Hi all, I have conducted a research to see the cross-cultural differences for gender bias in recruitment. Data is collected through Qualtrics. Participants were randomly allocated to see either a company scenario going successful or going bad - and they selected a candidate out of two options. And they completed another likert scale survey about mindfulness. Then they answered the questions about the company's future in Likert scale.
My hypothesis was there is going to be a difference between UK and Canada population and gender bias will be correlated with mindfulness scores (i.e. those who chose male for successful company will be less mindful) irrespective of country of residence.
Here is how all things measured:
variable1: residence, 2 levels (UK, Canada)
variable2: gender, 2 levels (male, female)
--randomization happened here--- half of the participants read a company history going successful, the other half read a company history going bad.
variable3: candidate choice (female candidate, male candidate) - both groups saw the same candidates.
variable4: question asked "the company will be able to solve their problems in the future" - measured with Likert scale with 7 options
variable5: question asked "the company will be successful in 10 years" measured with Likert scale 1 being bankrupted, 10 being one of the most successful companies.
variable6: mindfulness, measured with Likert scale and scored as means of the scale (1-7).
I believe I have to run multinomial logistic regression as I have 3 IVs and categorical DV. But not sure if I can include mindfulness scale in this as it is an ordinal measurement - how should I go with the analysis?
Many thanks in advance!
For variable 3 (candidate choice) you can run a logistic regression with residence, gender and company history (success/failure) as independent variables (+ their interactions) and male or female as the binary dependent variable.
See: https://statistics.laerd.com/spss-tutorials/binomial-logistic-regression-using-spss-statistics.php#procedure
For variables 4,5, and 6 you can a Residence X Gender X CompHistory ANOVA. But, as you observed, the scale is ordinal (Likert type). Yet, many researchers use these scales as if they were continuous. To my knowledge there are no non-parametric 3-way factorial procedures available. You could try multiple regressions. You might consider to run a PCA (principal component analysis) to see whether there is a common factor. If so, you could run an ANOVA on factor scores.
You can dichotomize the mindfulness scale: find a cutoff value that divides it into "low mindfulness" and "high mindfulness" for example. You may have to experiment to find a meaningful cutoff score. Alternatively you can use the mindfulness scale as-is in logistic regression, just as you can for age.
With all these predictor variables, make sure you have a large enough sample size to do logistic regression. Ten to 20 cases for each predictor variable is recommended.
If you have random allocation, I'd recommend M/AN/COVAS as that's what these analyses were originally purposed for (experimental designs).
Hope this helps,
Matt
Refer the following
Article An empirical assessment of Lean Six Sigma Awareness in manuf...
Have a look on this video on Choosing The Best Regression Model at https://youtu.be/i8tjLQUPc8Y
Thank you everyone for the helpful answers!
Matthew Kerry could you expand what you mean?
Have a look
Book Review: Using statistical methods in social work practice: A complete SPSS guide
Begun, A.
https://booksc.xyz/book/39007549/9d3ad0