Vanishing Hitchhikers: Questions of Methodology
DAVID MAIN and SANDY HOBBS
Although folklorists have played the most prominent roles in the development of
“contemporary” or “urban” legends as a field of scholarly study, the subject has also attracted
the interest of other investigators in the humanities and social sciences. Work by researchers
from different scholarly backgrounds may give rise to misinterpretation because of differing
assumptions, conventions and practices. This problem is illustrated by the response of Gillian
Bennett (2013) to our paper (Main and Hobbs, 2012). In our first paragraph we stated we
were “impressed” by her paper and in our conclusion we acknowledged that her thesis that
many Vanishing Hitchhiker texts “tell us that it is bad to die unfulfilled” (Bennett, 2011, 15)
adequately fitted her sample. However, she appears to regard our paper as an attack on hers.
Our aim here in this text is to correct some misunderstandings both by her and by ourselves,
which we believe may at least in part be due to our different scholarly backgrounds.
The first issue which we must confront is one of definition. There are many references
in the scholarly literature to a group of tales which are generally referred to as the Vanishing
Hitchhiker (VH). Bennett and Smith in their Contemporary Legend: A Folklore Bibliography
(1993) cited over 130 works dealing with it. It would seem that the VH term is sometimes
employed as a matter of convenience only, as some of the narratives do not deal with a
“hitchhiker” at all. Bennett (2011) considers a number of these tales and employs the label
“Phantom Hitchhiker”. However, she did not precisely specify how a “Phantom Hitchhiker”
text differs from other Vanishing Hitchhiker stories. On the basis of what she writes in her
response to our paper (Bennett, 2013), we now take it that her use of “phantom” means the
ghost of a dead person which consequently excludes other apparition possibilities, such as an
angel or Christ. However, the term “phantom” can certainly be construed as a broader
concept than a synonym for a ghost. Indeed the Oxford English Dictionary gives “ghost” as
only one of many meanings of “phantom”, for instance, “mental illusion” being amongst the
other potential meanings. Perhaps if Bennett had called her paper "The Revenant Hitchhiker",
her meaning would have been clearer. In suggesting that it would have been preferable for
Bennett to provide a clearer definition in her first paper we are probably displaying the
influence of our experiences in the traditions of empirical psychology. Psychological research
involves the careful collection and sifting of evidence, often involving statistical analysis. For
such processes to be meaningful requires precise specification of what is being studied, in
other words an operational definition. The concept of “operational definition” is discussed in
many texts on psychological research methods (see, for example, Breakwell et al., 2000,
Wilson and MacLean, 2011). The main aspect of it which we wish to stress is that it specifies
procedures which have been followed which allow one to identify and measure the concept
being studied.
In our paper we offered a definition of a Vanishing Hitchhiker story:
1. A protagonist (P) travels in a vehicle on a road.
22
2. P encounters an Other (O) whom they regard as a normal human being.
3. Later, P gains evidence that causes them to undergo a cognitive shift in their
understanding of O. This reappraisal results in reinterpreting the O as a
character who is not a normal human being.
We devised our definition in the belief that no widely agreed definition already existed
in the scholarly literature. Bennett disputes this and presents two quotations which she
apparently considers adequate definitions. One, by Baughman (1966), describes motif
E332.3.3.1 thus:
Ghost of a young woman asks for a ride in automobile, disappears from closed
car without the driver’s knowledge, after giving an address to which she wishes
to be taken. The driver asks person at the address about the rider, finds she has
been dead for some time.
This is followed by two qualifications indicating what the story “often” contains.
Contrast this description of the motif with our definition. We proposed that to decide
whether a text may be included in a sample of this type of tale requires consideration of
several questions. Is the protagonist travelling? Does the protagonist meet someone? Does the
protagonist eventually discover information that the encounter has been a supernatural one?
However, if one were to treat Baughman’s description as a “definition”, one would ask
different questions. Is the person who enters the vehicle a young woman? Does she give an
address? Does she disappear from the car? Does someone at that address reveal that she has
been dead for some time? Answering “no” to any of these questions would lead to the
exclusion of many texts that Beardsley and Hankey have included in their original VH study
(1941).
It seems clear to us that Baughman was not offering a “definition” with the same
purpose as our more psychologically shaped definition. We have attempted to employ terms
which would allow us to distinguish this type of story from other contemporary legend texts
which have at least some similarities with VH types of stories. For instance, “The Devil in
the Dancehall” (see, for example Glazer, 1984) shares with the Vanishing Hitchhiker the
protagonist’s encounter with someone beyond the normal, but differs from it in that the
encounter does not take place while P is travelling in a vehicle.
The “definition” which Bennett (2013) calls “perfectly adequate” would have led to the
exclusion of many of the texts in both her sample and in ours. Less than half of the
hitchhikers in our “Later” (collected post 1941) sample were female and thus do not conform
to Baughman’s description which specifies a “young woman”. Similarly, in her Table on
page 8 (Bennett, 2011) nine of the Hitchhikers are old women (or little old ladies), four are
men and three are a “little girl”.
Jones’s (1944) description is a second example of an “adequate” definition mentioned
by Bennett (2113). This definition is less specific in describing the hitchhiker, but contains
one feature which makes it incompatible as a definition with Baughman’s (1966) description.
23
For Baughman, the hitchhiker disappears from the car (on route to a destination) and the
driver then continues to the requested destination, whereas for Jones (1944) the hitchhiker is
present throughout the journey and only disappears after the destination has been reached.
Like Baughman, Jones does not appear to have had in mind “defining” what constitutes an
example of the VH legend and was merely giving some general guidance to what the term
covers.
On the other hand, Bennett (2013) acknowledges that there have been certain problems
in definition. One of us (Hobbs and Cornwell, 1991) discussed this in a paper on the “Corpse
in the Car” in terms which are similar to what Bennett (2013) states in her later paper. Jones
(1944, 285) gave “The Corpse in the Car” the code “BX”, suggesting it was simply a version
or subset of “The Vanishing Hitchhiker” version B. Yet, confusingly, on the same page he
also said that B and BX were “entirely separate” stories.
In Baughman’s Type and Motif-Index (1966), however, we find the prediction of a
corpse in the car reduced to an example of “other evidence of ghostly nature” which
sometimes replaces the sudden disappearance by a ghostly hitchhiker (Hobbs and Cornwell,
1991, 99).
One of Bennett’s criticisms of our study relates to our inclusion of “Corpse in the Car”
texts. She suggests that we have included them merely for polemical purposes (Bennett,
2013, 8). On the contrary, our decision to include them was a question resolved by examining
texts that were identified via our operational definition of a VH text. If the protagonist of a
story discovers that a stranger encountered has extraordinary prophetic powers we treat this
as falling within our specification of a supernatural encounter. We are surprised that Bennett
(2013) should be so keen to exclude the Corpse in the Car texts as being within the sphere of
VH stories. Not only was it treated as such by Baughman (1966) but by Bennett herself in the
book Urban Legends (Bennett and Smith, 2007). In chapter 9, which deals with “The
Supernatural”, such Corpse in the Car tales are included as a seventh type of Vanishing
Hitchhiker texts, under the label “Double Prophecy”.
Another source of confusion concerns sampling. In her original paper Bennett refers to
her sample as “random” (Bennett, 2011, 5), However, on the basis of what she says in her
response to our paper, we realise that she is not employing a “random sample” in the same
way as experimental psychologists do, where, methodologically, a random sample has a
specific meaning (see for example Wilson and MacLean, 2011). It would appear that rather
than randomly sampling a corpus of texts, Bennett actually “selected” texts for inclusion in
her analysis (Bennett, 2013, 1). The basis for this selection process is not reported. The
inclusion of such details would have been helpful in understanding Bennett’s sample of texts.
