• 𝕯𝖎𝖕𝖘𝖍𝖎𝖙@lemmy.world
        link
        fedilink
        English
        arrow-up
        0
        ·
        edit-2
        10 months ago

        We asked respondents to describe their current income situation.

        You mean 40% of forbes readers live paycheck to paycheck?!

        This may surprise you but not everyone reads forbes.

        (selection bias)

          • 𝕯𝖎𝖕𝖘𝖍𝖎𝖙@lemmy.world
            link
            fedilink
            English
            arrow-up
            0
            ·
            edit-2
            10 months ago

            I mean, forbes did the survey. Do you think they didn’t use their reader base?

            As for only surveying readers, lemme look at the survey and see if it said that. My concern is that a magazine performing a survey is going to have some issues with selection bias.

            edit:

            So, in this case, it suffers from a very small sample size. There may be funding concerns too.

            • iopq@lemmy.world
              link
              fedilink
              English
              arrow-up
              0
              ·
              10 months ago

              3000 is not a small sample size. 30 is a small sample size.

              Presidential polls are often done with this, or even smaller sample sizes.

              • 𝕯𝖎𝖕𝖘𝖍𝖎𝖙@lemmy.world
                link
                fedilink
                English
                arrow-up
                0
                ·
                10 months ago

                This may surprise you to know but America has a little over 3000 people in it. Some estimate the number to be over 1,000,000!

                Now, if we estimate the american population to be at 1,000,000, that means 3,000 is a sample size of only 0.3%.

                Now, 0.3% may seem like a large number and indeed it is when compared to 30, which would be a sample size of 0.03%.

                I wonder how big the working population of america is?

                United States Employed Persons was reported at 161,183,000.000 Person in Dec 2023 See the table below for more data.

                Hmm, so it seems 161,183,000 is a bit larger than 1,000,000. It’s actually about 161.183 times my original estimate! That would make 3,000 people bout 0.001%.

                Now, 0.001% is a very small number.

                • iopq@lemmy.world
                  link
                  fedilink
                  English
                  arrow-up
                  0
                  ·
                  10 months ago

                  You might want to look up sampling. If the sample is randomly distributed, you can calculate the chance that the given data is because of sampling error.

                  You see, it doesn’t matter what the total population is for the sampling error. It only matters what your number of samples is and the absolute difference between the groups.