I used the Marital Conflict Scale developed by the National Longitudinal Survey of Youth in the United States of America (Center for Human Resources Research, 2000) in order to measure marital conflict in my study. It is including 16 items (plus the items I added based on my objectives) on 4-point Likert scale from 1=never to 5=Often.
Would you please guide me how I can categorize the level of marital conflict among the respondents to low, moderate, and high levels? I could not find its categorization in the original file and the other papers which used this measurement.
First, you have to compute all the response gived from your participants generating a new variable called "marital conflict". In SPSS: Transform, Compute variable, target variable (marital conflict), item1+item2+...item16, Ok.
Second, you have to calculate Percentils for the new variable called Marital conflict. In SPSS: Analyze, Descriptive, Frequency, introduce marital conflict, statistics, Percentile 33 and Add, Percentile 66 and Add, Ok.
Third, values corresponding to percentile 33 and lower are your LOW LEVEL participants; between 34 and 66 are MODERATE LEVEL; and 67 and higher are HIGH LEVEL.
Fourth, you could generate a new variable called LEVEL: select participant with values corresponding to Low level and added it a number o word, for example 1 or low; then select participant with values corresponding to Moderate level and added it a number o word, for example 2 or moderate; select participant with values corresponding to high level and added it a number o word, for example 3 or high.
Doing this you wil can to manage directly each level or group of Marital conflict.
I helped you?
Best regards
Dear Dr. Javier Sánchez Rosas,
Thank you for your complete response. Yes, sure. It is very helpful.
Regarding to your explanation, the levels will be: between 16 & 27.68 (low), 27.69 & 35.04 (Moderate), and 35.05 & 64 (high).
So, for variables such as adult attachment that should be categorized in two levels (low & high) percentile should be 50. Am I right?
Would you please give me its reference?
Thank you in advance,
Dear F.,
16-28 (low), 29-35 (moderate), 36-64 (high).
Low and high could be splited by the percentile 50. Also, for any reason, you can decide to divide in three levels to Adult attachment and carrie out your analysis with the extreme, lower and higher groups calling them low and high.
Reference for this procedure or anything else related to me?
Thank you
Dear Dr.
Once again, thank you very much.
So, this method can be used for odd or even Likert scales. Am I right?
I mean, the reference that should be cited for using method in my thesis.
Best regards,
This method can be used in both cases.
I dont know references for this habitual practice. You could say: "The global score of marital conflilct was splited or percentil 33 and 66 was obtained to conform the .... groups. This procedure is tipycal in research who works with ... populations (and insert here references to papers where you saw this practices) because allows...
Hi Farimah,
Just another aspect of your question: why categorization at all?
Basically, if you have a continous variable (like marital conflict score or attachment) the principle is 'leave it as it is' and use appropriate statistical procedures, like correlation, regression, SEM, etc.
Categorization seems to make to interpret things easier but it also loose power, complexity, etc.
Even curvilinear models can be built in this way, if this is the concern.
Of course, there might be other aspects but I wanted to provide another frame. And feel free to choose that approach that best fit in your study.
Regards,
Dear Farimah,
If you want to know how to use your groups of low, moderate and high levels you can look at my hysterectomy paper where I used different levels of perioperative depression, anxiety, hostility and sexuality.
Dear Dr. Martos
Thank you for your attention. Yes, you are right. I am not going to analyse the data base on the categorizations. As I want to provide descriptive information about my sample in order to cover my first objective, I guess it can be good if I I categorize the variables in 2 or three levels. However, to understand the attachment styles among the sample (secure, fearful, ...), I guess I have to know at which level it is high or low.
Thank you,
Best regards,
Dear Dr. Ewalds-Kvist
Thank you for your reply. I
Please check your private message.
Thank you,
Best regards,
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis