Global sentiment: what can an open rating tell us?

Sentiment is not one single variable

Someone can be anxious about the news and satisfied with their own life. They can feel grateful for a personal event while angry about a public tragedy. They can give an ironic rating that looks numerically identical to someone else’s sincere judgement. A single response does not tell us which of these experiences produced it.

For that reason, global sentiment is best treated here as a research question rather than an established property of GodIndex. We are interested in how a short response might relate to a person’s circumstances and interpretations, not in relabelling every negative vote as a measurement of unhappiness.

The attention signal and the response signal

Imagine an index changes on a day when a widely shared article links to the site. Two things may have happened at once: more people noticed the project, and the responses of those who participated changed. Total activity is an attention signal; the distribution of ratings is a response signal. Neither should be used as a substitute for the other.

A future publication should keep these views together. A shift based on a small number of responses has a different evidential context from a shift during a large influx. Changes in recruitment can matter even when volume appears stable.

Three useful comparisons for a pilot

First, compare the same person’s responses over time with the aggregate movement. This can help distinguish within-person change from audience turnover, provided participation is appropriately consented and governed.

Second, compare the rating with a separately worded question about the participant’s current mood or life evaluation. Agreement is not guaranteed; disagreement would be informative about what the god question actually captures.

Third, compare responses with a pre-specified event calendar. Looking only for an event after noticing a dramatic movement creates an easy but unreliable explanation. A calendar established in advance can make negative findings visible too.

Useful to organisations, within limits

A public sentiment experiment could help a university teach measurement, a newsroom explore the relationship between news and response, or a company support independent research. It should not be sold as a validated forecasting tool, a substitute for customer research or a method of diagnosing individual wellbeing.

The present contribution is an accessible historical record and a set of study questions. New voting and research data are not yet live. See the research collaboration brief or partnership opportunities.