After an unnecessarily long delay, I am back to discuss the second study in the paper about seeing other's problems in your dreams. Before I get into the details of the study, I'd like to mention a few things that initially set off an alarm in my head. First, the researcher proposes to be testing a phenomena but the design of the study is entirely different. From the author's own description of the dream helper ceremony, the dreamers all know the problem of the individual involved and are purportedly trying to direct their dreams toward identifying solutions. This seems to me like another spin on a typical support group as opposed to a psychic experience. It is possible that the dream helper ceremony is poorly described in the paper but I find it more likely that the researcher is merely using this ceremony as a an outside reference point to the internal craziness of the proposed concepts.
In the second study, the researcher attempted to better experimentally manipulate the participants in the study. Whereas before there was a self-selection bias where only participants that claimed to have had a relevant dream submitted their journals for analysis, all participants had to submit their journals. Also, there was an experimental condition where half of the participants were given a picture of a fictional individual though they believed it was a real person. The participants recorded 2 dreams, were shown the picture of the target person. They were told that this person had several life problems but not specifically that they were medical. The participants then recorded 2 more dreams and the dreams were then compared for content.
The new target person was a woman who had a multitude of problems. She had multiple sclerosis, her mother was dying, her husband had died in an industrial accident, her son had been in a car accident, she had been in a car accident where her cousin had died, and her new partner was going through a messy divorce. The experimenter claims to have been blind to the person that was chosen; a friend of the experimenter had volunteered her. This raises a few problems. Whereas in the pilot, the person had 1 problem (breast cancer), this still allowed for the researchers to include a multitude of related codes (torso, limbs, torso, cancer, clinical setting). This person has so many potential problems in her life that a dream about nearly anything could possibly be coded as relevant. Also, if the experimenter's friend proposed this individual, I find it unlikely that the researcher had never heard about any of this person's many problems at some point. Though the researcher may not have realized it, they may have subconsciously considered that the woman with all the problems could be the target and may have directed the students toward her particular kinds of issues.
Additionally, the codes for the person include problems that are not her own, though they are problems she experiences. Her mother's lung cancer, for example, requires a respirator. The researcher proposes that the target's conscious mind is sending out information about her problems that the dreamers are able to connect to and interpret. Why would her mind think about inanimate objects unrelated to the target's problems? It is possible that the respirator is very prevalent but I don't see why it would be sent out as a problem unless the individual was having a problem with the respirator itself. This, as far as I can tell, is not known by the researcher.
Now we get into the results. I have a few problems with the way the analyses were handled. First, the researchers combined the two pre dreams and the two post dreams into a single dream value (one fore pre and one for post), aggregating the codes across the two dreams. The researchers claim that this provides a more conservative estimate because the sample size is lower since the dreams are combined instead of kept separate. This is an irrelevant point. The dreams were about different things and could reflect life events other than a connection to the target individual. Therefore, I feel that a more accurate data treatment would be to keep the values separate. I think aggregating the dreams is probably fine for a final analysis but there is valuable information in comparing the individual dreams pre and post. It makes no sense to aggregate two distinctly different dream types. Additionally, the post treatment dreams should be more similar to one another because they are focused on the target which would give more validity to the results (if they were possibly true)
The researcher compared each code in the dreams separately by both pre and post test values. Therefore, the researcher compares the pre-test values for the torso, head, etc. between the control and the experimental group. He found that there were significant differences for the post-tests for limb problems, breathing problems, and car/driving problems. Note that this is that the mean values in the post-tests were higher for those in the treatment condition than in the control condition. Those that saw a picture of a real woman were more likely to have dreams containing these components than people that saw a fictional person. While this is intriguing, the test the researcher used is rather poor. Since there was a pre-test the experimenter should have used this value as a control. This would wash out the individual characteristics of the dreamers dreams and make a stronger case that the manipulation did something. It would test whether there was a change in the kinds of things the dreamer dreamt instead of just comparing the dreams at each period. As is, the lack of this analysis throws up red flags for me.
From the increased likelihood of the dreamers dreaming about the 3 aspects the experimenter mentioned, he claims that this shows that individuals can accurately dream about other's problems. Though these results could be real, I don't think the interpretation the author makes really stands up. The content of the dreams is not definitively the problems of the target and even the excerpts the experimenter presents in the paper seem vastly different from the real problems of the target person.
The podcasters thought that the willingness of Psychology Today to reprint the findings of this study with little criticism was abhorrent. I honestly do not know why the author was not more skeptical of this piece when it has obvious theoretical and methodological concerns.
In the second study, the researcher attempted to better experimentally manipulate the participants in the study. Whereas before there was a self-selection bias where only participants that claimed to have had a relevant dream submitted their journals for analysis, all participants had to submit their journals. Also, there was an experimental condition where half of the participants were given a picture of a fictional individual though they believed it was a real person. The participants recorded 2 dreams, were shown the picture of the target person. They were told that this person had several life problems but not specifically that they were medical. The participants then recorded 2 more dreams and the dreams were then compared for content.
The new target person was a woman who had a multitude of problems. She had multiple sclerosis, her mother was dying, her husband had died in an industrial accident, her son had been in a car accident, she had been in a car accident where her cousin had died, and her new partner was going through a messy divorce. The experimenter claims to have been blind to the person that was chosen; a friend of the experimenter had volunteered her. This raises a few problems. Whereas in the pilot, the person had 1 problem (breast cancer), this still allowed for the researchers to include a multitude of related codes (torso, limbs, torso, cancer, clinical setting). This person has so many potential problems in her life that a dream about nearly anything could possibly be coded as relevant. Also, if the experimenter's friend proposed this individual, I find it unlikely that the researcher had never heard about any of this person's many problems at some point. Though the researcher may not have realized it, they may have subconsciously considered that the woman with all the problems could be the target and may have directed the students toward her particular kinds of issues.
Additionally, the codes for the person include problems that are not her own, though they are problems she experiences. Her mother's lung cancer, for example, requires a respirator. The researcher proposes that the target's conscious mind is sending out information about her problems that the dreamers are able to connect to and interpret. Why would her mind think about inanimate objects unrelated to the target's problems? It is possible that the respirator is very prevalent but I don't see why it would be sent out as a problem unless the individual was having a problem with the respirator itself. This, as far as I can tell, is not known by the researcher.
Now we get into the results. I have a few problems with the way the analyses were handled. First, the researchers combined the two pre dreams and the two post dreams into a single dream value (one fore pre and one for post), aggregating the codes across the two dreams. The researchers claim that this provides a more conservative estimate because the sample size is lower since the dreams are combined instead of kept separate. This is an irrelevant point. The dreams were about different things and could reflect life events other than a connection to the target individual. Therefore, I feel that a more accurate data treatment would be to keep the values separate. I think aggregating the dreams is probably fine for a final analysis but there is valuable information in comparing the individual dreams pre and post. It makes no sense to aggregate two distinctly different dream types. Additionally, the post treatment dreams should be more similar to one another because they are focused on the target which would give more validity to the results (if they were possibly true)
The researcher compared each code in the dreams separately by both pre and post test values. Therefore, the researcher compares the pre-test values for the torso, head, etc. between the control and the experimental group. He found that there were significant differences for the post-tests for limb problems, breathing problems, and car/driving problems. Note that this is that the mean values in the post-tests were higher for those in the treatment condition than in the control condition. Those that saw a picture of a real woman were more likely to have dreams containing these components than people that saw a fictional person. While this is intriguing, the test the researcher used is rather poor. Since there was a pre-test the experimenter should have used this value as a control. This would wash out the individual characteristics of the dreamers dreams and make a stronger case that the manipulation did something. It would test whether there was a change in the kinds of things the dreamer dreamt instead of just comparing the dreams at each period. As is, the lack of this analysis throws up red flags for me.
From the increased likelihood of the dreamers dreaming about the 3 aspects the experimenter mentioned, he claims that this shows that individuals can accurately dream about other's problems. Though these results could be real, I don't think the interpretation the author makes really stands up. The content of the dreams is not definitively the problems of the target and even the excerpts the experimenter presents in the paper seem vastly different from the real problems of the target person.
The podcasters thought that the willingness of Psychology Today to reprint the findings of this study with little criticism was abhorrent. I honestly do not know why the author was not more skeptical of this piece when it has obvious theoretical and methodological concerns.