Feature Update: Confidence Intervals/Statistical Significance

At Cleria, our mission has always been to provide you with the most accurate, actionable, and insightful data through our Brand Trackers. To that end, we are excited to introduce a significant feature update: Confidence Intervals.

Confidence Intervals: A Brief Overview

For those unfamiliar with the term, a confidence interval provides a range in which a metric is likely to fall, with a specified level of confidence. In other words, instead of relying solely on a single point estimate, confidence intervals give you a range that offers a clearer picture of the variability and reliability of your results.

Confidence Intervals: A Deeper Dive

At its core, a confidence interval provides a statistical range that, based on your data, an actual value is expected to fall within, given a specified level of confidence. But let's break that down a bit more:

• Interval Range: Instead of offering a singular data point or value, confidence intervals provide a range of values. For example, if you were examining brand perception in your brand tracker, instead of stating it as a singular value like 75% for a specific image item, a confidence interval might suggest the true value lies between 72% and 78%.
• Level of Confidence: The "90% confidence" we mentioned signifies the probability that the true value falls within the provided interval. So, if we repeatedly ran a specific brand tracking wave with different respondents we would expect the value to fall within this range 90% of the time.
• Why Not 100% Confidence?: It's a natural question to ask. However, in statistics, achieving 100% certainty often requires testing the entire population, which is not always feasible. Therefore, we typically use confidence levels like 90% (or, less often, 95% or 99%) to convey a high degree of certainty without asserting absolute perfection.
• What It Tells Us: Consider you're comparing two values, and their confidence intervals for a certain metric don't overlap. This can be a strong indication of a genuine difference between the two. On the other hand, if the intervals do overlap, it suggests that the difference might not be as pronounced or clear-cut.

The beauty of confidence intervals lies in their honesty. They openly acknowledge the inherent uncertainty in data collection and analysis, offering a more nuanced view than a single number could.

Why This Matters

1. Enhanced Data Reliability: Confidence intervals offer a more transparent view of the potential variability in the data. It helps in understanding the range within which the actual metric might lie.
2. Informed Decision Making: Instead of making decisions based on a single point estimate, you now have a range, enabling you to account for uncertainties. This means you can make more risk-informed choices.
3. Better Interpretation of Results: Confidence intervals can offer insights into the significance of your results. If two brands' intervals don't overlap, for example, it's a strong indication that there's a genuine difference between them.

Activating "Intervals" on the Dashboard

To make this feature as user-friendly as possible, we've incorporated a straightforward "Intervals" activation option (as part of the "dynamic" graph display) on the dashboard. Once toggled, every metric will display its corresponding confidence interval at a 90% confidence level.

To get the exact confidence interval, hover over an interval to see the lower bound value, the exact value from the survey and the upper bound value. You can change the color of the interval (as with any other color for any graph) under Administration --> Diagram Colors.

‍

Conclusion

While numbers are invaluable, understanding the context and reliability behind those numbers is equally critical. By introducing Confidence Intervals, we aim to bring more clarity, transparency, and robustness to your data-driven decisions.

As always, our team is on hand to assist with any questions or deep-dives into this new feature.

‍