, and the truth in the booth is the only number history will eventually compare us against.
Matrize was founded in 2014 with a single commitment: a firm where every figure that leaves the building is attributed to a named researcher and disclosed with its margin and field dates. The signature is not vanity. It is the mechanism by which the firm corrects itself. Anyone whose calls overrun the interval too often loses the right to sign.
This page is the page on which I am accountable to you, to the network we serve, and to the voter we measure. The pages that follow are the pages on which the methodology is accountable to the next reader. Please read both.
Everyone who signs for Matrize commits to these in writing. There is no fifth � and no exception. The error log on the methodology page is the public proof we keep to them.
Every published figure carries its sampling margin, field dates and the confidence interval the model assigns to it. Numbers without intervals are not numbers; they are claims. The firm publishes neither.
Sampling, weighting, calibration, modelling � every step is published with every wave. We discuss our method and aggregate data with academic and press partners on request.
We publish what the field returns. Findings are not adjusted for the political colour of who commissioned them. Anyone who edits a number for the client has no place at the firm.
Every call that fell outside its disclosed interval is logged on the methodology page, with the named researcher and the assigned cause. We publish our hits the same way we publish our misses.
� Noida.
I have been asked, with rising frequency over the last three years, whether public-opinion research in India can still be trusted. The question is fair. The trade has been damaged — by partisan over-promising, by margin-less televised slates, by the ease with which a single bad call now overwhelms a decade of careful work. The answer is that it can. But only if the firms who do it agree, in writing, to be measured.
Matrize publishes a signed estimate. The signature is the mechanism. When you read a Matrize figure, you are reading a number that a named researcher has been willing to attach themselves to publicly, in advance of the count. When the count comes in, that same name is the one beside the call in the error log. There is no way to take that back. That is the point.
"A signed estimate, an interval, a margin, a field log. A wave is not a number. It is a published commitment."
The firm is independent, deliberately. From the outset, the rule has been that no figure leaves the building unless someone is willing to put their name to it. That commitment costs us speed, sometimes; it has also saved us from avoidable misses. It is what the architecture of the firm exists to protect.
We also publish what we get wrong. The error log is the only honest mechanism by which a research firm earns its standing. When a call falls outside the interval we disclosed, it is logged on the methodology page � with the cause, and the name beside it. The next time we model that kind of contest, we model it differently because of the miss.
The pages of this site that matter most are not this one. They are the methodology page, where the sampling frame, the weighting scheme, the calibration formula, and the error log all sit in plain text; and the tracker page, where the count is mirrored in public exactly as our control room sees it. Read those first. If the working there is honest, the figures here will be too.
We will continue to publish, to sign, to disclose, and to correct. The firm asks only that the press, the political stakeholders and the academic community we serve hold us to it.
More than two decades across Indian television and print. A career spent reporting elections before measuring them.
The Director's office reads correspondence from press partners, academic reviewers and political commissioners. Objections to a published figure receive priority. Available on request
