"Subject: Qualifications of Therapists and Outcomes
Neil Jacobson claimed that the CR study "confirmed what
we already knew" that increasing levels of experience, skill, and
education do not make for better therapy. The CR study did nothing
of the sort, nor do "we already know it" from elsewhere. This is a
mischievous claim in today's marketplace. Profit-driven health care
schemes will send patients to the cheapest provider they can find--
particularly if scholarship can be invoked to justify the idea that less
qualified providers (less experience and less education) make for
just as good outcomes.
The CR dataset was conflicting on this issue. Social workers
(presumably mainly MSW's, although degree was not queried) did
as well as doctoral level providers. Marriage counselors
(presumably mainly MFC's, MFT's, and other nondoctoral
providers--although degree was not queried), on the other hand,
did significantly worse. This was not an artifact of the fact that
marriage counselors see couples in troubled relationships. CR
compared doctoral level providers treating marital and sexual
problems to marriage counselors treating the same problems, and
the doctoral level providers did significantly better. Here are some
relevant data: The marriage counselors (n=197) seeing
respondents who checked "marital or sexual problems" had
significantly poorer global outcome than other therapists (n=753)
with these patients (p<.0017, controlling for initial severity).
The argument for equivalence derives from Christensen and
Jacobson (1993) and their argument is seriously flawed. For they
review studies, some badly flawed, in which manuals are used,
mild and uncomplicated clinical problems are the diagnosis (but the
diagnosis is made by doctoral level providers), and duration of
therapy is brief and fixed. Precisely the situations in which clinical
judgment, experience, and education matter very little. A seven-
year-old may be able to fly a one-engine plane in clear weather, but
this does not mean she can handle a 747 in a thunderstorm.
The disinterested conclusion should be:
1) If a case is simple, if a manual must be followed, if the
diagnosis is made by a more experienced and better trained
clinician, and if treatment must be very brief, less experienced and
less well-educated providers may do as well as doctoral level
specialists.
2) It seems likely that in real therapy, where cases are complicated
and more severe, no manuals are used, diagnosis as well as
therapy must be done, and clinical judgment is important, that more
education and more experience of providers will improve outcomes.
3) Effectiveness studies of level of education and experience of
providers and the cost-benefit analysis are urgently needed.
Instead MCO's use this state of affairs to justify using less
experienced and less well-trained providers even in complicated
and severe cases. Again, I believe, patients are being deprived of
adequately skilled treatment on a massive scale. Until this issue is
resolved by the appropriate effectiveness study, I recommend that
MCO's err on the conservative side and provide experienced and
highly educated providers in all but the simplest and least severe
cases."
---------------------------------------------
BRAD REPLIES (with the truth):
Dear Martin Seligman,
I do not believe that it is so much a worry that clients are deprived
the "adequately skilled" treatment (that is in some way is supposedly
associated with "highly educated providers") AS MUCH AS we should be
concerned that they are denied (or largely denied) access to treatment
or
reasonable treatment IN GENERAL NEARLY ALTOGETHER. We need a mental
health
care SYSTEM with a variety of helpers, as may well come to be seen as
rational and justified (IF ONLY THE PREEMPTIVE "WORRIES" --AKA
PROVINCIALISM AND POLITICAL POWER-- OF CLINICIANS DOES NOT GET IN THE
WAY
OF DOING THE MUCH EASY, BASIC RESEARCH STILL NEEDED HERE). Do you have
any
data to the contrary? Apparently not.
RATHER THAN "clinical judgement" being education-related, it is just
as
likely related to the good adaptation and communication of the helper
(at
this point in much of our knowledge). AND, much of this may be a product
of
good develpment and perspective-taking abilities plus RELATED abilities
to
modulate emotional reactions. SELECTION OF GOOD CANDIDATES MAY WELL BE
AS
IMPORTANT AS TRAINING FOR SEVERAL FUNCTIONS (with training necessary,
but
not of the "level" you seem to insist on). I simply do not believe that
counselors (yes counselors, doing primarily counseling and not
"therapy")
apply but the simplest principles of psychology (OFTENTIMES) in any
clear
or rational way. (Appeals to scary cases of possible suicides aside.)
--
b jesness
P.S. *STILL*, ANYHOW, TO WORK ON YOUR WAVELENGTH: Ever heard of anyone
going into physics without taking physics in college? YET, a similar
thing
happens everyday in schools of clinical psychology. One must ask: How
much is pretense?
---------------------------------------------
P.S. On the problem of "highly trained" "therapists" doing
*everything*:
I have indicated that this arguably denies reasonable access to
treatment
for many. It possibly denies the "middle class" as much as the poor,
given
fee schedules -- though the poor also do not get the comfortable easy
access, the amount, or the variety of help needed.
But another problem is that we are doing little to clarify the cases
where special interventions are necessary. HERE AGAIN PROFESSIONAL
PSYCHOLOGY IS HOLDING UP EFFORTS, hurting and not helping. While these
interventions may not be well-defined or exist as standard (or as known
or
recognized) treatments at the present time, we are making little
progress
as "high level clinicians" diffuse there efforts on VIRTUALLY ALL CASES.
Signs of this are: There is relatively little specialization in the
field.
Likewise, relatedly, there is little true (scientifically meaningful)
science-practitioner work being done, where clinicians work closely
together developing and showing better inter-rater reliablilities. The
best example of a result of this short-coming is the very very little
work
on improving inter-rater reliabilities on diagnoses between the
publication
of the DSM-III and DSM-IV. It is as if we rely on the DSM committee for
all this work. Most reliablilty tests are done AFTER-THE-FACT, AFTER
new
options for diagnoses have been established (once a decade or so). The
research *they* examined IN ARRIVING AT THE SET OF DSM-IV DIAGNOSTIC
OPTIONS, i.e. the research they had to examine, contained *VERY,VERY*
little work on obtaining inter-rater reliabilities. THIS MAKES NO SENSE
FROM A GOOD-SCIENCE PERSPECTIVE. (For the source on this see the
Sourcebook for the DSM-IV, VOLUME 2.) Establishing better inter-rater
reliabilities for diagnoses should be an everyday activity.
The need for the field to BECOME a science is clear. the time for
pretending is over. -- b jesness
Then you haven't been paying attention to the discussion Leslie and I
were having.
ed
P.S. Poor Ed, the mistaken "student"
>Quoting Martin Seligman, from a mailing list:
>....The argument for equivalence derives from Christensen and
>Jacobson (1993) and their argument is seriously flawed. For they
>review studies, some badly flawed, in which manuals are used,
>mild and uncomplicated clinical problems are the diagnosis (but the
>diagnosis is made by doctoral level providers), and duration of
>therapy is brief and fixed. Precisely the situations in which clinical
>judgment, experience, and education matter very little. A seven-
>year-old may be able to fly a one-engine plane in clear weather, but
>this does not mean she can handle a 747 in a thunderstorm....
An apt analogy. Lord knows I've been in a number of thunderstorms.
ed
Dear Ed,
You say: "This lends support to Leslie's claim that PhDs do better
(in some cases)
than master's level therapists. "
I say: No, not really. The issue is so confounded in the comparison
Seligman cites (between marriage counselors and others) that it **is**
meaningless. Who knows what all the counfounds might be.
Also note: (quoting Seligman): " Marriage counselors (presumably
mainly MFC's, MFT's, and other nondoctoral providers--although degree
was not queried),..." (end quote). Presumably! (why so presumed?) This
is such bullshit, even that most biased Seligman should realize it. A
you student like you, Ed should know much better.
All the best evidence does indicate that experience (or staying in
the field) is somewhat related to effectiveness (maybe marriage
counselors hang around less long and end up having less experience on
average). BUT (in any case) nothing in the research well addresses the
issue of training. Also we have virtually no evidence on WHAT TRAINING
is *really* need for effectiveness for MOST clients. The research on
[did you see it? ; OF COURSE NOT , SILLY !)
regards, b jesness
P.S. On the other matter: Needless flaws exist in virtually all
"therapy" efficacy research (as well as in the research you *are
willing* to criticize) because of the unethical cowardice of the field.
E.G.: No reasonable controls, nor any test of therapists against
reasonable alternatives. A sham and an disgrace. I will keep saying it
until it simply makes sense. I shall break through the denial just as
one would with a classic alcoholic (though now it is the power-addicted
we are dealing with). Have a good day, Leslie.
>....The CR dataset was conflicting on this issue. Social workers
>(presumably mainly MSW's, although degree was not queried) did
>as well as doctoral level providers. Marriage counselors
>(presumably mainly MFC's, MFT's, and other nondoctoral
>providers--although degree was not queried), on the other hand,
>did significantly worse. This was not an artifact of the fact that
>marriage counselors see couples in troubled relationships. CR
>compared doctoral level providers treating marital and sexual
>problems to marriage counselors treating the same problems, and
>the doctoral level providers did significantly better. Here are some
>relevant data: The marriage counselors (n=197) seeing
>respondents who checked "marital or sexual problems" had
>significantly poorer global outcome than other therapists (n=753)
>with these patients (p<.0017, controlling for initial severity)....
This lends support to Leslie's claim that PhDs do better (in some cases)
than master's level therapists. And, earlier, I stated that the finding
that MFTs etc. did worse than PhDs did not control for the fact that
couples are harder to treat than individuals; I was mistaken about that.
ed
The recent Stein and Lambert "meta-analysis", in the Spring, 1995
Jour. of Consulting and Clinical Psyc., is a professional disgrace.
Most studies cited were not on the issue they were trying to address in
2 senses. The issue they were trying to get at was the effectiveness of
trained professional counselors (or "therapists") VS.
"paraprofessionals." The first big problem is that MOST of the studies
they examined (by far) AND INCLUDED in the "meta-analysis" involved
comparing EXPERIENCED professionals to
INTERNS or counselors-in-practicum. These are not studies of
professionals vs. "paras" IN ANY SENSE, EVEN WITH THE EXPERIENCE
CONFOUND ASIDE. Second, the vast majority of the studies
also had this issue of trained vs. less trained (or whatever) as an
issue on the side (often something just examined in passing); i.e. most
of the studies were not really studies of the question being examined NO
MATTER HOW POORLY YOU DEFINE "PARAS." Finally there were only 3 studies
that were both GOOD by normal standards and used measures that could be
considered objective. These were 2 in favor of the "paras" and one
showed a tie. In both the 2 best studies, actually looking at what
rational people would call "paras" and having this as the actual focus
(purpose) of the study, I believe the "paras" faired better. I shall
elaborate more below on the great likelihood of confounds.
The authors' conclusions from their "meta-analysis," which were
actually contrary to the facts when looked at the way I did above are
inexplicable. I can conclude only that these researchers are totally
incompetent AND/OR this was an effort at subterfuge. For a more
detailed look at the study, read on.
Below is a compliation of past postings I have made about this
"meta-analysis" last summer; because it is a compliation it is
somewhat redundant:
I would like to address the recent Stein and Lambert study in the
Jour. of Consulting and Clinical Psyc., Spring, 1995. They concluded in
this "meta-analysis" that grad. trained therapists yield modestly better
results in outcome measures from clients than paraprofessionals.
Confounds are a big issue and very much so here as I shall describe
below. Furthermore, not only is there likelihood of serious confounds
that abound but no rating system for study quality was involved in their
meta-analysis. Often the better studies (including previous
meta-analyses) indicated results contrary to what they reported for
overall conclusions.
In this Stein and Lambert review ("meta-analysis"): Even in the
SELECT group of studies that had objective measures and supposedly did
show some effect for more TRAINED vs. LESS TRAINED, it is good to give
people a realistic and meaningful idea of the magnitude of those
differences found. On objective outcome measures (objective exit
"tests") where differences were found, the "effect size" was .2 (once
the 1 outlier study of the 10 was thrown out as the authors suggest).
THIS MEANS a one fifth of a standard deviation difference on average
between the groups (please see the "FOOTNOTE" AT THE END)**. (TO MAKE
THIS MEANINGFUL: This is *less than half* [(note corrected statistic in
this new edition of the review)] the level of difference shown between
males and females (where males and females differ at all) on several
objectively measured interpersonal traits. IN THIS SPHERE this level
of difference is NOT considered impressive (certainly it is not
considered differentiating); .2 s.d. is at most about a quarter of the
difference between males and females shown on conglomerate scales set up
to differentiate them.)
Another thing to note w/r to these 9 studies from Stein and Lambert I
referred to above: with an s.d. of the groups at .31, 1 or 2 of these 9
studies likely showed the "paras" doing better (i.e. due to variability
in results amongst the 9 studies -- and recall this is the select group
of studies that showed more than the typical outcome difference). AND I
must add that in these studies showing a difference with grad. training:
These differences could very well be due to confounding factors (BIG
ONES): perceived status of therapist, age of therapist, experience (a
matter different from training) and OTHERS! None of this was
controlled. The only controlled study w/o confounds showed untrained
listeners superior to therapists for a BROAD RANGE of college student
problems.
Also remember these are group data and with just a .2 of a standard
deviation difference between the grad. trained and the "paras (their
definition)," a sizeable number of the "paras" (in EACH of the nine
studies) were doing better than the professionals on average (act. just
slightly less true than the other way around).
Because some personal uncertainty still remained for me with respect
to these studies: I went back and looked at all the particular
studies where objective measures were involved. In 6 out of nine, the
comparison was actually between late stage grad. students (in practicum
or interns) vs. EXPERIENCED degreed professions. Obviously this is not
the comparison either I or S & L were supposedly out to address with
this study (again, we are really not looking at training, but
experience). The 3 studies that remained using objective, typical
psychological measures of symptom change found no difference paras vs
profs. in 2 cases AND 1 study favoring the paras.
Furthermore, INDEED it is still true today that the best research
available on professional psychologists vs. others, and the only
controlled study, is one that compared professionals to untrained
individuals. This is the Strupp and Hadley study, 1979. And here, it
was shown that intelligent good-listeners could help college students
with a broad range of problems at least as well as professional
psychologists (Strupp and Hadley, 1979). It is completely unacceptable
that a study comparing professionals to "paras" without major confounds
has not been done. It is worse than if a drug company did not do
placebo studies. Worse because we do not know that "paras" would
materially or substantially provide anything different than grad.
trained psychologists.
In spite of the grave deficiencies, weaknesses of the studies
presented, and unjustified conlusions of the authors, still this S & L
"meta-analysis" has been heralded and is the study on the basis of which
John Grohol and others have argued that there is a modest difference in
outcomes with clients when grad. trained therapists and
paraprofessionals are compared. Readers can see for themselves after
going to the source (Stein and Lambert in Jour. of Couns. and Clin.
Psyc., Spring, 1995) that what I have been able to say about this study
is true and it is a mess. This is a much worse than usual meta-analysis
(many are very, very good and useful). Most meta-analyses are summaries
of studies that were on the actual matter of concern. Again, here in
the typical study included in this meta-analysis, the primary focus of
the study was not the question at hand (not even: the effectiveness of
more trained VS. less trained) but this was either a secondary
hypothesis of the study OR results "that were almost presented as an
aside." Again, in fact only 1 study included in the whole report had a
similar primary focus (still not identical to the question at hand) and
was controlled for confounds (this is the old Strupp and Hadley study,
'79; NO MORE RECENT ONES HAVE BEEN DONE). This study showed no
difference in counseling outcomes between trained psychologists and
totally untrained "nice guy" professors doing counseling with college
students with a BROAD RANGE of problems. To quote the authors of the
study itself on these matters:
"Readers familiar with the outcome research in this area are aware
that authors typically did not design their studies to primarily
investigate the effects of therapist training or experience. Indeed,
examining the relationship between training or experience and outcome
was usually a secondary hypothesis, or results were presented almost as
an aside. Thus, it appears that the investigation of the relationship
between level of training and outcome was not planned as carefully as
procedures designed to study the central hypothesis. For example, as
noted earlier in the article, the typical study did not adequately
isolate the issue of professional training from confounding variables.
We are aware of only one published study that has quite reasonably
isolated the ingredient of therapist training by controlling some of its
inherent confounds and correlates (e.g. age, status, perceived
expertness, interpersonal skills, etc.). This was the Strupp and
Hadley's (1979) classic study comparing male university professors, who
were selected because of their reputation among students as being
approachable and easy to talk to, and experienced male
psychotherapists."
It appears that while the issue of fully grad. trained therapists
vs, "paras" may not be a new one, THE RESEARCH WOULD BE. IT HAS NOT YET
BEEN DONE!! Also it is VERY likely that a number of the major confounds
I cited as possibly present probably were (and some I didn't think of).
Some (I won't say "a lot") of good and reasonable work is yet to be done
before we have a clue as to whether well-selected, reasonably-trained
and supervised "paras" do as well as clinical and counseling
psychologists with the terminal degree with the majority of clients.
The best evidence we have indicates that in general they will probably
do as well. There is some suggestive evidence from other studies cited
by Stein and Lambert that fully trained therapists may be helpful with
diagnoses and that dropout rates with "paras" become higher only when
more than 10 or 15 counseling sessions are required. This is suggestive
for the role of a new type of supervising clinical psychologist. Still
the full merit of reasonably selected and trained "paras" has not been
addressed. One wonders whether a major research issue will ever be
addressed when it is not in the vested interests of therapists. It is
surely overdo.
How these authors concluded in this "meta-analysis" that grad.
trained therapists yield modestly better results in outcome measures on
clients than paraprofessionals is very hard or impossible to understand
on a reasonable basis. As I indicated, bad definitions, poor measures,
and confounds are a big issue (and very much so here). As I indicated,
the likelihood of confounds abound; and no rating system for study
quality was involved in their meta-analysis. Often the better studies
(including previous meta-analyses) indicated results contrary to what
they reported for overall conclusions.
** FOOTNOTE:
The "effect size" in the Stein and Lambert research for the studies
using objective measures was defined as follows: ((mean of more highly
trained ON THE OBJECTIVE MEASURE)-(mean of less trained ON THE OBJECTIVE
MEASURE)) divided by the STANDARD DEVIATION OF THE LESS TRAINED ON THE
MEASURE. I tried to gage the magnitude of this effect in meaningful
terms by assuming the s.d. shown by the "experimental" ("para") group
would be about that of the general population. (It is actually likely
LESS thus I'M INFLATING the actual "effect" the way I represent things,
though this is not certain). Anyhow, assuming s.d. of the para group
equal to the general population on the measures of symptoms involved, I
argued tha .2 standard deviation difference (more trained vs less) would
not typically be consider meaningful on the measures.
I do admit that it may be too much of an assumption to assume that
the standard deviation shown by subjects in the control group (para
group) is equal in magnitude to the standard deviation of the general
population on the outcome measures. Yet, again, if anything a measure
of range shown by a select group on a pertinent measure is at least
typically (though not invariably) smaller than that in the general
population AND would come out that way here, I think, unless the shift
in therapy for some was dramatic, while for others none or for the
worse. Typically they would start AND (equally treated) END more
similar to one another than those in the general population. With this
in mind, as I've said, the real denominator in the formula for "effect
size" as they defined it would be even less than the s.d. of the
population sample on the test. And, this would mean the difference in
outcome measures is likely less than .2 s.d. of the measure (generally
speaking), that is, when looking at what this would mean in the actual
general population, on which the s.d. of the measure is based. This
would make the difference between paras and profs. even less meaningful.
The only thing to argue against this is if you can find out that there
was greater variability in the client sample than in the general
population.
---------------------
>>Dear Leslie Packer.Ph.D.
>>Regarding the demonstration you asked for: The demonstration of a
>>lack of evidence is easy. Here it is:
>>[did you see it? ; OF COURSE NOT , SILLY !)
>>regards, b jesness
[Uh? Regards? After _that_ string of _un_therapeutic abuse, or was it
'parapro', or "Professorial" abuse] <Noch'n mal; es +gibt+ keine
Professorenherrlichkeit hier, susses <VBG> kind: ;Åš+Åš >
>Reading comprehension problems acting up again, m'dear? <concerned look>
Dr. P., 'Prof'' J?sness was never too good with comprehension, and I
often fear for the condition of his striate cortex. <Or did I 'hear'
frontal lobe syndrome...? ...perseveration, fo' _sho'_...?>
>If you will go back and re-read my post, I did not ask for a demonstration
>of lack of evidence. I asked for a demonstration of equivalence. Perhaps
>the distinction was lost on you.
Distinctions are rarely emitted from the undistinguished, whether
'Professorial' or or not, and 'Prof.' B??? is no exception, Dr. Packer;
you might as well correspond with a sandbag, as well we both know.
>Kinda takes the sting out of your repartee when you're the only one who
>doesn't understand the point, doesn't it? Ah, but don't worry. Just go
>back to your seat and try again. We'll all wait patiently for you.
_Repartee_? Nope; _bland repeatee_. <But you knew already what to
expect...> <g> <Waiting behaviours commence.>
>>P.S. On the other matter: Needless flaws exist in virtually all
>>"therapy" efficacy research (as well as in the research you *are
>>willing* to criticize) because of the unethical cowardice of the field.
<Cor, I'd be careful about supporting one such as this, one might be
tarred with the same J?sness, no?... ...'many' and 'all', as two
different posters are wont to say...>
>How unfortunate for you that just repeating something over and over again
>(check the DSM for appropriate number... 300.03, perhaps?) doesn't give it
>veracity.
[Click, Clack]
300.03?; fur mesen, aah prefers the .303, it's terminal, it's mah
feryend, it sees and reaches for miles*, and from the faces of united
buttheads it always removes smahles.
* <'specially with a 'scope>
Absolute Disclaimer: 'Prof.' B.J and his young Canadian friend should
not even ideate <let alone falsely and _hysterically_ [g] assume> that I
own, or have _***access***_ to, the weapon on which I gained my
marksman's badge - 'cos I don't': I don't have and would reject such
access, or ownership; I believe in kill-files rather than killing or
maiming, or even psychologically hurting <altho', I believe that 'Prof
Jesness' has done some of the last himself, and I have some postings to
show...>. I also believe in second-hand access to postings - the latter
make me grin; that's enough for me. Savourrr the flavourrr....
