Dear Raija-Leena Punamaki,  (and all)

 Some of the list subscribers are using text only readers, so I have taken
the liberty of distributing the ascii version of your message - Richard
Wilkerson, technical support manager.

Dear Colleagues,
                                Helsinki, 19 May 1998

Hereunder are my (Punamaki - the author) responses to your comments on the
Article: "Determinants and mental health effect of dream recall among
children living in traumatic conditions".

First let me apologize for the delay for my response, due accumulation of

Now I would like thank you all for having devoted your time to comment the
article.  Getting feedback is a luxury to any researcher.  As a beginner in
dream research (although not in developmental psychology and stress
research), I felt like a student when "all the big names were commenting my
In this communication I will concentrate on your critical comments on
methodological problems and respond to each of you separately.  Later, it
would be intriguing to discusses specific themes, exchange ideas about the
problems and develop some dynamic hypotheses for forthcoming research.


>I personally do not like the mood congruency hypothesis since >it is not
testable. How should it be possible to measure dream >emotions
independently from mood after awakening.  I would like >to name it carry
over effect.  It seems more plausible to me >that dreams influence morning

My study setting was entirely based on one-source information concerning
the morning mood and dream content.  I fully agree that a setting involving
separate, more objective and independent measures of dream content (across
the night and various REM), and morning recall and emotion reporting would
have allowed us to test some of the hypotheses. Also the contemporary
emotion theories would brovide much more sophisticated tools to depict the
emotional dynamics than the one I used. You are right about "mood
congruency", it could be called "carry-over effect".  Apparently I was too
excited by the mood congruency literature, and would like to study it in
the dream context.  Let's discuss it later.

To Michael's comment on the Method Section:

>Personally, I think the participants knew very well the aims of >the study.

I would protest your argument that the participants knew very well the aims
of the study(Mental health function of dreaming, role of dream recall in
traumatic environment).  My text was apparently confusing in the previous
draft of the manuscript, which you must refer to: the purpose of the study
was NOT intensively explained to the participants.  I wrote that the
instructions were written in a tightly-scripted statement in order to
guarantee consistency from home to home.  (It went: "Nocturnal dreams are
familiar to all of us.  We would like to better understand childrens'
dreams and to know what they 'see' in their dreams. Therefore we kindly ask
you to report every morning what kind of dreams you had in the night...".
The instructions also included a clause about us maintaining the anonymity
of each child and technical advice as to how to report. I have the dream
diary forms in English (in addition to Arabic, Kurdish, Somalian, Spanish
and Finnish).  
We were conscious that we should absolutely avoid exerting ANY pressure on
substance, quality or quantity of dreams.  We never even hinted to the
children that they should report numerous or little dreams, and the
variation in dream recall was substantial and indeed proves that. 

>Second, the results may be influenced by experimenter effects >since the
groups were interviewed by two different persons.  >The researcher himself
(highly motivated) obtained the sample >with higher dream recall frequency.
I think a comment on this
>should be given.

It would have been ideal to use the same field workers in both places, but
sadly, for "security reasons", even a Gaza resident who may travel to work
in Israel must return daily to Gaza.  I would argue that fielworkers in
both Gaza and the Galilee, were highly motivated.  
>Personally, I like the funny sampling method ... There is, >however, one
weakness in this approach shortly addressed by >Punamäki.  She did not give
a response rate or a sample size of >all potential participants. This could
lead to severe biases in >the data.

We took every child (in the chosen age range) in every third house in a
given area.  Therefore it is impossible to report the sample size of all
potential participants, because we have no idea about how many potential
participants live in houses that we did not visit.

On results:
Please, see my response to Peretz Lavie concerning the confusing values in
the figures.

I would like to hear all of my colleagues solutions about how to take into
account the variation of dream recall in constructing dream content scores.
 I have chosen two ways: (1) using the recall score as a covariant when
possible, and (2) dividing the content scores by the number of dreams +
subsequent standardization.  The latter alternative allows us to use the
scores in a correlation analysis.

I would like to return in our next exchange to discuss the dynamics between
memory, saliency, bad dreams and nightmares.  Michael Schredl and Tore
Nielsen provide intriguing views of the problem.


>I read the paper several times and each time I read it I was >more
confused.  There are too many questions related to the >data presentation
and data collection procedure that cast some >doubt regarding the meaning
of the findings.

