Hi guys,

I'm so stressed out and feel like I could explode at any time. I really need some help. I'm doing (what feels like millions of) chi square tests exploring the relationships between respondents key characteristics and pro-environmental behaviour. I can do the test fine, its the interpretation of it. I want to know the proportion of respondents (categorised into 6 groups) that are more likely to act pro-environmentally. I don't know whether I have to use the percentages in rows or columns to figure this out! Some things have said columns, other have said rows.... and I just don't know!!

For example:

Category Act
No Yes
1 Count 1 167
% within row 1.2% 98.8%
% within column 1.1% 38.4%

2 Count 4 173
% within row 2.3% 97.7%
% within column 2.2% 39.8%

3 Count 24 35
% within row 40.7% 53.9%
% within column 13% 8%

4 Count 33 28
% within row 54.1% 45.9%
% within column 7.9% 6.4%

5 Count 74 12
% within row 86% 14%
% within column 40.2% 2.8%

6 Count 47 20
% within row 70.1% 29.9%
% within column 25.5% 4.6%

So do I say that those more likely to act pro-environmentally are those respondents in category 1 (98.8%) (row) OR do I say that those more likely to act are those in category 2 (39.8%) (column)?

Conversely, do I say that those least likely to act are those in category 5 (86%) (row) OR do I say again its those in category 5 (40.2%) (column)?

Please help me out... I'm struggling so badly. Everyone is confusing and my supervisors/university are pretty much useless and not supportive!

I would appreciate any help! Thank you in anticipation

Have uni not put you on to any statistical modelling/analysis software rather than doing this by hand? I don't do stats myself but know that lots of people use software for this. Although I can't tel you what software to use I'm sure there are lots of people on this forum who will know.

I am using software, I'm using SPSS. I'm not having an issue with doing the stats or using the software, just interpreting the results! Do I figure out the proportion more likely to take measures to protect the environment using the percentage in rows or columns (using the example I provided in my original post)?

As it stands, for this example.... I'm using column to state that those in category 1 and 2 are more likely to act pro-environmentally (78.2% - 38.4%+39.8%). I'm not sure whether this interpretation is correct though!! Can anyone clarify?

Hi Smiler,
Deep breath! I would really recommend that you sign up to the PSYCH-POSTGRADS list on JISCmail (Google it) and post your question there. There are many contributors there who have extensive knowledge of stats, particularly Jeremy Miles who has written quite a few stats books. They're usually really helpful.

Good luck!

Hi Noctu,

I've tried to access, and post to, PSYCH-POSTGRADS on JISCmail but unfortunately it won't let me post there. I'm not sure why! This just isn't my day!

Does anyone have any advice? Be much appreciated!

Hi Smiler ! I can't help you sadly, but I would certainly appreciate your help ? I am about to start my DBA within 4 months. I am not good in statistics actually that was one of the subjects that I barley passed. What sort of statistically knowledge you think I need ? I asked one of the DBA schools they said don't worry its only basic statistics!. What do you think, is a DBA just not for me :-(

Have you asked your sup or is there anyone at your uni that can help?

Hi Smiler,

I think you are right to look at the % in row rather than % in column.
The Percent in row means (for example) in group 1, 98.8% responded yes and 1.2% no. Whereas in group 3, only 53.9% responded yes and 40.7% responded no.
The percent in column means that of the people who responded no, 1.2 % were in group 1, 2.3% were in group 2 ..... 25.5% were in group 6. This statistic is less meaningful because because the numbers in each group are different.

Hope this makes sense.

CB

One final point, are the numbers/ percentages for group 1 correct? as they don't add up, 1/168*100 = 0.6% rather than 1.2. Should the count saying no be 2? this may be a typo in putting the data here.

Hi Obahamdoon, thanks for your comment. Yeah, statistics isn't my strongest point either. My statistics knowledge is at a very basic level, undergraduate really, and I'm not sure that it's decent enough, but I'll work this up. I definitely think that a DBA is for you if that's what you want to do, just make sure you know what you'll be doing during your DBA... stats, other research methods etc.

Hi Wowzers, unfortunately my supervisors are not so hot on stats. My dept is small and the person who was great on stats, took voluntary redundancy a few years back, and wasn't replaced.

Hi Catalinbond, thanks for the advice, that is very helpful. In what ways do mean this statistic is less meaningful because the numbers in each group are different? Also, you are very correct, it should be 2 instead of 1, I was typing this out in a bit of a panicked rush!

Thanks all :-)

"Hi Catalinbond, thanks for the advice, that is very helpful. In what ways do mean this statistic is less meaningful because the numbers in each group are different?"

What I mean is that groups with less people will always contribute a smaller percentage within the column.

To give an example, in group 2 173/177 (97.7%) said yes and they contribute 39.8% within the group of people who said yes. Now if in a smaller group had the same percentage within in the group saying yes (e.g. 43/44, 97.7%) they would only ever contribute a quarter as much to the column (or group of people who said yes overall) as there were fewer in that group in the first place.

Hopefully that is clear.