Printer Friendly

Bem's "feeling the future" (2011) five years later: its impact on scientific literature.

L'ARTICLE "PRESSENTIR LE FUTUR" DE BEM (2011) CINQ ANS PLUS TARD : SON IMPACT SUR LA LITTERATURE SCIENTIFIQUE

BEM'S "FUHLEN DER ZUKUNFT" FUNF JAHRE DANACH: IHR EINFLUSS AUF DIE WISSENSCHAFTLICHE LITERATUR

EL ARTICULO DE BEM "FEELING THE FUTURE" (2011) CINCO ANOS DESPUES: SU IMPACTO EN LA LITERATURA CIENTIFICA

Even before its publication, in March 2011, in the highly influential Journal of Personality and Social Psychology (JPSP), Daryl Bern's (2011) article "Feeling the Future: Experimental Evidence for Anomalous Retroactive Influences on Cognition and Affect'" had already initiated an intense debate in the (social) media, establishing a remarkable controversy (e.g., Gad-el-Hak, 2011; MSNBC, 2011). The main reason for this was the message conveyed by its findings: "Precognition is scientifically supported" (Duggan, 2015). Although there is great public interest in parapsychological (psi) research, controversy has always been an integral part of this field of study, such that Richard Broughton (1991) titled his book Parapsychology: The Controversial Science. Evidence for psi was reported in the extensive parapsychological literature of recent and past years. What, then, made the article by Bem so unique and interesting? According to Palmer (2015), its particular importance lies in the scientific status of Bem as a social psychologist at the prestigious Cornell University, where he works, and in the fact that in this article he adapted experiments from methodological paradigms that are common in psychology, although reversing the order in which the task stimuli were presented.

Although controversies in science are common and scientists often argue among themselves, this does not mean that scientists must argue or that controversies are essential to the development of science. The history of science shows that some important theoretical changes have been controversial (e.g., Darwinism), whereas other achievements (e.g., electricity-magnetism unification) have not (Machamer, Pera, & Baltas, 2000). There are also controversies that are explicitly avoided. By publishing in a highly regarded journal, and using well established methodological paradigms of mainstream psychology that were reversed, Bem also gave rise to controversy within the scientific community.

In parapsychology, some controversies have been documented and studied. Zingrone (2006) presents an impressive amount of data and an in-depth analysis of controversies throughout its history. Palmer (2015) analyzed Bern's (2011) controversy in some depth, namely the specific criticisms levelled at his studies and the meta-analysis conducted by Tressoldi, Rabeyron, Duggan, and Bem (2016; see also Bem, Tressoldi, Rabeyron, and Duggan, 2015), which supports the early findings of Bem.

The present study aims to complement this more in-depth approach to the controversies in parapsychology through an analysis of texts indexed in a scientific database that cited Bern's (2011) article, using the Iramuteq program for textual data analysis (Ratinaud, 2009). Using exploratory means, Iramuteq employs a quantitative approach that we used in order to understand the impact of this article on the scientific community.

We analyzed three supplementary variables representing the year of publication and subject area. The choice of the supplementary variables resulted from our expectations. Indeed, we did not have any particular predictions regarding the year of publication, but we had some hypotheses with respect to the subject areas in which the sources were indexed: We expected that a more critical point of view on Bern's article would be demonstrated by papers published in sources indexed in the subject area Psychology, and a more positive point of view would be found in texts published in sources indexed in other subject areas (see, e.g., McClenon, 1982; McClenon, Roig, Smith & Ferrier, 2003; McConnell & Clark, 1991; Roe, 2016).

Method

Data Collection

The texts that cite Bern's (2011) "Feeling the Future" were identified from the Elsevier Scopus database for the years 2011 to 2015. The export procedure was performed on August 24, 2015, resulting in the collection of 163 texts, but one that was duplicated was excluded from further analysis. The following information was registered for the 162 remaining texts: (a) title, (b) abstract (when available), (c) authors' keywords (when available), (d) year of publication, (e) authors' names, (f) source, and (g) subject area. A source could be indexed in more than one subject area. Very few errors were detected in the downloaded texts (eight words misspelled, two authors' keywords missing and one incomplete). These errors were manually corrected after consulting the original texts, and one title was translated from Italian to English in order to run the automatic data analysis on a reliable document. Titles, abstracts, authors' keywords, years of publication, and authors' names were then compiled into a single textual corpus, which constituted the material to be analyzed.

Summaries were separated by the supplementary variables necessary for the association of the lexical productions with the years and subject areas of publication. Three supplementary variables were added: first, the year of publication [variable with five modalities (or categories): from 2011 to 2015], and then two different (but complementary) variables for the definition of the subject areas (e.g.. Psychology, Medicine) in which the sources (academic journals, books, conference proceedings) were indexed: (a) a variable with two modalities: in the subject area Psychology, indexation (i) or not (ii), and (b) a variable with three modalities: indexation only in the subject area Psychology (i), or indexation in the subject area Psychology as well as in other subject areas, namely. Arts and Humanities, Medicine, Social Sciences, Mathematics, Neuroscience, Business, Management and Accounting, Decision Sciences, Economics, Econometrics and Finance (ii), or indexation not in subject area Psychology but in other subject areas, which were, in addition to the above-mentioned ones, Physics and Astronomy, Multidisciplinary, Nursing, Agricultural and Biological Sciences, Computer Science, Engineering, Materials Science, and Health Professions (iii). The introduction of this second variable for the subject areas in which the sources were indexed was meant to assess whether the lexical production of the texts included in sources indexed in subject area Psychology and also in other subject areas would modulate the broad difference expected between the texts indexed or not indexed in subject area Psychology.

Data Analysis

To analyze the very large corpus extracted from the database, we used the Iramuteq program (Interface de R pour les Analyses Multidimensionnelles de Textes et de Questionnaires). This program, anchored in R software, enables several different statistical analyses and processing of text (Ratinaud, 2009). In particular, Iramuteq enables the analysis of a corpus using a method developed by Reinert (e.g., 1993), named Alceste (Analyse des Lexemes Cooccurrents dans les Enonces Simples d'un Texte), which has been available for a long time in the software ALCESTE[R]. The method enables the study of the formal structure of the co-occurrence of the words ("the lexical worlds ") in a given text or set of texts: Classes of words are created and associated with supplementary variables likely to explain the conditions of their production. More precisely, the program performs a downward hierarchical classification, extracting (creating) classes of words that co-occur and that are most dissimilar from other classes of co-occurring words; it presents the resulting clustering tree (dendrogram) showing the proximity relationships between the classes; it also provides a list of the most representative words in each class, a list of text segments in which the representative words are included, and the modalities (i.e., the categories) of the supplementary variables associated with the classes. Iramuteq also performs an analysis of correspondence (Benzecri, 1973), which provides a schematic spatial representation of the relationships between the extracted classes, represented by their most representative words. Data analysis was performed using Iramuteq version 07.2, released on December 23, 2014.

Concretely, three steps are taken to perform the analyses. In Step 1 the different parts of a text or, in the case of a questionnaire, respondents' answers to an open-ended question (see, e.g., Poeschl, Valentim, & Silva, 2015), are divided into text segments. The text segments correspond, usually, to a sentence and they are the basis of all computed statistics. At that time, all words are registered to create a dictionary, as well as a dictionary of their lemmatized forms (words reduced to their roots), intended to avoid irrelevant differences (words written in singular or plural, verbs conjugated in different tenses, etc.; Chabchoub, 2008). The text segments are congregated in larger groups of text segments which form the rows of a table. The lemmatized forms form the columns of the table. The intersection of the rows and the columns registers either 1 if the form is present in the text segments or 0 if it is not.

In Step 2, the table is submitted to a downward hierarchical classification, which iteratively separates the text segments into classes in which specific lemmatized forms co-occur and classes in which they do not (Reinert, 1986). This procedure was developed to analyze large logical tables (codified 0 or 1) of scarce frequencies. The first iteration decomposes the original table, made up of all the text segments, into two subtables, one including text segments with mostly 0 in the cells in the original table, and the other including text segments with mostly 1 in the cells. Each of the subtables thus constitutes a class made up of a distinct vocabulary with the overlap of classes as small as possible. The following iterations further decompose the subclasses, increasing the intraclass homogeneity and the interclass differences until the iterations fails to improve the sorting (Reinert, 1986; for more details about the procedure, see also Reinert, 1983, 1985). The procedure results in a tree diagram.

In Step 3, the results that can be interpreted are displayed. The classes presented in the clustering tree are described by the list of their associated forms, which constitutes the classes 'profiles. The profile of a class is the list of the words associated with the class. The degree of association of a form with a class is assessed by applying the chi-square formula on the observed data. An illustration of the relation of Form 1 (Replication) with Class 1 (The Role of Replication in Psychology Research) is presented in Table 1 (Reinert, 1993; see Pelissier, 2016, for further examples). In this example, the chi square of association is significant: [chi square](1, N = 622)= 105.27, p <.0001.

To indicate the direction of the relation between a form and a class (in Table 1, e.g, whether there were, or not, more Form 1 present in Class 1 than in the other classes, or vice versa), the sign of the difference between the observed and the expected frequencies is added to the chi square value, which is then denominated chi square of association. The significance of the chi square value, with one degree of freedom, is assessed and the classes are described by their most representative forms. The larger the chi square value, the stronger the association of the form with the class, and each class is thus characterized by a set of forms (words) more representative of that class than the others (Viaud, Uribe, & Acosta, 2007).

A chi-square of association is also computed to assess the degree of association of the supplementary variables introduced in the corpus with the classes extracted by the program. An illustration of the relation of Modality 2 (year 2012) of Supplementary Variable 1 (Year of Publication) with Class 2 (Experimental Studies of Anomalous Experiences) is presented in Table 2. The chi square of association is significant: [chi square](1, N = 622) = 22.34, p < .0001.

We present hereafter the results obtained from the procedure. More specifically, we describe the profiles of the classes extracted by the downward hierarchical classification performed on the corpus, and their association with the modalities of the supplementary variables. Then we present the results of the correspondence analysis.

Results

The analysis of the corpus with the Iramuteq program was performed on the 162 texts that cited Bern's (2011) "Feeling the Future." In step 1, the corpus was subdivided by the program into 721 text segments containing, altogether, 4438 lemmatized forms. In step 2, the downward hierarchical classification sorted 622 text segments into four classes, classifying 86.27% of the 721 total text segments. The excluded segments contained forms (i.e., words) that were too rare to be taken into consideration (i.e., total frequency less than 3). The analyzed text segments were first divided into two main branches of the clustering tree (or tree diagram) on the basis of the dissimilarity of their vocabulary. Each one was subsequently decomposed further into two classes of words that did not require any further division (see Figure 1).

First Branch of the Dendrogram

We begin by describing the two classes of the first branch of the dendrogram presented in Figure 1, which includes 58.3% of the total text segments: the large Class 1 and the smaller Class 4. Table 3 presents a summary of the forms included in these two classes, that is, their 35 most representative forms, showing the frequency of each form in the class, its total frequency in the analyzed segments, the percentage of the form in the class, and the value of its chi square of association, calculated from a table such as Table 1 (for each value, df= 1;N= 622, p < .001). To make the reading easier, we substituted for each lemmatized form the word most frequently used in the corpus.

Class 1--The Role of Replication in Psychology Research. Class 1, one of the two largest classes, is composed of 195 text segments, that is, 31.4% of the corpus, organized around "replication": [chi square] = 105.27. As may be seen in Table 3, it also includes significant chi squares for the words "psychology" ([chi square] = 88.38),"science" (f = 76.15), "research" ([chi square] = 72.64), "article" ([chi square] = 37.84), "publication bias" ([chi square] = 31.99), "practice" ([chi square] = 31.76), "psychological" ([chi square] = 27.72), "important" ([chi square] = 27.44), "method" ([chi square] = 26.28), "empirical" ([chi square] = 25.96), "education" ([chi square] = 25.18), "field" ([chi square] = 22.72), "publication" ([chi square] = 20.68), "methodological" ([chi square] = 19.67), "recommendation" ([chi square] = 18.46), "improve" ([chi square] = 16.90), "credibility" ([chi square] = 15.50), "validity" ([chi square] = 13.27), "crisis" ([chi square] = 13.25), and "error" ([chi square] = 11.87). This class is characterized by a vocabulary that addresses the role of replication in psychology research. The lack of replication of some studies, along with questionable research practices, may have contributed to a crisis of credibility regarding psychological science. Publication bias is a major concern, and the perceived crisis is also viewed as an opportunity to improve research practices through implementation of recommendations that can increase the credibility of scientific evidence in psychology journals. Issues of validity of empirical findings and errors also characterize the vocabulary of this class.

This first class groups vocabulary that was referred to mostly in texts whose sources are indexed in subject area Psychology (f= 147, [chi square] = 36.96). Significant chi squares were found for only subject area Psychology (f= 68, [chi square] = 13.44); Psychology and Other Subject Areas (f= 79, [chi square] = 9.1); and texts published in 2015 (f= 20, [chi square]= 11.46). The other chi squares are nonsignificant. With regard to the subject areas, we note that, although globally the vocabulary stems from texts whose sources are indexed in subject area Psychology, it is more specifically characteristic of the texts whose sources are indexed only in this area than of the texts whose sources are indexed in the modality Psychology and also other subject areas.

Class 4--Bayesian Statistical Inference. Close to Class 1 in terms of vocabulary, Class 4, the second largest class, includes 167 text segments which correspond to 26.9% of the total segments. This class is structured around "Bayes" ([chi square] = 88.89). As shown in Table 3, it also comprises significant chi squares for the words "factor" ([chi square] = 73.17), "Bayesian" ([chi square] = 64.56), "test" (f = 56.11), "null" ([chi square] = 54.57), "Bayes factor" ([chi square] = 45.99), "statistical" (f = 45.11), "hypothesis" (f = 43.40), "data" (f = 42.59), "p values" ([chi square] = 40.27), "inference" (f = 27.41), "default" ([chi square] = 24.88), "set" (f = 22.46), "analyse" ([chi square] = 21.91), "significance" ([chi square]= 19.88), "frequentisi" ([chi square]= 19.29), "assessment" ([chi square] = 16.91), "effect size" ([chi square] = 16.91), "probability" (f = 16.91), "exploratory" (f = 13.73), and "one-sided" ([chi square] = 13.73). The vocabulary ofthis class points mainly to concepts related with the Bayesian approach. The segments of text mostly reflect the perceived merits of this approach when compared with the more traditional inferential statistics, namely those relying on p values. The words "Bayes" and "Bayesian" are used to express an inferential statistical method used to test hypotheses (null and alternative) in which the notion of probability is different from the frequentisi approach. On the other hand, the specific term "Bayes factor" is a value that results from one Bayesian approach to assessing null values, and it is used to infer which model (null or alternative) is more accurate given the data. With regard to effect size, it relates to both approaches.

Class 4 groups vocabulary that is referred to more in texts whose sources are indexed in subject area Psychology (f = 121, [chi square] = 20.74); Psychology and Other Subject Areas (f= 78, [chi square] = 22.16); and Texts Published in 2011 (f= 51, [chi square] = 14.39).

Second Branch of the Dendrogram

The second branch of the dendrogram presented in Figure 1 includes 41.8% of the text segments, also divided into two classes of unequal weight. Class 2 and the small Class 3. Table 4 presents a summary of the forms included in these two classes, that is, their 35 most representative forms, indicating for each form its frequency in the class, its total frequency in the analyzed text segments, its percentage in the class, and the value of its chi square of association (df= 1, N= 622, p < .001). We also substituted for each lemmatized form the word most frequently used in the corpus, to make the reading easier.

Class 2--Experimental Studies of Anomalous Experience. Class 2 includes 153 text segments, that is, 24.6% of the segments. It is organized around "participant" and "task" ([chi square] = 43.10). It significantly comprises the words "experience" ([chi square] = 34.67),"target" ([chi square] = 31.15), "response" ([chi square] = 29.14), "predict" ([chi square] = 28.44), "learn" ([chi square] = 27.99), "anomalous" ([chi square] = 26.55), "nonintentional" ([chi square] = 24.84), "spontaneous" ([chi square] = 24.84), "stimulus" ([chi square] = 22.40), "performance" ([chi square] = 21.70), "precognition" ([chi square] = 20.43), "subject" ([chi square] = 19.68), "emotional" ([chi square] = 18.57), "prestimulus" ([chi square] = 18.57), "mind" ([chi square] = 18.19), "control" ([chi square] = 17.69), "randomly" ([chi square] = 17.29), "go/nogo" ([chi square] = 15.45), "PMIR" ([chi square] = 15.45), and "remote viewing" ([chi square] = 15.45). This class includes terms from classical empirical research, applied in anomalous experiences studies, and also concepts, methodologies, and theories specific to parapsychological research (e.g., nonintentional precognition task, prestimulus, go/nogo task, remote viewing, and psi-mediated instrumental response--PMIR).

Class 2 groups vocabulary that was referred to mostly in texts whose sources are not indexed in subject area Psychology (f= 88, [chi square]= 18.87); and in Texts Published in 2012 (f= 63, [chi square] = 22.34).

Class 3--Quantum Phenomena and Theories. Class 3, with 107 text segments which correspond to 17.20% of the total analyzed segments, is the smallest of the extracted classes. It is significantly associated with "quantum" ([chi square] = 143.46), "brain" ([chi square] = 73.98), "consciousness" ([chi square] = 57.21), "Orch Or" ([Penrose-Hameroff theory of] orchestrated objective reduction; [chi square] = 39.01), "entanglement" ([chi square] = 37.81), "moment" ([chi square] = 37.81), "system" ([chi square] = 34.52), "retrocausality" ([chi square] = 34.08), "space time" ([chi square] = 34.08), "microtubule" ([chi square] = 29.16), "nonlocality" ([chi square] = 29.16), "superposition" ([chi square] = 29.16), "reality" ([chi square] = 28.12), "mechanic" ([chi square] = 24.26), "paa" (predictive anticipatory activity; [chi square] = 24.26), "causality" ([chi square] = 23.53), "principle" ([chi square] = 23.33), "anticipatory" ([chi square] = 19.38), "EKT" ([analysis of] Echeverria, Klinkhammer, and Thorne; [chi square] = 19.38), "paradox" ([chi square] = 19.38), "backward" ([chi square] = 18.60), and "bilk" ([chi square] = 18.60).

This last class grouped vocabulary that was mainly mentioned in texts whose sources are not indexed in subject area Psychology (f = 82, [chi square] = 61.84); and in Texts Published in 2011 (f= 34; [chi square] = 10.26).

Relationships Between Classes

The relationships of proximity and opposition between the classes may also be observed in the graphic provided by the correspondence analysis performed by the Iramuteq program. As may be seen in Figure 2, the most representative words of the four classes are positioned on two axes: the horizontal axis explains 40.75% of the inertia (i.e., variance) and the vertical axis 30.49%.

Figure 2 confirms, on the horizontal axis, the strong contrast between Classes 1 and 4 (The Role of Replication in Psychology Research, Bayesian Statistical Inference), and Classes 2 and 3 (Experimental Studies of Anomalous Experiences, Quantum Phenomena and Theories). If we take also into consideration the position of the words on the vertical axis, we may further note that the texts on Replication are the most closely related, in terms of vocabulary, with the texts on Bayesian Statistical Inference, and most distant from the texts on Experimental Studies of Anomalous Experiences. On the other hand, the texts on Quantum Phenomena and Theories are the most closely related, in terms of vocabulary, with the texts on Experimental Studies of Anomalous Experiences and most distant from the texts on Bayesian Statistical Inference.

Discussion

The analysis performed on the summaries of published texts with regard to Bern's controversy extracted four classes, which enabled us to identify four types of impact on the scientific literature on Bern's (2011) article "Feeling the Future." If Classes 1 and 4 (The Role of Replication in Psychology Research and Bayesian Statistical Inference) might be considered as coming from a more critical point of view, the data also show that they are close to each other and both are related with the identification of methodological problems and questionable research practices, including proposals for solving the problems. Classes 2 and 3 (Experimental Studies of Anomalous Experiences and Quantum Phenomena and Theories), on the other hand, do not share the same point of view, inasmuch as they are characterized by a specific vocabulary which indicates different approaches to Bern's article.

Data suggest that, subjacent to the Replication class, and to some extent also the Bayesian class, it is possible to consider the question of the authenticity or ontological reality of psi phenomena. Is there evidence or strong evidence that psi is real? The lack of replication of some studies (not only in psi research, but, for example, also in mainstream psychology and medicine), questionable research practices, and the debate around some statistical inference methods, fomented criticisms and negative reactions to Bern's findings. The Bayesian vocabulary was present more in 2011, when the debate began--for instance, in the same issue of JPSP in which the article by Bem appeared, Wagenmakers, Wetzels, Borsboom, and van der Maas (2011) published a paper on Bayesian statistics as a comment on Bem. On the contrary, 2015 was the year in which the question of replication seemed to be most prominent. In that same year, the journal Science listed the transformations made by psychologists in their research efforts (i.e., replications of key studies; creation of new models of scholarly publication and peer review, such as preregistration) as one of the notable scientific breakthroughs of 2015 (Bohannon, 2015).

The technical and complex replication topic, and subtopics such as meta-analysis, have been addressed by some parapsychologists over the years (e.g., Honorton & Ferrari, 1989; Watt & Kennedy, 2015, 2016). On the other hand, controversies between the supporters of frequentisi and Bayesian forms of interpreting probability have a long history in science, with the former seeing probability as a limiting ratio in a sequence of repeatable events, and the latter as a mental construct that represents uncertainty and which applies not directly to events but to our knowledge about them. As argued by Howey (2002), "... a choice between the two interpretations of probability is not forced by pure logic or the mathematics of the situation, but rather depends on the experiences and aims of the individuals involved and their views of the correct form of scientific inquiry" (p. i).

The fact that the Replication and Bayesian classes were represented more in psychology texts confirms our hypothesis that psychology sources have a more critical position concerning Bern's (2011) article. Firstly, it was published in one of the most influential journals in social psychology, using common methodological paradigms of mainstream psychology, which were reversed; secondly, it is well established that psychologists in general have had a strong critical position about the existence of psi phenomena; thus, it was expected that the most critical reaction would come from psychology sources (although the Journal of Parapsychology is also indexed in the subject area Psychology, the relatively small number of articles it published on this topic reduced its impact in the scientific literature indexed in the selected database). When the sources indexed in the subject area Psychology were divided into Only Subject Area Psychology and Psychology and Other Subject Areas, the Bayesian class was represented more in the latter, which is consistent with the usual technical argumentation, particularly in mathematical terms.

Classes 2, Experimental Studies of Anomalous Experiences, and 3, Quantum Phenomena and Theories, underlie two other types of impact of Bern's (2011) article. As pointed out before, the class Experimental Studies of Anomalous Experiences includes terms from classical empirical research, applied in anomalous experiences studies, and also concepts, methodologies, and theories specific to parapsychological research. The data suggest that, for the authors of these texts, the citation of Bern's article was applied in a context of process-oriented research on anomalous experiences (experimental work designed to find evidence about the characteristics of processes that underlie anomalous experiences), as opposed to proof-oriented research, that is, research that aims to find evidence for the existence of psi. This class is represented more in 2012 and in texts whose sources are indexed in subject areas other than psychology, thus confirming the hypothesis that sources from other areas of knowledge might have a more positive position concerning Bern's article.

The Quantum Phenomena and Theories class gives the same result, being more represented in texts whose sources are indexed in other areas of knowledge. In this case, data indicate that quantum theories of brain/consciousness function may leave the door open to the possibility of the existence of psi phenomena. For instance, retrocausality and predictive anticipatory activity (PAA) can be addressed in terms of quantum phenomena. This class, like the Bayesian Statistical Inference class, is represented more in the beginning of the debate (i.e., 2011). Scientists with a favorable position integrated the evidence of retrocausality phenomena as deserving further explanation, whereas the scientists with a more critical position pointed to the necessity for a new method of statistical inference to demonstrate that Bern's (2011) analysis had weaknesses and was not valid. The data trends in the present study suggest that the replication path is currently the most referred to in the attempt to reach closure in this controversy.