(Here I also respond to Deirdre Barrett comment)

Figures 3, 4a and 4b illustrate the interaction between two predicting
variables on one dependent variables.  Therefore raw scores cannot be shown
here.  Plotting the interaction reveals simple regression lines predicting
the level of dream recall (fig. 3) and symptoms (figs. 4a, 4b) in degree of
traumatic events and repressive tendencies (fig. 3), and in degree of death
and wounding and dream recall (figs. 4a, 4b). (Observe that "predicting" is
not a causal but rather a technical term here).  Significant interaction
means that the association between, for instance trauma and symptom, is
different among those with low and high recall, i.e. regression of Y on X
depends upon the value of Z.  

But because the independent and "moderating variable" are continuous
predictors in regression analysis, the "low" and "high" are not in their
original scores. They were chosen to be one standard deviation below the
mean, at mean, and one SD above the mean (Cohen & Cohen, 1983).

Researchers often choose to illustrate the interaction effect by using the
schematic or conceptual definitions (Low, Medium and High) of the X-axis.
This choice might have been less confusing.

There are some disagreements, as I understand, about how to construct the
interaction term in regression analyses.  I followed Aiken and West (1991),
who often refer to the highly influential Cohen & Cohen (1983) in
converting the interaction variables into centred scores. The X-axis scores
range accordingly.  [Another conversion possibility could be the use of
mean deviation scores or, as we did in the old days, just to use any
standardized variables in interaction (cross product term)].

>There is no information on the number of subjects in each >category of
traumatic events (Figures 1a 1b and 2)...  

The Traumatic Event Checklist was normally distributed, and therefore I
ranked in three percentiles that conceptually depict low, medium and high
levels of exposure.  Indeed the number of cases in each group in Figures
1a, 1b, and 2 would have helped to make any sense of the findings.
However, it is not customary, according to APA guidelines.

>What are the definitions of frequent and infrequent dream >recalls (every
night ? 6/7 nights? vs 0?, 1 2/7 nights???).  

The dream recall variable was used as a continiuous variable in all the
analysis (as dependendent and independent variables in regressions). If I
used it as grouping variable in ANOVA, for instance, I conceptualized it
high dream recall (7-6 nights), medium (5-4) and low recall (1-3) and no
recall at all.  
Dream recall variable was normally distributed, so the categorization is
not a great problem. You may be right that we use in liteture to careless
the term infrequent and frequent recallers. But if you have a continous
variable, it has become customary to name the phenomenon accoring to the

>The differences between the traumatized and non traumatized >groups are
far from being clear.

There is, indeed, a problem to objectively agree on the differences between
the so called traumatized (Gaza) and the so called non traumatized subjects
(Galilee).  The research setting was planned to increase understanding on
trauma (or stress)-dream association.  International, Israeli and
Palestinian human right documentation all indicate that childhood in Gaza
involved a substantial level of environmental violence, dangers and painful
experiences, greater than in Galilee.  However, I also measured the
personal exposure to traumatic events, so there are two different
indicators about the level of trauma.  I agree that these groups should not
be call traumatized or non-traumatized.  Perhaps "violence-group" and
"comparison group" would have been better. 

>I also found it particularly distressing that in the discussion >section,
Dagan Lavie and Bliech (1991) study is cited as >supporting the observation
that "the sleep of traumatized >people is lighter.." while this study
reported precisely the >opposite.

This was absolutely my mistake and misunderstanding to which I do not have
any rational explanation. Naturally, our results indicated a deeper sleep,
heightened thresholds and lower level of dream recall among PTSD patients.
I feel very embarrassed. 
But I think that we have not yet exhausted the question about active
blocking versus vigilant processes among trauma victims.
A theory of PTSD, as originally understood by Mardi Horowitz provides
evidence that intrusive and avoiding symptoms tend to oscillate across the
recovery process. The dream researchers Wright and Koulack (1987) suggest
in the similar lines that mastery dreams incorporate disruptive events and
affects (intrusive), and compensatory dreams avoiding painful events.
So the field is interesting, and our results on trauma-recall differ. I
would like to heard your further comments, but the scientific level of my
paper seemed to frustrate you. 

>... it is almost certainly premature to argue from >correlational data
that infrequent dream recall protects >children from developing depressive
symptoms but makes them >more susceptible to somatic and anxiety symptoms.

