I am conducting mediation analysis where M mediate between X and Y and gender is entered as covariate using PROCESS SPSS. There is difference in the results of Mediation analysis in terms of negative and positive signs of coeff of covariate in two ways of coding of the variable "gender"( 0."Female" and 1 "Male") and (1="Male" and 2 "Female"). This difference in the signs could cause any difference in interpretation? and what the values for coding should be adopted in this case.
If you'd like, you can code female as 100 and male as -7 and you'll still interpret your results the same in the end! The significance and standardized beta will always be the same (except +/-). Consider another example: let's say you use height to predict temperature. Whether you measure meters & Celsius, feet & Fahrenheit, or cubits & Kelvin, you'd have to get the exact same correlation and significance. If not, something's very wrong with stats! Standardized beta is a correlation. In your case, the correlation is positive or negative, depending on which gender you code as higher. The best-fit line always goes "up" from male to female (or female to male) and that's either a positive or negative slope depending on which gender is coded with a bigger number. Hope this helps some, Sara! ~ Katie
- Badges
- Science topic
- Similar topics
- Mathematics