Chi-Square

Hi,

I was just wondering whether anyone could help me with this.

In order to calculate chi-square, you have to calculate the expected value - so I would have thought that the expected value would match up to the hypothesis. However, in a calculation I have done my expected value is different to the hypothesis. It mathematically seems correct as I have checked my calculation but theoretically seems incorrect? :confused:

Is it just that I'm interpreting it wrong? Is the expected value just some sort of statistically expected value and not related to what you're actually predicting?

It's a bit complicated to explain, but I hope that makes sense!

Thanks,
Natalie

Reply 1

beautifulbigmacs

You need to double check this with someone but I think if the number for the hypothesis is different to the actual outcome then it suggests that the hypothesis turned out not to be.

I've not done chi square tests for a while so be sure to check this with someone.

Reply 2

iammichealjackson

Original post by NatalieMLW

I think your confused about what the expected values mean.

The expected values are the values you would expect to see under the null hypothesis (that there is no effect). The expected values are entirely determined by what your null hypothesis is- SPSS sets the null hypothesis automatically, but you can change it to what ever you want. The expected values are not the values that you predict with your hypothesis, but the values you expect when there is no effect (under the null hypothesis).

When the observed values (i.e. the data you collected) strongly deviate from the expected values, the test statistic increases. You get a "signfiicant" result when the test signifcant is high and there is a large difference between the observed and expected values.

(edited 8 years ago)

Reply 3

beautifulbigmacs

Sidenote....is chi square covered at A level or is it only encountered once at uni?

Reply 4

iammichealjackson

Original post by beautifulbigmacs

Sidenote....is chi square covered at A level or is it only encountered once at uni?

Both? I mean the statistics in psychology a level is pretty pointless memorising. I dunno if its in statistics a level. Its a fairly standard test in psychology and you probably encounter it at uni too.

Reply 5

beautifulbigmacs

Original post by iammichealjackson

Thanks. So for A level, are you saying that although chisquare is encountered, you just have to memorise concepts relating to them rather than actually do them?

Edit: i want to teach sociology A level but am weighing up if i would be suitable for studying a bit of psychology at uni to have the option of perhaps teaching that also.

(edited 8 years ago)

Reply 6

iammichealjackson

Original post by beautifulbigmacs

Yeah at least when i did it it was pretty much like you only have to remember the name of the statistical test (e.g. t-test, chi-square, ANOVA, pearson correlation) to use in a particular circumstance. It may change in the future though, i really doubt statistics at a level psychology will ever get more complicated than that, although i think its a complete waste of time to learn the names of particular statistical tests at a level so maybe it will change in the future.

Reply 7

NatalieMLW

Original post by iammichealjackson

Ahh that makes sense. Thank you so much!

Reply 8

powerofanalytics

Hi, the chi-square statistics test is widely used. Most common example is in A/B tests for web pages (example from everyday practice: http://getdatadriven.com/ab-significance-test). To use this case to explain it: your null hypothesis is, that you expect the two different A/B web pages will generate the same % of conversions. Your alternative hypothesis, you would like to prove is, that the observed values generated by each of the web page are acctually statistically different for A/B page from the expected ones.

I hope this helps. I am sure chi-square test will come in handy for you also in social sciences :smile: