The data is likert and i want to apply the logistic regression. How should I enter the data in spss?
Attaching the questionnaires which i used for collecting the data.
(QOL= quality of life)
Well, you should have a row for each participant and a column for each question. In each cell within those columns would be the score for a participant on that question.
What is your outcome variable? What are your predictors?
~Michael
well I am not sure about the outcome variables and predictors. Its just my initial step. I've entered the data like in the attachment.
Logistic regression uses 0 or 1. so in here 1 means diabetic, 0 non diabetic. so how should I proceed?
There are a number of questions:
1. Is this questionnaire validated? If you are computing total score than you need a valid and reliable tool.
2. I am assuming your dependent variable is QoL. How are you determining that using the tool? What is your cut-off score to group either as poor or good QoL?
3. What are your independent variables?
4. Sample size?
Hi Hamna,
Three comments.
1) If you're interested in comparing QOL between diabetics and non-diabetics, then your outcome(s) are one or more QOL measures. If these QOL measures are based on a published, validated scale, then you should be able to locate the original validation and scoring information online. For some very basic, but helpful reading on Likert scales, see http://www.simplypsychology.org/likert-scale.html
2) In terms of data entry, try setting up your data in an Excel spreasheet, something like this, where the first row has the variable names (column headers), and each line has data for one person:
ID Diab Var1 Var2 Var3 (etc)
Use numeric coding (e.g., 0/1 instead of No/Yes) whenever possible.
SPSS can import data from Excel; it's painless.
3) Logistic regression models can handle dependent variables on the (0,1) (binary) scale, or an ordinal scale (often, 0, 1, 2, 3,..., but may be something else wherein the order has meaning). Conceptually, you could use a QOL measure as your dependent outcome in a logistic analysis. It depends. Do you want to report on the odds of being diabetic, given a certain level of QOL? Or do you want to report on the odds of having higher or lower levels of QOL, given diabetes status?
good luck.
Your outcome should be QOL and you can use ordinal regression. So you will be comparing the odds of increasing a level in the ordinal scale for QOL (your Likert scale) between diabetic and non diabetic patients.
You measure your items/statement using an interval scale (not an ordinal or dichotomous scale).
Obtain data from your respondents (minimum of 100 respondents), enter data into Spss file. In "Analyze", choose "dimension reduction" and enter items score for that particular construct. Tick the appropriate box and execute. Spss output will give you the dimensions where the items are segregated together with factor loading. The items with low factor loading (normally less than 0.6) should be eliminated. Thus you will get the reduced number of items to work further.
Remember for Logistic Regression, the DV is categorized into two possibilities such as "Success" or "Failure" while your IV is not necessarily categorical.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS