Cross Tabulations and P Values?

12-Jul-17, 11:37
edited about 19 seconds later
by Anna10
Avatar for Anna10
posted about 2 months ago

I got feedback from a journal reviewer about some cross tabulations I did in a paper I wrote from my PhD research (which I completed 2 years ago). I did a mixed method exploratory study (interviews followed by a survey). My sampling method was non probability for the survey.

The reviewer commented on the analysis of the survey data that I did using cross tabulations in SPSS. The reviewer wrote;
“The 'crosstab' analysis should be supported by p values to show strength of correlations.”

I included the cross tabulations graphics only in the paper.

When I look back at my data analysis I cannot see any p values and there must have been a sound reason why I did not do this (as I spend weeks deciding on what statistical tests stats to do based on the data I had). For the life of me, however, I cannot remember why there are no p values now. Instead of spending another few weeks reading up on it all again I am hoping there might be a stats expert on here that might know the answer immediately?

The reason I used cross-tabulations because I wanted to do multivariate analysis. The cross tabulations were done to examine the relationship between categorical variables in greater detail than frequencies for individual variables.

All help welcome!
13-Jul-17, 14:30
edited about 1 minute later
Avatar for Tudor_Queen
posted about 2 months ago
Hi Anna

I have a feeling that you did cross tabs and eyeballed the table for relationships. And additional step you could have asked SPSS to take would be Chi Square analysis, which would give you p-values - which you could interpret to suggest whether those relationships were likely to be significant (not just at chance).

As far as I understand it is your choice whether or not you took that extra step, and it seems perfectly valid to have just eye-balled the data. But if the reviewer wants p values and you have the dataset handy then you can easily go back and re run the cross tabs PLUS Chi Square analysis.

Or if you see other papers where it wasn't done and you really don't think it necessary in your paper/can't access the data at the moment then I guess you could just try to justify not presenting them.

Hope this helps.



Copyright ©2011
All rights reserved

Postgraduate Forum

Masters Degrees

PhD Opportunities

PostgraduateForum is a trading name of FindAUniversity Ltd
FindAUniversity Ltd, Sellers Wheel, 151 Arundel Street, Sheffield, S1 2NU, United Kingdom. Tel +44 (0) 114 268 4940 Fax: +44 (0) 114 268 5766