I am trying to assess various factors affecting the wellbeing of communities living around a protected area in Africa. I developed a set of eight wellbeing indicators for this assessment. I conducted a household survey and my sample comprised of more males (n=309, 84%) and fewer females (n=58, 16%).
In my analysis, I have included a number of factors including gender and age of respondent, employment, education, household size, receiving benefits from conservation and experiencing conflict with wildlife. My results indicate that education, employment and age of respondents influenced different wellbeing indicators. However, gender did not appear to influence any of the wellbeing indicators.
Could it be possible that my unequal numbers of male and female respondents affected the gender variable? Any literature to support or dispute this thinking?
Thanks
If the proportion of male to female in the population is the same as in your sample (i.e. 309:58), then no. Otherwise, you will have to correct the bias by weighting your data to reflect the actual population ratio, which usually is 50:50 :)
Unequally sized gender groups can lead to unequal variances, a general loss of power and issues with confounding variables. Gender in your study still may have confounding effect, despite of the fact that at the moment you didn’t found any influence of gender on wellbeing indicators.
I agree with Andrew that one of the solutions is to correct the bias by weighting your data to reflect the 50:50 population ratio.
Another commonly used strategy is to analyse your data by using some statistical tests that are set up specifically to deal with this type of problem (unequal sample sizes).
I would recommend reading:
http://www.statisticshowto.com/unequal-sample-sizes/
Yes I think. Even within families, the response to questions vary significantly based on gender. So the balancing will be required.
Thank you Andrew, Sabina, Dimitar and Emmanuel for such great comments. I now have some ideas on how to proceed.
If you got these results, why do you worry unless you have reason to worry. If you have carefully designed the study and done apriori sample size calculation and standardized your wellbeing measurement scale (preferably the one already published in indexed papers). The only thing you can do is to perform statistics on the data. If these suggest that gender does not have an effect, you have to agree unless there is a compelling reason to suggest otherwise.
The whole purpose of your research was to know what is the case not to verify what your initial gut feeling was. You will have to respect the data that you have generated.
Just write the manuscript and communicate it in some good journal. All the best.
Hi all,
I have the same question too. My respondents were 75% males and 25% females. and gender is the moderating variable. based on the findings, gender does not moderate the relationship. does this finding may due to the unequal size of gender groups?
Thank you.
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude