1
$\begingroup$

I have been working on this for the past 4 hours.. sadly. I need to get the null and alternative hypothesis from an article that states a p value, and I chose this one.

http://eds.b.ebscohost.com/eds/pdfviewer/pdfviewer?sid=b537533e-4934-4987-acf8-b9e1e3e12d34%40sessionmgr113&vid=1&hid=104

Volume 39, Number 2, March-April 2015, American Journal of Health behavior, "Subjective Social Status and Readiness to Quit among Homeless Smokers."

It seems that the hypothesis (the one of many I chose) is "Higher SSS Community rankings will relate to greater readiness to quit." However, I am not sure how I could convert that to a null/alt hypothesis seeing as I don't know what numbers for me to set a mean equal to. The paper contains a lot but in my attempt to summarize, it attempts to show that a higher SSS rating (perception of your own status in the social hierarchy) among homeless people is linked to a higher readiness to smoke.

To simplify, how to I get both hypotheses (null/alt) in an article such as the listed when it isn't layed out in a nice question format such as what I am used to? What is the null and alternative in this situation?

What I understand so far and seem to be the information necessary to know is this:

The readiness to quit smoking at the lowest SSS Community (1) is -.03 and for the highest SSS Community (9) is .32, so the hypothesis is true since more are ready to quit if the SSS is higher. The probability is is "p < .001"

My problem is, I don't know how to put it into the format I was taught, which is Ho: (Mu) = x I am not sure how to take the information I have and get the two hypotheses.

$\endgroup$
  • $\begingroup$ Please add the [self-study] tag & read its wiki. Then tell us what you understand thus far, what you've tried & where you're stuck. We'll provide hints to help you get unstuck. $\endgroup$ – gung - Reinstate Monica Aug 5 '15 at 23:05
  • $\begingroup$ Also, please paste into the body of your question whatever context is necessary to understand it. Eg, provide a complete citation for the paper (in case the link goes dead), & paste quoted text from the paper (perhaps w/ terms defined or whatever is necessary). $\endgroup$ – gung - Reinstate Monica Aug 5 '15 at 23:07
  • $\begingroup$ The "format you were taught" seems to be for a very specific kind of test, where you're comparing a mean to a prespecified value (likely a one-sample t-test). There are dozens of other tests in common use. The paper looks like it's testing whether a correlation is zero (or not zero under the alternative), which is similar to what you're familiar with, but somewhat different. $\endgroup$ – Glen_b -Reinstate Monica Aug 6 '15 at 7:38
0
$\begingroup$

The notation for a null hypothesis that you have listed seems to be for a one-sample t-test. The description of the content of the paper (which I have not read) seems to be about a correlation.

$\endgroup$
  • $\begingroup$ So it isn't possible with this article? Most of the articles I have found were like this, as in it didn't lay them out as simply as what I've seen so far. And your question seems to imply that there is a way to do it, but not in that format? Is that right? $\endgroup$ – Tony Kay Aug 5 '15 at 23:18
  • $\begingroup$ @TonyKay, the null hypotheses for different tests differs. So you have 1 type of null for a 1-sample t-test, a different type for a 2-sample t-test, the null for a correlation is different again, etc. I don't see a problem w/ using this paper unless your assignment specifies a 1-sample t-test. $\endgroup$ – gung - Reinstate Monica Aug 5 '15 at 23:34
  • $\begingroup$ It doesn't specify, so I guess I would need to calculate the null for both cases and compare them? That somewhat makes sense. So say I took the lowest level (1). And, then I would do: Ho: Mu = -.03, right? If that't the case, then would would I try to do for alt? Alt is where I would compare right? $\endgroup$ – Tony Kay Aug 5 '15 at 23:40
  • $\begingroup$ I wouldn't try to come up with a bunch of 1-sample t-test nulls. If what they did was something ultimately related to a correlation, I would come up with a null for that. As for the alternative hypothesis, I would figure out the null 1st & then think about how the null & alternative have to relate logically. $\endgroup$ – gung - Reinstate Monica Aug 5 '15 at 23:52
  • 1
    $\begingroup$ @gung: sorry I deleted the comment $\endgroup$ – user83346 Aug 6 '15 at 6:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.