References

Bem, D. J. (2011). Feeling the future: Experimental evidence for anomalous retroactive influences on cognition and affect. Journal of Personality and Social Psychology, 100, 407-425.

Bem, D., Tressoldi, P., Rabeyron, T., & Duggan, M. (2015). Feeling the future: A meta-analysis of 90 experiments on the anomalous anticipation of random future events. F1000 Research, 4, 1188. doi: 10.12688/fl000research.7177.1

Benzecri, J.-P. (1973). L'analyse des donnees. Volume II. L'Analyse des Correspondances [Data analysis Volume II. The correspondence analysis]. Paris, France: Dunod.

Bohannon, J. (2015). Reproducibility in psychology. Science, 350(6267), 1459.

Broughton, R. S. (1991). Parapsychology: The controversial science. New York, NY: Ballantine.

Chabchoub, A. S. (2008). Une application de la statistique textuelle a l'exces de confiance des dirigeants: Cas des acquisitions.[An application of the textual statistic to managers' excess of confidence: The case of acquisitions], 9es Journees Internationales d'Analyse Statistique des Donnees Textuelles, 1081-1091.

Duggan, M. (2015). Feeling the future of Bern's findings. Retrieved from http://tanclab.org/blog

Gad-el-Hak, M. (2011, January 7). Room for debate: When peer review falters. New York Times. Retrieved from http:// www.nytimes.com/roomfordebate/2011/01 /06/the-esp-study-when-science-goes-psychic/when-peer-reviewfalters

Honorton, C., & Ferrari, D. C. (1989). "Future telling": A meta-analysis of forced-choice precognition experiments, 1935-1987. Journal of Parapsychology, 53, 281-308.

Howie, D. (2002). Interpreting probability: Controversies and developments in the early Twentieth Century. Cambridge, UK: Cambridge University Press.

Machamer, P., Pera, M.,& Baltas, A. (Eds.) (2000). Scientific controversies: Philosophical and historical perspectives. New York, NY: Oxford University Press.

McClenon, J., Roig, M.. Smith, M. D., & Ferrier, G. (2003). The coverage of parapsychology in introductory psychology textbooks: 1990-2002. Journal of Parapsychology, 67, 167-179.

McClenon, J. (1982). A survey of elite scientists: Their attitudes toward ESP and parapsychology. Journal of Parapsychology, 46, 127-152.

McConnell, R. A., & Clark, T. K. (1991). National Academy of Sciences' opinion on parapsychology. Journal of the American Society for Psychical Research, 85, 333-365.

MSNBC (2011). Professor: Strong evidence ESP is real. Retrieved from http://www.nbcnews.com/video/msnbcnews/40965642

Palmer, J. (2015). Implicit anomalous cognition. In E. Cardena, J. Palmer, & D. Marcusson-Clavertz (Eds.), Parapsychology: A handbook for the 21st century (pp. 215-229). Jefferson, NC: McFarland.

Pelissier, D. (2016). Initiation a la lexicometrie. Approchepedagogique apartir de l'etude d'un corpus avec le logiciel IraMuTeq [Initiation to lexicometric analysis. Pedagogical approach using the study of a corpus with the software IraMuTed]. Retrieved from http://www.iramuteq.org/documentation

Poeschl, G., Valentim. J., & Silva, B. (2015). The thousand shades of the financial and economic crisis: A study of the social representations of the crisis in Portugal. Spanish Journal of Psychology, 18, 1-10. DOI: http://dx.doi. org/10.1017/sjp. 2015.73.

Ratinaud, P. (2009). IRAMUTEQ: Interface de Rpour les Analyses Multidimensionnelles de Textes et de Questionnaires [IRAMUTEQ: Interface of R for the multidimensional analyses of texts and questionnaires] [computer software]. Retrieved from http://www.Iramuteq.org

Reinert, M. (1983). Une methode de classification descendante hierarchique: Application a l'analyse lexicale par contexte [A method of downward hierarchical classification: Application to the lexical analysis by context]. Cahiers de 1'Analyse des Donnees, 3, 187-198.

Reinert, M. (1985). Un logiciel d'analyse lexicale: ALCESTE. [A software for lexical analysis: ALCESTE]. Cahiers de l'Analyse des Donnees, 4, 471-484.

Reinert, M. (1986). Classification descendante hierarchique: Un algorithme pour le traitement des tableaux logiques de grandes dimensions [Downward hierarchical classification: An algorithm for the analysis of logical tables of large dimensions]. In E. Diday et al. (Eds), Data analysis and informatics (pp. 23-28). Amsterdam, Netherlands: Elsevier Science.

