4 ms·

I was once one of these panelists. As one, I represented 0000’s of other people via their extrapolation. While my experiences at the time (2009) were common, th

by 12xo 6y ago

I was once one of these panelists. As one, I represented 0000’s of other people via their extrapolation. While my experiences at the time (2009) were common, they did not represent that of many of my fellow unemployed colleagues and friends. In other words: these surveys are guesses. Nothing more... take them as estimates only and know that the truth is always different.

mywittyname 6y ago
Is there a better alternative? I understand your criticism, but this is not a dataset that can be reasonably obtained in its entirety, so sampling is basically required. Their methods have been largely consistent over the years, so even if it isn't complete, it's easy to compare to historical trends, which is more important in this context.
- 12xo 6y ago
  Interesting that you feel this data isn’t available on a per case basis... it’s even more interesting that I end up with downvotes foe sharing my actual experiences. HN is tone deaf. Most of you folks work in modern tech! Modern in the sense that your company and job are likely based on big data... ugggh
  - govg 6y ago
    ....what they're saying is that there's no feasible way of getting an exact count for unemployment and you have to use some extrapolation.
  - skybrian 6y ago
    I think you got downvoted for dismissing the whole concept of random sampling. In a random survey you're not representing your friends. They will presumably be represented by randomly picking someone who happens to be more like them. (Also unumployed.)
    - 12xo 6y ago
      I didn’t dismiss anything of the sort, that’s what’s funny. I shared my actual experience of being chosen as a panelist and have a vastly different outcome and outlook than many of my friends. It’s not an uncommon assertion that the Fed data is not accurate. Number 1 it’s political. And perhaps more importantly, it’s designed to obfuscate the individual and that, while statistically accurate for static data is anything but for this type of situation. People and specific employment issues are anything but static...
      - IkmoIkmo 6y ago
        You literally said it's just a guess. It's not. It's a scientific method of estimation which has known expectations of accuracy, and when done right, has a track-record of being able to provide reasonable approximations. Saying it's 'just a guess' implies something very different for which you are rightly downvoted. Second, explaining the basics of random sampling by saying what applies to you, did not happen to apply to your friends/family, implies you have no understanding of the concept of random sampling either. There is absolutely no requirement for your situation to be the same as your friends/family to make this method useful, but by posting it as an argument, you imply that this does matter. For this, your post is also rightly downvoted. As for 'static data', it's a sample to provide insight in a snapshot in time. It doesn't claim to be anything but static. It does exactly what it sets out to do.
        12xo 6y ago
        Garbage in, garbage out.... of course the data over time will correct but the anomalies created by a few outliers are in fact a representation of bad practices. And those outliers are not measured because one cannot ever account for every potential situation when it comes to people. So they create a margin.... works for figures but NOT for policy!! My point,by the way. No wonder SV is still chasing ad optimization as it’s primary business. It’s impossible to ever become efficient, but that doesn’t stop people from chasing the paradigm. Because it’s the cracks and crevices that allow for exploitation, that’s why.... oh never mind, I’m talking psychology to people who think in binary and never the two shall twain
  - take_a_breath 6y ago
    People who work in big data understand how statistics, extrapolation and time-series work. You might be more interested in the ADP data, as they use actual payroll data. Their payroll research shows 20.2 million jobs lost [1]. [1] https://adpemploymentreport.com/2020/April/NER/NER-April-2020.aspx https://adpemploymentreport.com/2020/April/NER/NER-April-202...
6gvONxR4sf7o 6y ago
The plural of datum is data. If you are trying to correctly estimate a number like 14.7% to within 0.1%, you only need to survey maybe half a million people, making each person "represent" 600 others in your words. If you want it within just 1%, you only need about 5 thousand people, making each person "represent" about sixty thousand others. That doesn't make it any less correct, nor is it "extrapolation" in the usual sense. Closer to interpolation. It's definitely not just a guess.
- 12xo 6y ago
  I never mentioned the figures... I was referencing the panel. These were questions, not facts. I was asked a series of questions related to being unemployed and / or seeking employment. That data was extrapolated to millions of others and presented to the public. It was not accurate. My experiences were not common so their random use of my data meant the report was skewed. Even if only slightly. Math is objective. Words are not.
godelski 6y ago
> these surveys are guesses. Well yeah, that's what statistics is. What matters is your confidence level. Logistically I can't sample the entire population of a city, let alone a country. Even if I could I can't guarantee that people are answering honestly. Guesses are all you are ever going to get.

by 12xo 6y ago

mywittyname 6y ago

Is there a better alternative? I understand your criticism, but this is not a dataset that can be reasonably obtained in its entirety, so sampling is basically required. Their methods have been largely consistent over the years, so even if it isn't complete, it's easy to compare to historical trends, which is more important in this context.

