Year 12s: what will be the first way you research your future options? >>

Is my statistical analysis (SPSS intepretation) correct?

Hello,
Could anyone check whether my interpretation is correct? Attached is a Word document containing a table of frequencies and the results of a statistical test imported from the SPSS programme, so all the work is done; I just need someone to make sure I have read it all properly.

My interpretation is that Chi2 = 8.663, df = 1, p< .003 and as such the null hypothesis can be rejected.

Is this correct? Has this interpretation been written out properly?

Any help is massively appreciated and might even earn you a rep.

Statistical Interpretation from SPSS.docx13.4KB

Reply 1

Wiggly

Bump (a bit urgent) :wink:

Reply 2

_Sinnie_

It looks fine to me, which shouldn't be taken as gospel. Though you could specify the null hypothesis in the text, just so it is clear what you are rejecting.

I wasn't sure about one tailed and two tailed - but google tells me that a Chi square is always one tailed.

Reply 3

Gregorius

Your interpretatio0n is basically correct. Only note that it is telling you that p=0.003 (rather than p < 0.003) to the degree of acuracy that is given.

You do get the two-sided test here as the chi-squared statistic can be "too small to be true" (i.e. observed are too close to the expected for it to be truly by chance).

Note that you also get a variety of other statistical tests; strictly speaking, the one that is appropriate depends upon the design of the experiment in question.

Reply 4

iammichealjackson

Seems fine, i doubled checked using an online chi-square calculator aswell.

You also need to state the direction of the statistical effect you found. (i.e. that smokers are more likely to have greater caffeine consumption). You're right that the two or one tailed distinction doesn't apply to the chi-square test.

As the above poster mentioned, you should write p=.003 not p < .003. Traditionally, psychology papers did used to write (p < .05 OR p < .01 OR p < .001) however this is not done any more because we have computers that give us exact p values rather than old-school tables in books which authors used to use. Therefore, the recent APA publication manual tells authors to give exact p values (it also helps for researchers reading papers since they can understand the size of the effect when doing things called meta-analysis).

If this was for a research paper, you would also need to give an effect size statistic (in this case the odds ratio statistic http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2938757/)

So in your case the odds of an individual being a smoker if they have high caffeine consumption is 4.86x greater than if they have low caffieine consumption (you would report this as OR = 4.86). This kind of thing may not be expected/included in an intro stats course though but would look a lot more professional!