....OTOH... ...wot's the position, re. falsely declaring oneself to be a
Professor of Psychology?
[Click, clack]
I was just more than a tad p****d off with seeing direct posts by people
who have nothing new to offer, who will not entertain the restrictions
laid down by IRB's <problems with reality testing?>, and who
consistently come up with bullship when confronted with data. So, I
placed them in my kill-file. Now, because of the quality of responding
beh's., I wonder whether I should refrain from killing threads... ...the
eternal NG dilemma.
Responses? OK. They tend to teach me. I am also amazed that we can,
from a sow's ear, a silk purse make. My 'umble 'gratulatins to the
respondents, you are better than 'Bad' <or was that 'B???'?>
<Why don't we have a stats thread?>
B??? I wouldn't mind if he was 'RET to the bone', but he's not. He's
?r?d to the bone, and so it would seem is the axon grinder/rent-a-quote.
How _could_ _anyone_ defend such a _plonker_, I ask? Well? _HOW_?
<Sorry, shouting.> Not good for credibility, what, what, what?
>>E.G.: No reasonable controls, nor any test of therapists against
>>reasonable alternatives.
>While I actually agree with you about the absence of necessary controls and
>adequate designs, I disagree with you strongly as to why.
IRB's, ethics comittees, the actual _experiences_ and _feelings_ of
_patients <other side of psychology being the subject>, near whom I
wouldn't even let a parapro _imagine_ a fart?
B???, you're dangerous, and that's the truth. You have overextended an
idea; you have cuddled up to those who extol the virtues of social
medicine, you have cuddled up to those who attempt to propagate self
help, and you also ride on the bandwaggon of meta-analyses <been taking
tuition recently?>... ...never mind your earlier <uninformed> attempts
to snuggle up to those who opposed those whom you opposed... ...not
realising that said opponents actually had a clearly defined theoretical
viewpoint of their own... ...not yours.
>>I shall break through the denial
<This from a man in <g> denial.> <Or was that 'renial'?>
>Well, now you're in real trouble. Some of us don't believe in denial <G>.
Shame, there's a _good_ few miles in this one... ...Monsignor J?sness
_loathes_ denial! He's a BTTB'er, as opposed to an RET'er...
>> Have a good day, Leslie.
<Have a nice 'phone bill Bradley, using a web browser is costly, and
probably requires 'BTTB power', eh?>
><chuckling> Thanks for the laugh.
Moi aussi. [;Åš+Åš to you, B?????y]
CHEERS, eh? <Stick on the ice? PTSD? Explanation?>
--
Peter
(end quote) It was addressed to you (sorry for my lack of salutation)
-- b jesness
>so far -- that you can personally take credit for?
And I am still waiting for your reply, Cognitee. I really want to know
what _positive_ changes you think you've personally made for mental health
patients using your approach and arguments.
It is a client and science advocacy group,
dedicated to furthering science standards and practices in the therapy
field. We insist on fair and proper representation of treatments and on
providing information about costly or limited treatment options
available to clients "up front". We believe options and evidence of
their efficacies should be provided to clients before they enter a
course of counseling or therapy. The various treatments and programs
offered by each professional mental health service provider should be
outlined in some detail in a booklet made available to clients. Only
this would provide reasonable information before the expense of and
commitment to a course of treatment.
Also, techniques or methods used that have NOT been clearly shown to
have efficacy AND validated for a particular, reliably-identifiable
problem type (i.e. showing blind inter-rater reliability) are NOT be
referred to as "therapy." Correspondingly, when what is done is
COUNSELING, the cooperative nature of this should be made clear and it
should be properly represented, engendering appropriate expectations.
Counseling is considered a most noble cooperative endeavor, requiring
the most consideration, judgement, and intelligence. Those who are
well-adapted will be better counselors. For this reason, and considering
the rest of the evidence, counselors/therapists should have a long
history of good adaptation.
Moreover, BRAD believes daily standards in practice should provide
for on-going research (such as for the development of reliable
diagnoses) and this should be done within each large mental health
service agency. Furthermore, basic foundation research definitively
showing that graduate-school-trained counselors are superior to other
sources of help must be done to establish the range of problems for
which special treatment by professionals is actually better (and not
inferior to other more accessible and less costly sources of help, e.g.
peer counselors or paraprofessionals). BRAD also supports (given at
present there is no evidence against it and some good evidence in its
favor): peer counseling programs and counseling programs for
paraprofessionals. BRAD seeks to demystify mental health professions
and rid it of great myths. We hope for a sensible, delineated mental
health care SYSTEM, with the care often involving peers and
paraprofessionals and for care to be provided by individuals within a
client's working community.
"Cognitee <Cogn...@aol.com> wrote:
>No Ed,
> I think all the concerns about design matter. All these concerns
>must be honored strictly. I do. My point is that there is more to good
>science than "research design" as currrently considered.
Of course there is. My objection was to the way you offhandedly
dismissed the importance of research design." (end quote)
ED !, I, in no way, in any part "dismissed" research design. BUT this is usually a relatively easy matter, typically involving less=
theory or judgement. What to study and where to try to improve inter-rater reliabilities systematically is a greater matter and inv=
olves more judgement and wisdom. Still it is not hard if done (as it has to be) by a local working group. It is impossible "by com=
mittee" in most significant ways. And this is exactly why our progress in diagnoses has been advancing (at best) at a snail's pace:=
too much committee work ALONE. But (of course) the proof of quality of the sharing of the results is always in the inter-rater rel=
iabilities.
