1. 6
  1. 13

    It’s a beautiful site, and obviously a fair amount of work went into this.

    But when you got 2,200 respondents from the US, and mostly people heard about it from Twitter, your sample size is so small and biased that how can you really draw any conclusions? There’s probably 10x the total US survey correspondents writing JS in just the Bay Area. Or in Austin, Seattle, NYC, or LA.

    only 21% of you used TypeScript compared to 69% today

    Look I love Typescript, and I think it’s going to replace vanilla JS eventually, I don’t think there’s any way that 69% of all JS devs are now using Typescript. 30% of people switched to Vite already? I bet less than 30% of all JS devs have even heard of Vite.

    I think there’s a lot of engineers who’re stuck with whatever was current in JS when their application was created. Maybe they’re slowly working-in some React, but half the codebase is still jQuery and handlebars. If they heard about Vite they’d be interested, but they haven’t even got the time to upgrade from Webpack 3.

    1. 4

      But when you got 2,200 respondents from the US, and mostly people heard about it from Twitter, your sample size is so small and biased that how can you really draw any conclusions?

      There are two different things here. One is the number of respondents. 2,200 respondents is a lot! As you can see by playing with https://www.calculator.net/sample-size-calculator.html, you get around a 2% margin of error at that size. For something like a US election where the final vote tends to be within 5% for either party, you need close margins of error, but for something like “do you use React?” two percent is more than sufficient. Also the size of the total population only matters when the size is so small that the sample is a sizeable portion of the total population. For something as large as JS developers, there’s no statistical difference in necessary sample size between if there are 100,000 JS developers in the world or 100 million.

      So the other issue is how is the sample drawn. And here is the survey is crap, obviously. When you draw an interested sample from Twitter, you are not getting a random sample, so the results aren’t useful. To work around that what you would want to do is to have a smaller, truly random sample, use that to get ballpark figures, and then have that rebalance your non-random sample. But these results are probably just worth what you paid for them. :-)

      1. 1

        When you got 2,200 respondents from the US, and mostly people heard about it from Twitter, your sample size is so small and biased that how can you really draw any conclusions?

        That’s a fair point.

        30% of people switched to Vite already? I bet less than 30% of all JS devs have even heard of Vite.

        Your point about whether this is a representative sample is taken, but “switched” is not what’s being claimed. “Used” is. One need merely run npm init vue@latest (Vue 3 ships with Vite), play with the scaffolding for a few minutes, and skim the Vite guide to get the gist of it.

        Taking myself for example, I’m working on several repositories with JS code, two of which have branches in which I’ve started down the path of upgrading to Vue 3. I’ve found Vite to be significantly faster and easier to configure than webpack. (Actually, I haven’t really had any need to “configure” it at all so far.) But I’m not done with the upgrade in either repo (for reasons unrelated to Vite). So have I “switched” to Vite? No. But have I “used” Vite? Yes.

        1. 1

          Fair point. But I claimed fewer than 30% of all JS devs have even heard of Vite, so whether it’s “switched” or “tried” their number is still off base 🙂

          1. 2

            A very back-of-napkin way to see this: there are 6 pages of vue.js questions asked on StackOverflow in the last 24 hours, 26 pages of react questions, and half a page of vite questions. Such-a-small number of questions should make us suspicious of such-a-large percent of people having heard of vite.

            1. 2

              Or Vite is just so easy to use no-one needs to ask SO questions ;)