# Visualize Raw Data to Make Better Decisions

## Метаданные

- **Канал:** NNgroup
- **YouTube:** https://www.youtube.com/watch?v=JLdQeE4I5YM

## Содержание

### [0:00](https://www.youtube.com/watch?v=JLdQeE4I5YM) Segment 1 (00:00 - 04:00)

If there's one thing I learned in my PhD, it was never get bit by a monkey. But if there was another thing I learned, it would be to look at your raw data. I know the temptation to rush through data analysis, especially when you're on a deadline or under pressure. But whether you're conducting qualitative or quantitative research, checking and visualizing your raw data is the best way to make better analysis decisions and critical errors that might otherwise go unnoticed. I want to show you an example of how this might happen with quantitative analysis. Imagine you've just finished collecting data from a usability study. 40 participants completed your e-commerce checkout task and you're eager to analyze the results. What's the first thing you do? Well, if you're like most researchers, you probably start with summary statistics. So, let's look at our data. The mean time on task is 141 seconds. The median 105 seconds. standard deviation of 67 seconds. These numbers seem perfectly normal. Nothing alarming here. You might think, great, let's run a t test and compare this to our benchmark. But wait, let me show you what actually happens when we visualize this data. Look at this histogram. Do you see what I see? That's not one bell curve. That's two distinct peaks. We have a biodal distribution hiding in plain sight. One group of users, about 25 people, completed the checkout in around 90 seconds. But another group, 15 users, took nearly 4 minutes. Think about what we almost missed. If we had rushed into statistical testing, we'd have violated basic assumptions about normal distributions. Our results would have been meaningless. But more importantly, we would have missed a critical UX problem. Why are 37% of our users taking almost three times longer to check out? Maybe they're hitting a confusing error message. Maybe the shipping options are unclear. Maybe there's a technical bug affecting certain browsers. Whatever it is, it's potentially costing us over a third of our customers. The average told us nothing. The visualization told us everything. So here's your takeaway. Before you run a single statistical test, before you calculate any complex metrics, just look at your data, create a histogram, plot the individual points. Because sometimes the most sophisticated analysis you can do is simply opening your eyes. This can also happen with qualitative studies. If you jump to using transcript analysis tools or AI powered thematic analysis, you might miss out on crucial patterns that you would have detected by reading through transcripts or watching recordings yourself. Looking at and visualizing your raw primary data isn't a waste of time. Quite the contrary. In fact, it can be the highest value action and the biggest return on investment that you can do. You spent so much time and so many resources gathering this data. You should become intimately acquainted with it. Remember, visualization isn't just a nice to have. It's your first and best defense against misleading yourself. Thanks for watching. If you want to see more of our UX videos, take a look at these over here and consider subscribing to our channel. On our website, nnngroup. com, you can access our free library of over 2,000 articles. You can also register for one of our virtual UX conferences that offer live hands-on UX training.

---
*Источник: https://ekstraktznaniy.ru/video/43265*