It is for sure a mistake to say that recall 'protects from developing
symptoms', as developing apparently refers into future.  An interaction
effect refers, however, to a moderation, which means that the negative
(vulnerability) effect of trauma (on well-being) is attenuated or
cancelled.  You are right, one should be more clear, and we (psychologists)
always dispute about the moderating and mediating effects with

>Also, I would be interested in the authors' response to my view >that a
necessary distinction between TRAUMATIC events and major >life
changes/stressors has been obscured in the Introduction.  

The conceptualization of life stress, traumatic stress, trauma and
traumatic events surely requires more work.  I do not think that I equated
bereavement or divorce with traumatic events of this sample (detections,
night raids, deaths).  I used the results of the few empirical research on
stress-dream associaiton to learn about the possible link.  The content of
the stressful events do not have to be exactly the same, the other research
still gives you some ideas when constructing hypotheses.  The more
interesting question might be the subjective appraisal of trauma or stress
as Kathy Belicki noted.

>I want to know how much these effects derive from the fact that >her
traumatized subjects simply experienced more nightmares >(possibly PTSD
nightmares) and that, by definition I think,
>nightmares are both salient (awaken the person) and mood >congruent (cause
post awakening distress)?

The question about nightmares, PTSD-nightmares 'traumatic dreaming' and bad
dreams is a crucial issue in populations exposed to painful experiences.  I
agree that they should have conceptualized more clearly in the present
The issue also relates to the controversial issue whether  
dreams play any adaptive role, have some psychological function or any
meaning in information processing.  In trauma research, there is a tendency
to equate trauma related, repetitious and bad dreams with symptoms.
Conceptually they are confused with nightmares, because they, indeed,
constantly wake up the dreamer and increase distress.

When starting the research I was (and still am) very impressed by the work
by Hartmann and Belicki about nightmares.  Basing on clinical knowledge I
strongly believe that nightmares also deserve contextual analysis, they
cannot be considered just as one category (without content) in dream
research.  In diagnosis it might be appropriate.  The empirical question
should be about variety, dream features (actors, scenes, controllability,
narrative fragmentation) and function of vicious vs. benevolent dream.

The occurrence of nightmares can be calculated in the data, using one of
the Hartmann criteria, according to which the dream is interrupted.  There
were two indicators of that:
A person spontaneously reports him/herself waking up by the dream: "I
screamed loudly (in the dream) and thereafter I woke up screaming".  We
also explicitly asked the child to report whether he/she woke up in the
night, and why.  I agree that it is a very good idea to analyze more
carefully the nightmare dynamics separately in the comparison and 'trauma'

>... on page 254, the author states 'infrequent dream recall >protected
children from developing depressive symptoms when >they were exposed to
experiences of death and wounding, but >made them more susceptible to
somatic and anxiety symptoms'.
>This sounds like causal reasoning does it not? 

I am aware of the danger of understanding some psychological responses as
'protective' or vulnerable in a cross sectional study setting.  The
moderating role of a variable refers, however, to the fact that the
association between explaining and criterion variables (e.g. trauma and
symptoms) is different in different conditions (e.g. remembering/not
remembering dreams). So, even if it sounds like playing with words, I would
argue that the recall protects (or makes more vulnerable) a child from
negative impact of trauma.

Finally, I agree with Tore, who would like to see descriptive results about
nightmares and bad dream prevalence in these diaries.  Unfortunately, for
objective considerations of space, editors usually omit those few samples.
As you all know, writing for scientific journals often means reducing your
reporting to a limited number of pages, a phenomena which sometimes gives
your paper a 'choppy' feeling.
You wrote so kindly that I hardly can react. Being in science is to be on
constant guard, so for me it easier to defend and argue. I have to say I
never suspected some colleaques' "jealousy", because from our point of view
psychological field work in current political situation in Gaza is
troublesam, using Palestinians as sample is like academic suicide, and this
was my first dream research. It was based very much on clinical stories and
the fact that Middle Eastern people seemed to express their pain naturally
through dreams. I lived a part of my life in Jerusalem, and this explains
the professional contants with Palestinians and Israelis. 
        I would like to continue discussion. And thank for you patiency.  

Raija-Leena Punamaki, Psychologist, Ph. D.
Senior Research at the Finnish Academy of Sciences

Department of Psychology
Division of Applied Psychology
Meritullinkatu 1 B
00014 University of Helsinki

Tel: +358-9-1912 3253   Fax: +358-9-1912 3379
home email: [EMAIL PROTECTED]

Reply via email to