Along with gender and race, I am using summated scores from the Climate Change Anxiety Scale and the ASES (academic self-efficacy inventory) all as predictor variables. My dependent variable is YES/NO to choosing a Natural Science major.
Yes, it is acceptable to use summated Likert scaling scores as continuous data in a logistic regression. However, it is important to keep in mind that when using Likert scales, the numeric values assigned to the response categories are arbitrary and do not necessarily represent a linear increase in the underlying construct. Therefore, it would be important to analyze the data carefully to ensure that the results are valid and meaningful.
No, Likert scores, even summated, are not continuous data :)
As Peter Donkor opined, output scores from Likert scale can be used as continuous data . Transformation of
a data set to scores gives it random characteristics. That is why data scores from PCA can be used to create new variables which are independent of each other.
Harold Chike
Thank you for the response, Harold. What do you mean by PCA? Are my summated scores the new variable?
Harold Chike
I expect to run the logistic regression in SPSS. So I assume that I can still treat the summated scores as an ordinal measure. Is Peter suggesting that I examine the distribution of summated scores for some measure of normality? I don't understand what Peter means by a linear increase. Of course, the distribution of summated scores may be non-normal, depending on the responses to anxiety levels from the Climate Change Anxiety Scaling and the scaling from the self-efficacy inventory.
@Hamilton, PCA is the abbreviation for an analysis called principal component analysis in statistical analyses. Do not worry about if you don't yet know about it.
Your used summated scores as input data in your analysis. This not the same as the output scores after your analysis.
Harold Chike
Thank you sir. Yes, the summated scores, along with gender and race, are predictor variables to be entered into the equation.
In effect, @Hamilton, your raw summated scores should not be used as they are in nominal data form for regression analysis. They need to be transformed to interval data.
Harold Chike
How do I transform raw summated scores into interval data?
With that said, Tabachnick & Fidell (Using multivariate statistics, 4th ed.) state that predictor variables can be continuous, discrete, dichotomous or mixed...and that the predictors do not have to be normally distributed, linearly related, or of equal variance. I would think that summated scores from Likert scaling responses would be deemed viable under these conditions. Ordinal data are discrete data...and almost all studies that I have read accept summated scores as much more valid and reliable than a single score from one item. When we summate Likert scaling scores, we are already treating the data as interval.
Any answer for the first question and any clarifications would be appreciated.
@Hamilton, your problem is that you want you want to run logistic regression using the SPSS package
However, logistics regression is not yet available in SPSS. But it is available in the R statistical package.
Try and have access to the R packages.
Transformation of.data to interval level can only be done using principal component analysis (PCA)
with which you are not familiar. That is only relevant for the SPSS package.
So consult your supervisor for advice on some appled statistical texts.
Harold Chike
> ANALYZE > REGRESSION > BINARY LOGISTIC > LOGISTIC REGRESSION
This is possible with SPSS version 27.0.
- Badges
- Science topic
- Similar topics
- Filing