SEM Series Part 4: Convert Continuous to Categorical

[@Gaskination]

Hi Rana. The way to do it is to either create a composite (see imputing composites video) and then split it along the median (see splitting continuous variable into high and low values video). Or you can just create an index (sum or average) of all the indicator response values, and then split that along the median.

[@drranaayyub]

Dear James, Your tutorials are so nice that one can learn these essential skills from these alone. If you would be able to add such a tutorial for guiding such moderation by Construct having factors so that will be great. GOD bless you. Thanks again for your help all the times.......

[@drranaayyub]

Dear Dr. James, Would you please guide me about a reference regarding doing multigroup moderation after converting continuous moderators to categorical moderators and then splitting it into high and low groups based on median.I need it to defend this kind of moderation in my thesis. Thanks in advance.

[@drranaayyub]

Dear James, Thank you you are great. I will try these videos to learn it. Thanks again for your guidance. GOD bless you.

[@Met201101M]

I wonder if you have any video explaining how to enter and analyze categorical variables (i.e. control vs treatment groups) in AMOS. Thanks

[@Gaskination]

You might take a look at logistic regression for using categoricals in regressions. You can look at ANOVA for comparing means across control and treatment groups.

[@PhilippBrauner]

I usually use the Transform -> Rank function. If you select "N-Percentiles = 2" you automatically get a new variable with the lower half and the upper half of your cases.

[@drranaayyub]

Dear James Thank you very much for adding such a nice tutorial. As you know i have been learning SEM (AMOS) by following your tutorials. Would you please guide me that how we can convert our construct (having 6 factors) to a categorical variable? I mean if i have a Latent construct of acculturation (ACC) which has 6 variables i.e. ACC1, ACC2, ACC3,ACC4,ACC5 and ACC6 and all of these have been measured on 7 points Likert scale. So how i can convert this to see moderation effect in AMOS.

[@sowmya310]

How do I convert a continuous variable into categorical, if the variable has multiple items measuring it?

[@Gaskination]

You could first average them. Or, you could do an EFA with those items and then save factor scores as variables (in the scores option in EFA window). Then split the resulting variable.

[@ath1na]

Hi there, I need to split a variable measured on a 7-point scale on high and low based on the median in order to run two different regressions (one for the cases that are high and one for those who are low). I am not sure what the right way is. Moreover, how could I get two different files (one including all the cases for high and another one for low)? Could you please help me out?

[@Gaskination]

The easiest way is to use the Rank Cases function in SPSS. This will split it into two even groups. If you are using SPSS or AMOS for the regressions, then you don't need two separate files. The Rank Cases will create a new variable with the group membership (e.g., 1 for low and 2 for high). You can then use this as a moderating variable in AMOS or as a splitting or factoring variable in SPSS.

[@ath1na]

James Gaskin Would it be possible please to describe the exact process and steps I have to take?My understanding about this process is that I have to use the mean scores and not the median ones. I need to split my sample based on the median scores and the scales are 7-point ones. I am using SPSS. Thank you very much for your help and reply!

[@RitaRita-le2or]

Hi James, thanks for the useful tutorials. I wonder the method you used in this tutorial is "median split"? if so, what about those whose numbers are exactly the median value? they don't belong to the low or high group, what do we do with them?

[@Gaskination]

You can choose which group the median is assigned to. You can also use this simple procedure: kzbin.info/www/bejne/bH6ug4Wjp9KgpMk

[@drranaayyub]

Dear Dr. thanks for your reply,Yes you are right it makes sense but to cite a particular reference i need it, OK, so it means that i should also try to search it from google scholar...Then i will be coming back to you.Thanks again for your help as usual.

[@danjonathan2763]

Hi James! Im working on my MBA thesis these days, I have one major problem that is hard to solve! In my model, I have one "binary predictor"(negative/positive), and my dependent variable is continuous. Can I use Amos to run SEM for this model? or Amos with Bayesian Estimation? I try to use 0 and 1 to my independent variable(0=positive,1=negative), but I think it's illogical. Could you help me to solve my problem? Thank you so so so so so much!!!

[@Gaskination]

If you have only this one IV, then it makes more sense to use t test or anova, where the factoring/grouping variable is the IV.

[@danjonathan2763]

Ok! thank you so much!

[@sarikajain744]

Hello sir, I request you to help me out in how to control age and gender(demographic variables) in AMOS. These variables are showing significant correlation on my 3 exogenous variables. However I don't know how to control their effects in AMOS.

[@Gaskination]

+Sarika jain You can include both just like you would a regular IV. Gender is binary, so it will just act like a dummy variable. Age is either ordinal or continuous, so it can be treated like any other variable. Just include regression lines from these to the DVs, and covariances to the IVs.

[@thaermajali1514]

hello sir, first I just want to ask if I have latent variable as moderater with five points scale how I can make it categorical like high & low e.g personality traits. second is it enough to analyze the to set of data and compare the results or I have to create moderate effects option

[@Gaskination]

+Thaer Majali Moderation is a theory driven test. If your theory includes moderation, then you must test for it, otherwise no need. Here is a video showing how to convert a variable into high/low: kzbin.info/www/bejne/bH6ug4Wjp9KgpMk

[@renubhardwaj7506]

hello sir, thank you for such a nice tutorial..... can u please help me in knowing whether the analysis would be same if endogenous variable is categorical

[@Gaskination]

If the variable is already categorical, then you don't need to do this, because it is already split in groups.

[@Gaskination]

slick! I'll be using that from now on. Thanks!

[@Gaskination]

Just makes sense... sorry that's not a good answer. I'm sure there is precedent out there (but nothing comes to mind). I searched for the following on google scholar, and found many likely candidates: moderation, standard deviation below above mean, hi low (just typed all that in, no quotes).

[@Gaskination]

ath1na Here is a new video explaining how to do this: kzbin.info/www/bejne/bH6ug4Wjp9KgpMk