Sample Size

Posted: February 3, 2012 in Uncategorized

I don’t like maths, simple as. Especially as you can abuse maths to prove pretty much anything. LIES you say but check this out. Have you ever noticed that the number 11 has a powerful and mysterious connection to the 9/11 attacks.

  • The date of the attack: 9/11 – 9 + 1 + 1 = 11
  • *September 11th is the 254th day of the year: 2+5+4 = 11
  • *After September 11th there are 111 days left to the end of the year
  • .Twin Towers – standing side by side, looks like the number 11
  • *The first plane to hit the towers was Flight 11 by
    *American Airlines or AA – A=1st letter in alphabet so we have again 11:11
  • *State of New York – The 11th State added to the Union
  • *New York City – 11 Letters
    Afghanistan – 11 Letters
  • *The Pentagon – 11 Letters
  • *Ramzi Yousef – 11 Letters (convicted or orchestrating the attack on the WTC in 1993)
  • *Flight 11 – 92 on board – 9 + 2 = 11
  • *Flight 77 – 65 on board – 6 + 5 = 11

excuse me but i think that’s bulls*%t, now why have i used such an extreme example, because this next bit is boring, unlucky reader. I realized when playing with some data in out last small groups session that you can sometimes create a significant difference where there wasn’t one before by simply increasing your sample size. Now don’t jump to any conclusions i wouldnt just make up data im not going to lie. But why not just copy and paste the data you already have, its not like your inventing numbers its all true. so why not?
Comments welcome.

  1. psucb2 says:

    I loved this! not only did it prove a point but was funny :) However, i also think that sampls size is important when testing a hypothesis. I mean, yes increasing sample size should increase the chances of significant finding, so surely thats more incentive to do it rather than just copy and paste as you said? Plus, imagine going for surgery and it having a 90% success rate. Great right? but what if you find that it was only trialled on 10 people. that means for every 1 person in 10 in doesnt work. i wouldnt want to be that 1 in 10. I’d be way more happy if they had tested it on say 1000 people and still found the same results. So although i agree that stats can be manipulated fairly easily, i also think it is incredibly important to have a large sample size, even if the results are the same.

  2. tinkybelle says:

    I LOVED your blog… it really made me laugh :-D. However, I do think you raise an interesting and important issue about statistical manipulation. We too touched on the method using your data set 2 or 3 times if your effect was almost within significance to see if it affected the significance, but as I understood it, this was to be used in a purely preliminary manner, and if there was a change in the significance level, then the researcher must then go and gather more data and NOT just use their one set of data 3 times over for instance! The problem with doing this of-course, is that it increases the chances of a false positive result (Type I error) which we all know is very very bad!! I think this experiment-resources website explains it well. Of course, it works the other around too. A larger sample size is more representative of the population, whether the outcome is significant or not, it’s more likely to be accurate. I thought your point about “not making up numbers” and “using the data you already have because it’s not inventing numbers” was brilliant ☺ because I’ve found this article about something called Multiple Imputation for Missing Data which is a method of ‘filling in’ missing data with a number of plausible values, then making statistical inferences from that. So maybe you’re not that far off the mark when you talk about “making up data”…. maybe you should….then just give it a fancy name!!

  3. tinkybelle says:

    Ok so it didn’t like the links to the websites…. here they are (Sorry)

    “this experiment resources website” =

    Multiple Imputation for Missing Data article =

  4. Yes you can create a significant effect by increasing sample size, but that doesn’t mean you should do it. If you choose a big sample size to begin with then fair enough but if you do it specifically under the pretenses of making your non-significant result significant your going to create error in your results. So I guess under the circumstance of initially starting with a big sample to increase your chance of getting a significant result then fair enough. But if your just coping and pasting results, it kinda is almost like making up results. You can’t guarantee that the other people you could of tested would of got the exact same results as the other participants in your study. Which means effectively you are making up results as you are not accounting for the variability of the other individuals that may have been in your study if you had initially ran a bigger sample. Statistics as so easy to manipulate, but in manipulating results your cheating yourself and others. You made it up so the credit you get is false. Plus if it is for a treatment you may be providing something potentially harmful or something that doesn’t work at all.

  5. […] 6. […]

  6. psucc3 says:

    you have both raised interesting points, but i can reveal why you cant really get away with this, other than the fact its a bit wrong. When you run a T-test its under the assumption that “Samples must be independent”. So unfortunately repeating data just isn’t allowed in stats land.
    some other important things to note are that the sample has to be normally distributed and you need a homogeneity of variance , all things that are you don’t want to be messing around with if you want to be taken seriously.

  7. ellislee15 says:

    Funny blog, housemate! Your points made me think about how much numbers can be manipulated to make us believe things, for example in terms of psychology participants’ data can be removed out of results in order to make it seem as though there is a significant effect size. I think it shows us how careful we need to be when looking at research, and actually have a real look into the numbers in the results. Your idea of copying the results can be used in a good way to help us see if it would benefit the research by studying more people. I found this meme… valentines day is cancelled.

  8. kiwifruit8 says:

    well dave firstly your a tart, secondly did you really say:
    “But why not just copy and paste the data you already have, its not like your inventing numbers its
    all true. so why not?”
    Isnt the whole point of finding data is for it to be reliable,valid and generalisable? By copying and pasting previous data your showing the same audience twice, not finding more people to show how good your results are. I get that your saying the data is true but as we all know people believe bigger results because we tend to go with the majority but yours would be l-l-l-lying xx

  9. psud74 says:

    Interesting blog.. however just one thing you didn’t explain why increasing the sample size can give you a significant difference. Its because the higher sample size the lower the group variance and therefore a significant difference will be clearer. Also its funny how you said about making up data because it actually does happen out there a psychology professor, Prof. Diederik Stapel in nether lands was recently found out as he had made up a lot of data that ended up in published research papers. Its people like that tho that make us psychologist look bad.

    Andy Fields (2009) Discovering statistics using SPSS.

  10. […] Share this:TwitterFacebookLike this:LikeBe the first to like this post. […]

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s