12xo 6y ago
Interesting that you feel this data isn’t available on a per case basis... it’s even more interesting that I end up with downvotes foe sharing my actual experiences. HN is tone deaf. Most of you folks work in modern tech! Modern in the sense that your company and job are likely based on big data... ugggh
- govg 6y ago
  ....what they're saying is that there's no feasible way of getting an exact count for unemployment and you have to use some extrapolation.
- skybrian 6y ago
  I think you got downvoted for dismissing the whole concept of random sampling. In a random survey you're not representing your friends. They will presumably be represented by randomly picking someone who happens to be more like them. (Also unumployed.)
  - 12xo 6y ago
    I didn’t dismiss anything of the sort, that’s what’s funny. I shared my actual experience of being chosen as a panelist and have a vastly different outcome and outlook than many of my friends. It’s not an uncommon assertion that the Fed data is not accurate. Number 1 it’s political. And perhaps more importantly, it’s designed to obfuscate the individual and that, while statistically accurate for static data is anything but for this type of situation. People and specific employment issues are anything but static...
    - IkmoIkmo 6y ago
      You literally said it's just a guess. It's not. It's a scientific method of estimation which has known expectations of accuracy, and when done right, has a track-record of being able to provide reasonable approximations. Saying it's 'just a guess' implies something very different for which you are rightly downvoted. Second, explaining the basics of random sampling by saying what applies to you, did not happen to apply to your friends/family, implies you have no understanding of the concept of random sampling either. There is absolutely no requirement for your situation to be the same as your friends/family to make this method useful, but by posting it as an argument, you imply that this does matter. For this, your post is also rightly downvoted. As for 'static data', it's a sample to provide insight in a snapshot in time. It doesn't claim to be anything but static. It does exactly what it sets out to do.
      - 12xo 6y ago
        Garbage in, garbage out.... of course the data over time will correct but the anomalies created by a few outliers are in fact a representation of bad practices. And those outliers are not measured because one cannot ever account for every potential situation when it comes to people. So they create a margin.... works for figures but NOT for policy!! My point,by the way. No wonder SV is still chasing ad optimization as it’s primary business. It’s impossible to ever become efficient, but that doesn’t stop people from chasing the paradigm. Because it’s the cracks and crevices that allow for exploitation, that’s why.... oh never mind, I’m talking psychology to people who think in binary and never the two shall twain
- take_a_breath 6y ago
  People who work in big data understand how statistics, extrapolation and time-series work. You might be more interested in the ADP data, as they use actual payroll data. Their payroll research shows 20.2 million jobs lost [1]. [1] https://adpemploymentreport.com/2020/April/NER/NER-April-2020.aspx https://adpemploymentreport.com/2020/April/NER/NER-April-202...

6gvONxR4sf7o 6y ago

The plural of datum is data. If you are trying to correctly estimate a number like 14.7% to within 0.1%, you only need to survey maybe half a million people, making each person "represent" 600 others in your words. If you want it within just 1%, you only need about 5 thousand people, making each person "represent" about sixty thousand others. That doesn't make it any less correct, nor is it "extrapolation" in the usual sense. Closer to interpolation. It's definitely not just a guess.

12xo 6y ago
I never mentioned the figures... I was referencing the panel. These were questions, not facts. I was asked a series of questions related to being unemployed and / or seeking employment. That data was extrapolated to millions of others and presented to the public. It was not accurate. My experiences were not common so their random use of my data meant the report was skewed. Even if only slightly. Math is objective. Words are not.

godelski 6y ago

> these surveys are guesses. Well yeah, that's what statistics is. What matters is your confidence level. Logistically I can't sample the entire population of a city, let alone a country. Even if I could I can't guarantee that people are answering honestly. Guesses are all you are ever going to get.