Are SEM / HLM required or regularly offered in your program?

[roubs]

---

Members don't see this ad.

The questions about stats consulting and where our field is going made me wonder what the level of SEM/HLM training is in various programs.

It seems like they may not make it into the APA requirements for stats training but anyone planning on a research career or even anyone who wants to be a future stats guru / consultant in an AMC should have a good understanding of how to conceptualize/run/write up multi-level and SEM.

As far as options for where your program stands I was thinking

1. Required

2. Electives regularly offered and nearly everyone takes them -- meaning to capture the situation where students are strongly encouraged to get formal course work in SEM/HLM and nearly everyone follows the advice.

3. Electives regularly offered; a minority of students take advantage of them

4. Electives offered very rarely or not at all

My program would probably fall into group 3.

 

[Pragma]

I voted "required" for my grad program, because you had to take some advanced stats (of which SEM was an option) beyond the basic statistics courses. But I will note that some people fulfilled their requirement with a different advanced stats course. I'd say a minority of students took advantage of all advanced stats courses offered (i.e., took more of them as electives).

 

[AcronymAllergy]

Mine is pretty much akin to a 3. No HLM classes were offered that I know of, but I took SEM along with about 10 other folks, and found it to be a great class. I'm hoping they'll keep up with that, and will start to offer some additional advanced stats electives (although I won't be around to see any of them).

 

[roubs]

I voted "required" for my grad program, because you had to take some advanced stats (of which SEM was an option) beyond the basic statistics courses. But I will note that some people fulfilled their requirement with a different advanced stats course. I'd say a minority of students took advantage of all advanced stats courses offered (i.e., took more of them as electives).

Was the basic stats sequence 2 or 3 semesters?

 

[Pragma]

Was the basic stats sequence 2 or 3 semesters?

Just two.

We also didn't have an HLM class at the time i was there, but I hear they have one now. I learned HLM from a stats consultant on a grant.

 

[cara susanna]

Mine is "4," but keep in mind that I'm not at a research-heavy program. We actually did have an elective offered through another dept, but it was more of a workshop that met twice. I'm working on a study that is using SEM and I pretty much had to figure it out on my own (although profs did give me some assistance).

 

[Ollie123]

I voted #2, though the situation is similar to Pragma's. We're required to take four stats classes...one is the basics (Regression/ANOVA), one is termed multivariate but is pretty broad and includes some discussion of SEM/HLM, and one is psychometrics. The final is elective, but has to be either one of the stats classes, or grant-writing.

SEM and HLM are both offered, and are very popular choices. Other commonly chosen options are meta-analysis, factor analysis, IRT, survival analysis, an ANOVA-specific class, and they just started offering a two-course sequence in Bayesian approaches that I am absolutely giddy about. Not all are through the psych department though. A lot of students will take workshops or self-teach even if they don't take a course. I've recently had to self-teach GEE, though after having taken HLM it wasn't that bad.

A few of us go well beyond the required 4 though...I should be up to 8 or 9 by the time I'm done, which is not unheard of. I agree that SEM and HLM are both extremely useful... have yet to take SEM actually, though am at a point I could probably self-teach if needed (though may take it anyways depending on how long I stick around). Of course, I also think its frequently used as "turd-polish" and is probably the one I see getting the most abuse in the literature Of all of them though, I'd actually like to see meta-analysis become required (or at least heavily discussed in more general stats courses). I think its going to become more and more necessary to know regardless of whether one is applied or research-focused.

The major issue I've found is that these courses tend not to be taught by folks who do the sort of work that I do. There is a lot of nuance to HLM when running it on tightly controlled experimental data (commonly with smaller n's) rather than typical longitudinal studies, so I've found it difficult to get support. While I generally consider myself a careful, competent statistician, I am always afraid I'm going to make an error no one will know enough to catch, and journals certainly don't look at things in that level of detail. Doesn't help that many of the folks doing this work were born and raised on repeated measures ANOVA and know almost nothing about other techniques. I'm hoping to find get a post-doc that will allow me to work with the biostats department, since stats is definitely one of those things where the more you learn, the more you realize you don't know.

 

[roubs]

Just two.

We also didn't have an HLM class at the time i was there, but I hear they have one now. I learned HLM from a stats consultant on a grant.

Ah, our required sequence is 3, third one being multivariate which I guess I thought was required everywhere. My program is balanced with probably only a minority of students going into academia so I would have thought we were on the low end of stats requirements.

Given the discussion/issues in the psychology future directions thread I think it's kinda short sighted for individuals to decline advanced stats training.

 

[Pragma]

Ah, our required sequence is 3, third one being multivariate which I guess I thought was required everywhere. My program is balanced with probably only a minority of students going into academia so I would have thought we were on the low end of stats requirements.