Quoting you again (at first quoting me):
">Good
>observations and with the continuous establishment of a series of
>excellent inter-rater reliabilities (and thus a basis for good validity)
>DEPEND *AS YOU MUST KNOW OR SENSE* ON MORE THAN JUST "DESIGN."
Yes, but let's see you get "good observations," "excellent inter-rater
reliabilities," or any sort of validity without a good research design.
ed" (end quote)
I AGREE you must have good design. You also must breathe, eat, and sleep. -- b jesness
Of course there is. My objection was to the way you offhandedly
dismissed the importance of research design.
>Good
That's a good point. The size of the treatment groups were 16 and 15. A
couple other things about that oft-cited study:
1. The "paraprofessionals" in Strupp & Hadley were university professors
specially selected by students "on the basis of their reputation for
warmth, trustworthiness, and interest in students." These were not even
average professors, much less average paraprofessionals.
2. The number of professional therapists in the study was five; the
number of "parapros" was four (I think -- it may have been five also).
3. The measures used were not particularly sensitive to change (e.g.,
MMPI).
4. While no significant differences appeared between pros & "parapros,"
differences in group means were all in the expected direction.
ed
Interesting, Ed, and I agree with you. I was under the impression that
all scientific experiments need to have a sound design, otherwise one
might as well go play in s.p.p.? <G>
OTOH, one could always become a parapro, or an adherent of German
Speculative Idealism. How apt.
Yours, in tears of laughter.... .....just like B??? is, I suspect, since
he's probably enjoying himself soooo much. This is just a game.
--
Peter
<trying hard to lurk, but failing intermittently, it's that damn 'send' button
thangy, it looks so professional!>
Well, since I have one handy...
>4. Didn't Strupp and Hadley report that the group means were
>misleading because of the large within-group variability? Again, this
>would argue for a real problem with drawing _any_ conclusions from their
>study.
Yes, they did: "...considerable variability in individual dyads.... As
our work progressed, we became impressed with the fact that group
comparisons obscure the very phenomenon that must be understood..."
(p. 1135).
Strupp published four articles in the 1980 Archives of General
Psychiatry, which examined these "individual dyads." In my opinion,
these are much more interesting articles (methodologically and
substantively) than S & H (1979). Furthermore, these 1980 articles
document (in places) how, confronted with difficult clients, the
university professors ("parapros") were often ineffective (not that the
pros had much luck with these cases, either, but the parapro's mistakes
were particularly glaring).
>5. You keep repeating that the paraprofessionals performed equally well
>over a "BROAD" range of problems exhibited by college students. That is
>not my recollection. I cannot remember offhand _how_ the student
>participants were recruited, and perhaps you could refresh my memory...
Allow me. They were selected based on elevations of depression, anxiety
(psychasthenia), and social introversion scores on the MMPI. Average T
scores on these were, respectively, 87, 82, and 70 for the
professional therapist group and 76, 74, and 69 for the "parapro" group.
Note, incidentally, the higher scores on intake among those assigned to
the therapists... As S & H note, randomization was not entirely
successful.
ed
P.S. I will not be pleasant in the face of the "pleasnant acquiescence"
to the inadequate status quo. "I'm mad as hell and I'm not going to
take it anymore."
But then what of book writers who base theories on one or two patients,
who are not in any way 'picked at random'.
It seems that other that the Horatio Alger appeal, no other claim could be
made.
It is more than mildly hypocritical of you, Mr. Jesness, to talk
of "science" all the time and have done so little yourself.
As many people have pointed out many times in the past, why don't
you get off your ass and do some? It's one thing to sit there and
bitch and complain for months (almost a year now) on an Internet
newsgroup and quite another to get funding and subjects to actually
do all the "science" you claim is needed. If you're such the
expert in science, Mr. Jesness, you should truly be leading the
field in directing others and obtaining the means to do the research.
I'd very much like to see you in such a leadership role in this
field.
Last time I looked, research doesn't grow on trees or fall out
of the sky. Or get done by harassing others to do it.
-John
--
Mental Health Net
http://www.cmhc.com/
Top 5% of the Web - Point
"On Jul 23, 1996 15:42:32 in article <Re: Another debate between BRAD
and
Seligman, with BRAD the victor>, 'Cognitee <Cogn...@aol.com>' wrote:
>P.S. Especially those who engage in denial would not believe in it.
--
Sure... but so will those who know that they don't.
(Isn't this fun?)
Leslie E. Packer, PhD" (end quote)
What is this the completion of the proof? Like: All male humans are
The study you're referring to gives no more than half-hearted,
mild support for the relationship between therapist efficacy and
therapist experience. That fact alone contradicts your assertion that
the reason it was accepted was because it supports the status quo. The
"status quo," to the extent that such a thing exists in a field this new
and divisive, is that experienced therapists are MUCH, MUCH better than
inexperienced ones. Which is hardly what Stein & Lambert concluded.
ed
P.S. When you realize, as Allen Ivey does, that the counseling/therapy
field is operating "in a narrow fog." A very apt metaphor. List to
BRAD.
I think your dismissal of the importance of design is typical. As a
"scientist," you suck.
ed
P.S. You are not too smart when you're cogent, are you Pete?
P.S. I've expressed some regard for single sugject designs elsewhere.
In today's mental health care system, I guess I love them, though
reversing conditions seems UNFORTUNATE. It would be often unnecessary
IF ONLY ONE HAD AMASSED DATA THAT SPOKE WELL GENERALLY TO A
WELL-IDENTIFIED (defined) GROUP OF PROBLEMS.