Reinert, M. (1993). Les "mondes lexicaux" et leur "logique" a travers l'analyse statistique d'un corpus de recits de cauchemars [The "lexical worlds" and their "logic" through the statistical analysis of a corpus made of accounts of nightmares], Langage et Societe, 66, 5-39.

Roe, C. (2016). What are psychology students told about the current state of parapsychology? Mindfteld, 7, 86-91.

Tressoldi, P., Rabeyron, T., Duggan, M., & Bem, D. J. (2016). Feeling the future: A meta-analysis of 82 experiments on the anomalous anticipation of random future events. F1000 Research. Retrieved from https://fl000research.com/articles/4-1188/v2

Vaud, J., Uribe, J., Acosta T. (2007). Representations et lieux communs de la mondialisation [Representations and commonly accepted views of globalization]. Bulletin de Psychologie, 60, 21-33.

Wagenmakers, E. J., Wetzels, R., Borsboom, D., & van der Maas, H. (2011). Why psychologists must change the way they analyze their data: The case of psi: Comment on Bem (2011). Journal of Personality and Social Psychology, 100, 426-432.

Watt, C., & Kennedy, J. (2015). Lessons from the first two years of operating a study registry. Frontiers in Psychology, 7. DOI: http://dx.doi.org/10.3389/fpsyg.2015.00173

Watt, C., & Kennedy, J. (2016). How prospective meta-analysis could be useful for parapsychology. Abstracts of Presented Papers: The First Combined Convention of the Parapsychological Association (59th) and the Society for Scientific Exploration (35th), p. 49.

Zingrone, N. L. (2006). From text to self: The interplay of criticism and response in the history of parapsychology (Doctoral dissertation, University of Edinburgh). Retrieved from https://www.era.lib.ed.ac.Uk/handle/l 842/7722

Center for Psychology at University of Porto

University of Porto

Porto, Portugal

bruno_silva@fpce.up.pt

Caption: Figure 1. Dendrogram and relative weights of the classes based on the percentage of text segments.

Caption: Figure 2. Correspondence analysis
Table 1

Presence Versus Absence of Form 1 in the Text Segments
of Class 1 Versus Other Classes in the Observed Data

                 Number of text   Number of text   Total
                 segments with    segments with
                 Form 1 present   Form 1 absent

Class 1                55              140          195
Other classes          7               420          427
Total                  62              560          622

Table 2

Presence Versus Absence of Modality 2 of Supplementary
Variable 1 in the Text Segments of Class 2 Versus
Other Classes in the Observed Data

                   Number of text      Number of text     Total
                   segments with        segments with
                 Modality 2 present   Modality 2 absent

Class 2                  63                  90            153
Other classes           102                  367           469
Total                   165                  457           622

Table 3
Class profiles: 35 Most Representative Words
Associated With Class 1 and Class

                        Class 1

Form                 TV in    Total    % in     [chi square]
                     class      N      class

Replication           55       62      88.71       105.27
Psychology            62       81      76.54       88.38
Science               47       57      82.46       76.15
Research              61       86      70.93       72.64
Scientific            31       40      77.50       42.30
Article               30       40      75.00       37.84
Publication bias      16       17      94.12       31.99
Practice              31       45      68.89       31.76
Psychological         24       33      72.73       27.72
Important             14       15      93.33       27.44
Method                29       44      65.91       26.28
Empirical             21       28      75.00       25.96
Education             13       14      92.86       25.18
Field                 16       20      80.00       22.72
Publication           11       12      91.67       20.68
Methodological        12       14      85.71       19.67
Recommendation        10       11      90.91       18.46
Core                   8        8     100.00       17.75
Fact                   8        8     100.00       17.75
Improve               12       15      80.00       16.90
Movement               9       10      90.00       16.24
Credibility            7        7     100.00       15.50
Issue                 12       16      75.00       14.54
Replicability         11       14      78.57       14.84
Journal               14       20      70.00       14.34
Developmental          6        6     100.00       13.27
Guide                  6        6     100.00       13.27
Validity               6        6     100.00       13.27
Crisis                 9       11      81.82       13.25
Investigation          9       11      81.82       13.25
Argue                 16       25      64.00       12.90
Error                  7        8      87.50       11.87
Similar                7        8      87.50       11.87
Finding               17       28      60.71       11.75
Experimental          26       49      53.06       11.65

                        Class 4

Form                 TV in    Total    % in     [chi square]
                     class      N      class

