Big Five Personality in 2005 of BHPS and Wave 3 of UKHLS
I have two questions regarding the Big5 personality traits in the BHPS and Understanding Society:
1. I am currently using all waves from BHPS and Understanding Society to understand the impact of personality (the Big Five) on labour market outcomes by occupation. I am aware that the Big Five were asked in wave 14 of BHPS and wave 3 of Understanding Society. I am now unsure on how to merge those two variables. Given that personality is often assumed to be constant across years I would like to extend this measurement to more than just the year of the wave they were asked. However, I am unsure as to how to treat those values for BHPS versus Understanding Society and how they differ across waves by individual. Is there a best practice?
2. Derived personality variables from wave 3 of UKHLS: I have tried to replicate those with the non-derived individual responses (following the description on the Understanding Society) but I do not get the same but similar results. It would be important to know how those are coded for me to do the same for the 2005 responses. I have followed the instruction on your website: "Component score calculated as the average item response if no more than one of the three input responses is missing". And then rounded those values. But my derived variable is different from yours.
#1 Updated by Stephanie Auty about 2 months ago
- Private changed from Yes to No
- % Done changed from 0 to 50
- Assignee set to C Josten
- Status changed from New to In Progress
- Category set to Derived variables
Dear C Josten,
1. We are not able to advise on this and suggest looking at previous research using personality measures. If you would like to discuss it with other researchers who have used our data, you could join our jiscmail list, by emailing UKHLS-REQUEST@JISCMAIL.AC.UK
2. The negatively worded components of a positively worded attribute (and vice versa) will need to be reverse coded, that is scptrt5a1 (rude) scptrt5c2 (lazy), scptrt5e3 (reserved) and scptrt5n3 (relaxed). Then, if one value of a component of an attribute is missng but the others are not, it is imputed using the mean of the two that are present. After that the mean for each attribute is calculated for those with three non-missing components and rounded. If you are still getting different results then please let us know.