Given the discussion/issues in the psychology future directions thread I think it's kinda short sighted for individuals to decline advanced stats training.

Yeah in my program I thought it was maybe lighter than was needed, although there was a heavy emphasis on learning stats in the research lab.

In my case, I took the two basic semesters, then added on SEM and Multivariate courses. HLM was learned independently. As I mentioned though, it sounds like they added some options. There was also a workshop in Kansas (I think) that was popular.

It doesn't sound like there were nearly as many options as Ollie's program.

 

[Pragma]

I have a related question. How common is it for other research related coursework to be required?

I ask because the 3 required semesters of stats in my program does seem light. Many of us took more as electives, but practice oriented students usually took other electives. However, we had some other requirements. For example, we all had to take a grantwriting course.

 

[AcronymAllergy]

I have a related question. How common is it for other research related coursework to be required?

I ask because the 3 required semesters of stats in my program does seem light. Many of us took more as electives, but practice oriented students usually took other electives. However, we had some other requirements. For example, we all had to take a grantwriting course.

That's actually even more than was required for us, although by "stats classes," I'm specifically talking ONLY about statistics (i.e., courses with the word "statistics" in the title). Psychometrics and the like were covered in our measurement class. Beyond those two, though, there wasn't anything required, although most students would also take advanced/multivariate stats. Everything else was essentially learned through self-study.

There were many very strong areas in my program, but I'd say stats was probably one of the weaker portions, at least as far as formalized learning went. Part of that was probably just related to actually finding the people to teach it. And there was always a stats department if anyone REALLY wanted to learn a whole lot more.

 

[Pragma]

That's actually even more than was required for us, although by "stats classes," I'm specifically talking ONLY about statistics (i.e., courses with the word "statistics" in the title). Psychometrics and the like were covered in our measurement class. Beyond those two, though, there wasn't anything required, although most students would also take advanced/multivariate stats. Everything else was essentially learned through self-study.

There were many very strong areas in my program, but I'd say stats was probably one of the weaker portions, at least as far as formalized learning went. Part of that was probably just related to actually finding the people to teach it. And there was always a stats department if anyone REALLY wanted to learn a whole lot more.

Good point, although I don't know a lot of people that went to the stats dept.

I think this is an important thread because stats is one of those things where I think the coursework DOES matter. I can't imagine having been able to learn multivariate or sem on my own. However, I did benefit substantially in my lab from having a stats consultant on staff for grants. Part of their job was to teach us about tricky methods, and I still am very grateful to my mentor for prioritizing this.

People I knew who got F grants also did attend workshops because they were paid for. I wish I had tried for this funding, in retrospect.

 

[paramour]

I voted "required" for my grad program, because you had to take some advanced stats (of which SEM was an option) beyond the basic statistics courses. But I will note that some people fulfilled their requirement with a different advanced stats course. I'd say a minority of students took advantage of all advanced stats courses offered (i.e., took more of them as electives).

This is the way our program works. We have two basic stats courses everyone is required to take and then we're required to take at least two "advanced" stats classes above & beyond. Quite a few folks end up taking SEM as one of the electives... There's another class that quite a few folks take that also has some brief exposure to SEM/HLM near the end, so some folks get double exposure if they take both (or if they don't take the full SEM, they still get a brief dose of it...). Oooor, there's usually a SEM/HLM seminar/workshop that's periodically offered that some folks will take. The HLM class is not offered as frequently...

 

[FadedC]

I'm at 4, although I'm school psych, and it's possible that I might be able to take a course as an elective alongside people in the clinical program.

It's actually been a bit of a problem in my research group, because we've needed to do HLM and lacked sufficient experience/training. Although with enough time, effort, books, and consultation, we were able to perform the necesary analysis.

 

[PsychPhDone]

We have a quant psych program so all our stats are through them. We have a year of required stats + 1 elective. HLM/SEM are popular for that.

 

[IOPsych]

I chose option 3.

I took HLM, but it was offered as a seminar.

Our basic stats sequence is 5 classes.

Advanced Methods I
Advanced Methods II
Advanced Methods III
Psychometrics
Applied Research Methods

and I took a seminar on HLM.

 

[futureapppsy2]

At my old program, we had to tale a two semester stats sequence (in addition research methods and advanced psychometrics) and then 2-3 advanced stats classes and could choose from HLM, SEM, meta-analysis, multivariate,advanced SEM, or longitudinal data analysis. We could also take single-case (which was pretty stats heavy, surprisingly) or qualitative courses in addition to those courses. Yeah... I'm going to miss the stats training here. 🙁