I am in a leadership role. You just don't realize it yet. -- b j
In article: <31F6DB...@coil.com> "John M. Grohol Psy.D." <gro...@coil.com> writes:
> It is more than mildly hypocritical of you, Mr. Jesness, to talk
> of "science" all the time and have done so little yourself.
> As many people have pointed out many times in the past, why don't
> you get off your ass and do some? It's one thing to sit there and
> bitch and complain for months (almost a year now) on an Internet
> newsgroup and quite another to get funding and subjects to actually
> do all the "science" you claim is needed. If you're such the
> expert in science, Mr. Jesness, you should truly be leading the
> field in directing others and obtaining the means to do the research.
> I'd very much like to see you in such a leadership role in this
> field.
> Last time I looked, research doesn't grow on trees or fall out
> of the sky. Or get done by harassing others to do it.
" 'Tis the voice of the Para; I heard him declare,
'You have baked me too brown, I must sugar my hair.'
As a duck with its eyelids, so he with his nose
Trims his belt and his buttons, and turns out his toes.
When the sands are all dry, he is gay as a lark,
And will talk in contemptuous tones of the Thera-Shark,
But, when the tide rises and thera-sharks abound,
His voice has a timid and tremulous sound.
I passed by his garden, and marked, with one eye,
How the Student and the Professor were sharing a pie.
The Professor took pie-crust, and gravy, and meat,
While the Student had the dish as its share of the treat.
When the pie was all finished, the Student, as a boon,
Was kindly permitted to pocket the spoon;
While the Professor received knife and fork with a growl,
And concluded the banquet -" ...
That'sa spicy para-meata-balla, as all Alice afficionados will know.
Still laughing in my sleep.
--
Peter
<relurks>
** Here's a very apt analogy: It's easier to tell a meter from a yard
than it is to tell a 1/2-inch from a centimeter.
1. What type of alternative classroom environment(s) do we need to develop
so that children who have ADHD can function to their potential without
medication?
2. Some SSRIs that are highly selective for the 5HT-1A receptor have been
demonstrated to help reduce alcohol consumption in animal models. Can we
develop a comprehensive program of medication and environmental support
that will reduce alcoholism, a disease that is extremely costly to our
society and that wrecks a lot of lives?
3. What child-rearing and/or school-based experiences are effective in
promoting the "just say no" attitude towards drugs in our children?
4. With so many people dying of AIDS, how do we effectively promote safe
sex?
Shall I continue? Your concerns are not necessarily my concerns or
priorities, Brad. And my priorities and concerns are probably no one
else's.
So instead of screaming that people are being defensive, biased, etc. just
because they don't jump on your bandwagon, perhaps you might step back and
try to see other perspectives.
I am more concerned in seeing that everyone who needs help has access to
the system. But I am not willing to compromise and say that means that we
start by giving them less than professional help.
Leslie
P.S. I do somewhat like your single subject research outlook at times and
your attitude towards being responsible in client assessment. Too bad
that is all I like about you
>4. Additionally, didn't Strupp and Hadley report that the group means were
>misleading because of the large within-group variability? Again, this
>would argue for a real problem with drawing _any_ conclusions from their
>study.
I think I can confirm this. Strupp & Binder (1984; _Time Limited
Dynamic Therapy, Basic Books, I think) commented on this program of
research and did mention that they suspected the large within group
variability in outcome prevented between group differences from being
detected.
Another interesting comment they made was the reports of the
untrained college professors discomfort at running out of material to talk
about before the end of the allotted sessions.
>If I followed your argument, the small effect size (which you estimated at
>2 SD but suggest might be an overestimate) shouldn't be considered
Are you all talking about 2.0 SD? As in (control group mean) -
(treatment group mean)/sd(control-group scores) = 2.0? If I'm reading
this right, this is a huge effect size! (thinking of Cohen's d here)
>important because it is small. Was I understanding your correctly?
^^^^^?
>been in the other direction. There is an important difference between
>statistical significance and clinical significance, but one cannot truly
>have the latter if the former does not exist. So demonstrating a reliable
I don't think I understand this Dr. Packer: are you saying that
statistical significance is necessary for clinical significance? If you
mean that reliability of measurement and reliability of effect are
necessary to show clinical significance then I'm sure I agree.
But if you really mean statistical significance, as in some p <
alpha, I think I disagree. Since statistical significance is directly
influenced by one's decision about the level of Type 1 error one is
willing to make in one's inferences on the basis of a particular design
before the fact, and clinical significance is a judgment about the amount
of change in a patient or group of patients determined after the fact, I
think these are apples and oranges.
I imagine a counterexample where a small, but reliable, effect
would not attain significance due to low power in the design because of a)
small samples b) small Type 1 error rate, and/or c) high within-group
variability due to other reasons.
>than values such as "extremely highly trained vs. Zero training), they have
>produced a _lower_ estimate of the effect of the independent variable. Had
>they actually selected the levels properly (by your definition of what
>levels they should have used), it is more than likely that there would have
>been an even _greater_ effect (in the direction of trained >
>paraprofessionals).
This is a really interesting point.
>Regards,
>
>Leslie
>---
>
>Leslie E. Packer, PhD
Really interesting review, thanks.
Scot
--
\----/ \----/ \----/
|||| Scot McNary |||| sc...@wam.umd.edu ||||
|||| Graduate Student, UMCP |||| (bethany too) ||||
/----\ /----\ /----\
Yes, and in the Strupp (1980) articles, there is the report of the
professor/"parapro" who got locked in a power struggle with an angry,
resistant client and kept trying to feed him advice (condescendingly).
> Are you all talking about 2.0 SD? As in (control group mean) -
>(treatment group mean)/sd(control-group scores) = 2.0? If I'm reading
>this right, this is a huge effect size! (thinking of Cohen's d here)
That was a typo, I think, Scot -- it was supposed to be 0.2.
ed
Boy, no matter how many times we go over this territory, you just can't
get it right, can you?
0.82. Memorize it.
ed
ps, and Leslie, none of your snide remarks about meta-analyses <s>.
I am continually struck by how nonresponsive you are to the very good
points Leslie raises. She takes care to quote you at length and respond
specifically to each of the issues you raise. Why do you not
do the same? Why do you repeatedly fail to respond substantively to her
critiques? [rhetorical question]
ed
>couple other things about that oft-cited study:
>
>1. The "paraprofessionals" in Strupp & Hadley were university professors
>specially selected by students "on the basis of their reputation for
>warmth, trustworthiness, and interest in students." These were not even
>average professors, much less average paraprofessionals.
Right... so there was some "stacking" or optimizing in terms of the
paraprofessionals used. But that doesn't concern me so much as long as we
recognize that these "paras" had been selected this way. We also need to
realize that as university professors who routinely provide academic advice
and counseling to their students, most of these professors already had some
experience in counseling. As a former academic, I know that students
frequently drop in to talk when they're upset or in trouble if they
like/trust the faculty member. Hence, these "paras" _may_ not have been
equivalent to "new" paras in their abilities.
>
>3. The measures used were not particularly sensitive to change (e.g.,
>MMPI).
I couldn't remember the measures. Thanks, Ed.
>
>4. While no significant differences appeared between pros & "parapros,"
>differences in group means were all in the expected direction.
>
Which supports the concern about the power of the design and the tests.
Thanks again.
Hopes this helps to put Brad's statement re. a "broad range of clients"
into perspective. S & H themselves note that they made a "concerted
effort to obtain a truly homogenous patient population..." (p. 1126)
ed
>>on a large N). Insisting on the true experiment as the only thing
>>that's good is certainly true *only* in certain contexts (AND NOT
>>OFTEN). I was topic of my graduate class in research methodology. -- b
>>jesness
>>
>>P.S. I've expressed some regard for single sugject designs elsewhere.
>>In today's mental health care system, I guess I love them, though
>>reversing conditions seems UNFORTUNATE. It would be often unnecessary
>>IF ONLY ONE HAD AMASSED DATA THAT SPOKE WELL GENERALLY TO A
>>WELL-IDENTIFIED (defined) GROUP OF PROBLEMS.
>>
>>
>In your statement above, you stated that you would "take a very strong
>correlation over .... significant results." Surely you know that one
>cannot tell cause and effect with a correlation (or have I totally missed
>your point?).
Nope, you didn't misread him. Amazing, huh? And don't you just love his
statement about reversing conditions? Shows a real... how shall we say
this diplomatically.... lack of sophistication about research? And based
on these impeccable credentials, he naturally presumes to tell people how
to run research <laughing>.
Pull up a chair, Mark, and put up your feet. By the time you get done
reading the rest of his posts, you'll probably be dizzy <G>.
>progress (or lack thereof). Again, just a little reminder that the issue
>of effectively evaluating treatment is often done one client/patient at a
>time and administered scientifically.
Well put, Mark.
One must appreciate that "significance" is a statistical concept
having to do only with the statistical likelihood of finding the
between-group difference again (either 99 or 95 times out of 100). This
has to do with: group variabilities, difference in means, and group
size. It's easier to find differences reliably, that is again and
again, with large groups even if the difference is small. (see footnote)
In fact, if very large groups are used very, very small differences
between the group means can be found reliably (and this is all
"statisical significance" means). It is possible that I could get
"significant result" showing that eating beans affects school
performance (after lunch on a standardized test of some kind) in a
controlled study if I had 10,000 children in each group. So you see,
"statistical significance" does not equal (or mean) meaningfulness (in
a practical sense) UNLESS realistic sizes are compared.
Correlation coefficients are better indicators of absolute STRENGTH
of the actual results or findings. They do not rely on group size
(though the certainty of the exact value of the correlation increases
with group size). Correlation coefficients could be reported with
experimental results and probably should to allow for clearer
comparisons of strenth of results. (Correlation coefficients can be used
with experimenatal designs as well -- it is a statistic that is
calculated and is not linked up with a particular design). Hope this
helps. -- b jesness
** Here's a very apt analogy: It's easier to tell a meter from a yard
than it is to tell a 1/2-inch from a centimeter.
NOW THERE ISN'T TOO MUCH "AMAZING" ABOUT WHAT I SAY, HUH? -- b jesness
>small samples b) small Type 1 error rate, and/or c) high within-group
>variability due to other reasons.
Whoa. You just made an error above. "Reliable" is the _same_ as
"statistically significant." "Reliable" is one of a number of synonyms
that are used for "statistically significant," the other most common phrase
being "not due to chance or random error." I think that perhaps what you
were trying to say is that there is a real effect (in the population) but
you don't recognize it in your research because.... etc.
All of the parameters you cited in your example above would, indeed, work
against an outcome achieving statistical significance. That's why I always
taught my students to _set_ their power and then use power analysis
procedures to determine their N. I am probably one of the few people who
do not automatically set alpha and/or beta. I really try to think of the
implications of each type of error for the particular question I'm looking
at.
>
><snip>
>
>Really interesting review, thanks.
Thanks for the kind words.
--
Leslie
Leslie E. Packer, PhD