#stats 101

Anthro: lang and cultures edition

Stats 101

Anthro: bio edition

Intro to Psych

Writ & Comp

Hello friends!  Though you don’t technically need to know statistics to understand the next chapter (because it will be explained in dialogue), I thought that it would be nice to include some real world examples and definitions.  So welcome to Statistics 101!  You can call me Professor Aisling.

What is the study of statistics?

Statistics are formulas and methods that allow you to analyse data from your research as well as understand the research of others.  Specific formulas are used depending on what is being analysed and what information you want to get from the raw data.

Why is it important?

Statistics allow you to understand scientific research, graphs, data, and information around you.  Scientific literacy through the use of statistics makes the difference between understanding junk science versus the real deal.  So many graphs and stats used in news and advertising are purposefully deceiving, and statistics allows you to read this information, understand what the data is really telling you, and then apply this information to your life.  it makes you a smarter consumer, voter, and citizen.

Hypothesis vs Null Hypothesis

Your hypothesis is a predictive statement or your educated guess about what will happen as a result of your experiment.  It acknowledges that your independent variable will cause your dependent variable.  Or it acknowledges what you believe to be fact.

The null hypothesis is the opposite statement.  The null hypothesis states that the independent variable does not cause the dependent variable, as far as the research shows.  This is like a default understanding of the research because you don’t have any evidence to show otherwise.

Your goal during research is to reject the null hypothesis.  You want to say that the current understanding of the research is incorrect, and your hypothesis is going to replace the old knowledge.  When you reject the null hypothesis, you are confirming that your hypothesis is significant.  It’s not proof that you are correct, but you are able to say that your data shows an outcome with statistical significance.

So how do you find statistical significance?  By using a statistical formula that fits the data you are using.

P-Values: What are they?

Today, we will be focussing on p-values, which come about as a result of a Chi Squared test when analysing data.  A p-value is your gateway to understanding whether you have reached the threshold for statistical significance in your research.

Your p-value is a number that must be smaller than a certain cut-off point in order for you to reject your null hypothesis.  P-values run between 0 and 1.  The lower, the more significant your findings.  The higher, the less significant.  You want to get the lowest number possible so that you can say, “Hey, my data shows pretty definitively that my hypothesis is correct.”  The number itself represents the chance of error that your hypothesis could be a false positive, so a low number is a low chance.

Most areas of science use a threshold of 95% for statistical significance, meaning that you can reject your null hypothesis and accept your hypothesis if there is a 95% or higher chance of it being correct.  So think of the 95% as .95.  Meanwhile, the p-value needs to be 5% or lower for a chance of error, which can be thought of as .05.  A p-value of .05 or lower is cause for celebration, because it means that there is less than a 5% chance that you totally messed up.

As an example, say that you have a p-value of .001 (p<0.001).  That translates into this statement: There is less than a 0.1% chance that my hypothesis is not correct.  That is less than a 1 in 1000 chance.  That’s pretty damning evidence in your favour.  Now, say that your p-value was .39 (p=0.39).  That means that there is a 39% chance that you are wrong, which is not good enough for science.  You would have to accept your null hypothesis in this case.

REAL WORLD EXAMPLE TIME!

So, let’s put this into practice!  Say that I have a hypothesis that Kylo Ren fans are more likely than the general population to have a Tumblr account.  The null hypothesis would be that there is no evidence that Kylo Ren fans are any more likely than the general population to have a Tumblr account.

You collect your data through surveys of Kylo Ren fans and place it into a statistical analysis programme such as SPSS or MiniTab, which gives you a big list of statistical data.  Most importantly, you find that the p-value is .67.  Well darn!  That is way higher than our p<.05 threshold, so we would not reject our null hypothesis and would say that there is no statistical evidence to support that Kylo Ren fans are more likely to be on Tumblr.

But say that your p-value comes back as .003.  Wow!  That’s great!  Way lower than p<.05.  In this case, you can reject your null hypothesis and accept the hypothesis that Kylo Ren fans are more likely to be on Tumblr than the general population.