Hi everyone,
I'm a newbie here and to writing researches as well. I'm currently engaged in writing my first research (dissertation) for academic purposes and need some help with the same.
Here is the problem:
I have two scales that I want to correlate. One is a 60-item True/False Codependency Scale with a scoring range like this: Below 20 You have Little Need for Concern 21 - 30 Moderate Need for Concern. 31 - 45 Moderate to Severe Need for Concern. 46 or over Severe Need for Concern
The other one is a 36-item 7-point Likert Attachment scale with 2 further subscales (Avoidant & Anxious) of 18 items each. The scoring of this is based on calculating the averages of the responses.
Can somebody please confirm the level of the measurement of both scales (i.e., the four categories of the 1st scale and the 2 subscales of the 2nd scale) as well as the correlation method that I can use in Spss? Although correlation is strongly preferred but if it can't be used, please advise on another method.
P.S.- My supervisor suggested they're both interval but didn't explain how. I'd like to know this too.
Thanks a lot! :)
Lavanya Bahl First, it is important to distinguish between the ordinal item rating OPTIONS and the summation of the ORDINAL item ratings into the overall rating SCALE. The rating scale ranges from lowest to highest possible TOTAL scores. Under certain conditions, that summed total score represents an INTERVAL-level score.
With this in mind, you have produced multiple scales for which the conditions described by Likert may or may not be met. The dichotomous True/False options do not adhere to a Likert item structure. The rating options for the 7-point attachment items are not shared so they may or may not adhere to Likert's level of agreement construct. You did, however, sum the scores (to get the average) which is the customary process in developing an interval level scale under the Likert conditions.
Is it possible to run correlations between columns of summed scores in SPSS? Yes, the dutiful software will do what it is commanded (within reason) and produce correlation coefficient estimates. However, it is important for you to explain your process prior to generating these estimates in order for the reader to fully appreciate the correlation estimates.
- Badges
- Science topic
- Similar topics
- Developmental Psychology
- Attachment