Bayes                 31       31     100.00       88.89
Factor                35       42      83.33       73.17
Bayesian              32       39      82.05       64.56
Test                  41       61      67.21       56.11
Null                  25       29      86.21       54.57
Bayes factor          18       19      94.74       45.99
Statistical           27       36      75.00       45.11
Hypothesis            35       54      64.81       43.40
Datum                 42       71      59.15       42.59
p values              16       17      94.12       40.27
Inference             14       17      82.35       27.41
Default                9        9     100.00       24.88
Set                   17       25      68.00       22.46
Analyse               13       17      76.47       21.91
Provide               14       19      73.68       21.89
Significance          10       12      83.33       19.88
Calorie                7        7     100.00       19.29
Frequentisi            7        7     100.00       19.29
Food                   7        7     100.00       19.29
Analyze                8        9      88.89       17.90
Effect                31       63      49.21       17.84
Obtain                 9       11      81.82       17.23
Evidence              34       72      47.22       17.21
Assessment            11       15      73.33       16.91
Effect size           11       15      73.33       16.91
Probability           11       15      73.33       16.91
Stop                   6        6     100.00       16.51
Yield                  6        6     100.00       16.51
Estimate               8       10      80.00       14.62
Favor                  8       10      80.00       14.62
Confirmatory           5        5     100.00       13.73
Exploratory            5        5     100.00       13.73
One-sided              5        5     100.00       13.73
Small                  5        5     100.00       13.73
Value                 12       19      63.13       13.16

Note. N = number. Critical values of [chi square] with 1
df: 19.51 for p <.00001: 15.14 for p <.0001; 10.83 for p <. 001.

Table 4
Class Profiles: 35 Most Representative
Words Associated With Class 2 and Class 3

                     Class 2

Form              N in     Total    % in     [chi square]
                  class      N      class

Participant        22       29      75.86       43.10
Task               22       29      75.86       43.10
Experience         20       28      71.43       34.67
Target             10       10     100.00       31.15
Response           18       26      69.23       29.14
Predict            12       14      85.71       28.84
Learn               9        9     100.00       27.99
Anomalous          10       11      90.91       26.55
Person             14       19      73.68       25.46
Time               21       35      60.00       25.06
Nonintentional      8        8     100.00       24.84
Spontaneous         8        8     100.00       24.84
Investigate        10       12      83.33       22.76
Stimulus           17       27      62.96       22.40
Performance         7        7     100.00       21.70
Precognition       19       33      57.58       20.43
Subject            11       15      73.33       19.68
Emotional           6        6     100.00       18.57
Prestimulus         6        6     100.00       18.57
Mind               13       20      65.00       18.19
Control            12       18      66.67       17.69
Feel                7        8      87.50       17.29
Randomly            7        8      87.50       17.29
Read                7        8      87.50       17.29
Word                7        8      87.50       17.29
Reward              8       10      80.00       16.82
Belief             13       21      61.90       16.31
Human              12       19      63.16       15.71
Go nogo             5        5     100.00       15.45
Hit                 5        5     100.00       15.45
Trust               5        5     100.00       15.45
Twin                5        5     100.00       15.45
Wander              5        5     100.00       15.45
PMIR                5        5     100.00       15.45
Remote viewing      5        5     100.00       15.45

                     Class 3

Form              N in     Total    % in     [chi square]
                  class      N      class

Quantum            32       35      91.43       143.46
Brain              15       15     100.00       73.98
Consciousness      16       20      80.00       57.21
Conscious          12       15      80.00       42.56
Orch or             8        8     100.00       39.01
Entanglement        9       10      90.00       37.81
Moment              9       10      90.00       37.81
System             14       22      63.64       34.52
Retrocausality      7        7     100.00       34.08
Space time          7        7     100.00       34.08
Microtubule         6        6     100.00       29.16
Nonlocality         6        6     100.00       29.16
Superposition       6        6     100.00       29.16
Reality             7        8      87.50       28.12
Imply               5        5     100.00       24.26
Macroscopic         5        5     100.00       24.26
Mechanic            5        5     100.00       24.26
Paa                 5        5     100.00       24.26
Universe            5        5     100.00       24.26
Basic               6        7      85.71       23.33
Causality           7        9      77.78       23.53
Framework           7        9      77.78       23.53
Fundamental         7        9      77.78       23.53
Principle           6        7      85.71       23.33
Paper              10       17      58.82       21.26
Anticipatory        4        4     100.00       19.38
Biology             4        4     100.00       19.38
Ekt                 4        4     100.00       19.38
Paradox             4        4     100.00       19.38
Nature             10       18      55.56       19.14
Physical            6        8      75.00       19.01
Backward            5        6      83.33       18.60
Bilk                5        6      83.33       18.60
Usual               5        6      83.33       18.60
Causal              8       13      61.64       18.32

Note. N = number. Critical values of [chi square] with 1 df:
19.51 for p < .00001: 15.14 for p <.0001; 10.83 for p < .001.
COPYRIGHT 2017 Parapsychology Press
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2017 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Daryl Bern
Author:Silva, Bruno A.; Poeschl, Gabrielle
Publication:The Journal of Parapsychology
Article Type:Critical essay
Date:Mar 22, 2017
Words:6658
Previous Article:Individual difference correlates of psi performance in forced-choice precognition experiments: a meta-analysis (1945-2016).
Next Article:Anomalous cognition, dissociation, and motor automatisms.
Topics:

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |