To produce the output you uploaded, you clicked on Analyze > Compare Means > Means, and then selected Options and checked a box for Test for linearity. (You left out that final Means about.)

For the analysis you posted, Exam Mark was treated as a categorical variable, or more precisely, as a set of ordered categories, and (polynomial) trend analysis was performed. The Linearity line shows the test for a linear trend, and the Deviation from linearity line shows a test on all other (nonlinear) trends combined. (The combined df = 3, which means you have 4 Mark groups. So the deviation from linearity term will test the combined quadratic and cubic trends.) Linear regression, on the other hand, would treat Marks as a quantitative variable.

So, the question for you is, how were marks measured? Do you have percentage grades (or some other quantitative measure), or do you have only a set of 4 ordered categories? That will be the main determinant of which approach you use, IMO. HTH.

Bruce Weaver Thanks for your reply. Yeah, I forgot the Means part 🙂

My independent variable (Depression) is categorical (4 categories）and my dependent variable is Marks (out of 100). As you can see from the table I shared, Sig. value of 0.14 is for Linearity and 0.866 for deviation from Linearity. Can you guide me how to interpret these results?

Before we get to interpretation, here is an example I generated. It is structurally the same as yours in that the Group variable has 4 levels. Here is the code:

* Example similar to the one Rehan posted.

NEW FILE.

DATASET CLOSE ALL.

* Modify the path below to point to the folder

* with the sample datasets that come with SPSS.

GET FILE = "C:/SPSSdata/survey_sample.sav".

FREQUENCIES childcat.

* Variable childcat has 4 categories, so use it

* to mimic Rehan's 4 depression categories.

* Use age as the DV.

MEANS TABLES=age BY childcat

/CELLS=MEAN COUNT STDDEV

/STATISTICS ANOVA LINEARITY.

* Notice that the output shows Age * Childcat in

* the ANOVA table--i.e., DV * Group Variable.

* Rehan's output shows Depression * Marks, which

* indicates that Depression was treated as a

* quantitative DV and Marks as a set of 4 ordered

* categories.

As you can see in my comment lines, your output shows Depression * Exam Marks as the title in the ANOVA table. And that your MEANS command treated Depression as the quantitative DV and Exam Marks as a set of 4 ordered groups. I.e., your command looked something like this:

MEANS TABLES=Depression BY ExamMark

/CELLS=MEAN COUNT STDDEV

/STATISTICS ANOVA LINEARITY.

So, I think you need to check again what you said before. It looks to me as if depression is a quantitative variable and exam mark is a set of ordered categories (e.g., 4 letter grades, A-D, perhaps).

Rehan Qayyum, I just noticed that the same question was posted here:

• https://www.researchgate.net/post/What_is_the_difference_between_linearity_test_and_linear_regression_test2

As no one has responded there yet, I suggest you delete that version of the question. Thanks.

Bruce Weaver . Thank you very much for your prompt response. I just checked the table I uploaded and I observed a mistake. The table I shared was a screenshot from a word file. I copied the table from SPSS to word and mistakenly typed Depression* Exam Marks instead of Exam Marks* Depression. Sorry my bad 😞.

The Marks are from 0 to 100 and Depression is 4 levels (Very Low，Low， Moderate, High)

I am sharing the new screenshots from SPSS.

Thanks once again. Really appreciate your help.

Bruce Weaver . I have deleted the repeated question from here:

https://www.researchgate.net/post/What_is_the_difference_between_linearity_test_and_linear_regression_test2

Thanks

Hello Rehan Qayyum. Your new output shows that the overall effect of Depression (the ordered set of 4 groups) is not statistically significant (p = .478). It also shows that 88.54% of the between-groups variability (471.353*100/532.370=88.54) is due to a linear trend; and 11.46% (61.017*100/532.370= 11.46) is due to the quadratic and cubic trends combined. But neither of those tests is statistically significant either. So apart from speculating about why the expected results did not materialize, I don't think you have much to talk about.

Bruce Weaver Thank you very much. Just one last question. I conducted the same test to compare the overall effect Depression (the ordered set of 4 groups) with another variable Exam2 . I have attached the test result. The results are significant. Can you guide me how to write the interpretation for that result? Also how can I compare the interpretation for Depression*Exam and Depression*Exam2 ?

Thank you

I don't really want to write a discussion of the new results, because it would feel as if I was doing your homework. But here is an outline based on what I posted for the previous analysis. You can carry out the needed calculations (of percentages) and fill in the relevant statistics.

• The overall effect of Depression (the ordered set of 4 groups) is or is not statistically significant (report the relevant statistics).
• The test for a linear trend accounts for X% of the variability between depression groups. It is or is not statistically significant (report the relevant statistics).
• The test for all nonlinear trends combined accounts for X% of the variability between depression groups. It is or is not statistically significant (report the relevant statistics).

You also asked:

Also how can I compare the interpretation for Depression*Exam and Depression*Exam2 ?

I'm not entirely sure what you mean. Are you thinking of a quantitative comparison, or a qualitative comparison? Maybe someone else will understand what you mean.