Help! Psych Stats Question

[nerdlynoel]

Hi!

Right, so I have what is probably a really stupid question with a really obvious answer to it, but I am eye-ball deep in the midst of all the data and can't see the wood for the trees anymore.

Okay, the overall question of interest in my latest study was if rape survivors were seen as a separate group or if they are seen as a subgroup. The way I figured to best tell this was if they were rated highly in entitativity (read as 'group-ness') and low in group overlap. I figured the way to determine if the ratings were high or low would be to compare the mean scores to the midpoints of both scales. Here's where I get stuck. How? I mean, just looking at the means themselves, I can already tell that subgroup is the answer, but I need to run an analysis to justify this in my defence.

Anyone? Anyone please?

In advance, thankyouthankyouthankyou!

(Oh, and please don't get on a high horse about 'how dare you try to talk about survivors like that! Everyone is their own little snowflake!' I hear it from participants all the time and I promise I have a method to my madness.)

---

Members don't see this ad.

 

[WisNeuro]

Are you asking what kinds of tests you can run to evaluate mean differences between groups?

 

[Justanothergrad]

That's the only thing I can get out of that question too....

 

[EmotRegulation]

I can't follow this question at all, as I don't think we have enough information about "group" or about what measures you're discussing. That said, what I have been able to glean is a "how can I compare a mean score to the midpoint of a scale?" The answer to that is probably a one-sample t-test...where you can see if a single group's mean differs from a defined number. So if the midpoint of the scale is 3 (like on a 1-5 scale), you could set to see if one group's mean is significantly different from 3 using a one-sample t-test.

That said, it sounds like you are dealing with multiple groups, or subgroups, so that suggestion is probably not a good one, but it's all I could get from your question.

 

[Ollie123]

Agree with the above - I think if you give us more details about what your data looks like and what question you want to ask, we will be better able to advise. The way you phrased the question ("is there a subgroup") has me thinking about a number of much more advanced techniques than what is described above (taxometric analysis, latent class analysis, segmentation analysis). I suspect you are actually looking at answering a much simpler question (i.e. do rape survivors differ from <other group> on <variable name>), but I'm not certain.