In our paper, we tried to establish for the reader what our samples included. First, a text had
to fit our definition. Secondly it fell into the corpus from which the sample was drawn,
comprising three subgroups: texts studied by Beardsley and Hankey (1941), texts reported by
Bonaparte (1947), and an “opportunity” sample. This last category, as we reported in our
previous paper (Main and Hobbs, 2012, 25), was the result of a search in Sandy Hobbs’s
personal files and library to obtain a sample of the same size as that of Beardsley and Hankey
(1941) but containing only texts which had been collected more recently than 1941. Note too
24
that, in deference to the uncertainty amongst folklorists about the relationship between
Beardsley and Hankey texts on the one hand and those of Bonaparte (1947) on the other, we
present our analysis of these sources separately, thus allowing anyone who prefers a different
way of categorising the texts to be able treat them as separate narrative groups.
One of the advantages of presenting an ”operational definition” as we have done is that
it makes clear to all interested parties the key parameters used to select a sample of texts for
inclusion in our study. An operational definition does not close down debate but rather it
allows debate to progress. If a critic does not like an operational definition, he or she may
propose a different one and seek to demonstrate that it allows for a more meaningful
interpretation of the evidence.
We have two further issues to consider concerning sampling. Bennett’s approach to
texts is to distinguish between “full” and “summary” texts, and use only the former. This
seems to us problematic. We must all rely to some extent on other collectors, who do not
always give a clear indication of the relationship between what they actually heard and what
they actually report. We suggest that distinguishing a “summary” from a full text is not a
straightforward matter. It is possible that a collector has reported only a summary of what an
informant has provided, but it is also possible that what appears to us as a “summary”
actually amounts to all that the informant said. Simply because we are aware of fuller
versions of a legend does not in itself justify our labelling a short version a “summary”. The
version may be short for many different reasons. The story as heard by the informant may
have been short. The informant’s memory may be at fault. The informant may choose to tell
the story in that way because of the circumstances. (For example, the informant might cut a
story short if the audience shows signs of having heard it before.) Given that these are all
significant elements in the transmission of legends, and given that we have limited
information on the relationship between what the informant said and what the collector
reports, we do not see how it can be justifiable to treat an apparent “summary” as less worthy
of our attention than a fuller text.
The final issue concerning sampling we have already hinted at. We analysed the texts
on the assumption that they deal with a supernatural encounter. In contrast to the Bennett and
Smith (2007) source already cited, Bennett argues that a double prophecy text she deals with
is not about the supernatural but “what gypsies are traditionally supposed to do” (Bennett,
2013, 7). We suggest that most concepts of the supernatural would include powers of
prophecy beyond the normal range of predictions based on common experience. With this in
mind, we consider gypsy “prophecies” as implicitly invoking a character who possesses a
paranormal ability and as such these narratives fall firmly within the sphere of tales where a
normal protagonist is involved in a supernatural encounter.
We sympathise with Bennett’s unwillingness to become involved in debates about the
interpretation of statistical analysis. However, there is a weakness in her strategy of taking
specific examples as a way of undermining an argument. None of the arguments in our paper
are couched in terms which imply that some feature applies to every case. This, of course, is
true of her arguments too. Not all of her texts deal with the return of a young woman cut
down in her prime. The use of statistical analysis to compare different samples allows us to
25
make judgements about whether a feature of a group is distinctive enough to merit special
explanation.
We believe it may be helpful for us to take this opportunity to restate our position on
the nature of contemporary legends generally, and Vanishing Hitchhiker texts in particular, in
a way which we hope folklorists and those working in other disciplines will find
understandable, even if they disagree with it in whole or in part.
1. There is no scholarly agreement on the definition of a contemporary legend, and no
agreed corpus of texts. We note, for example, that Bennett and Smith in their Urban
Legends (2007, pp. xvi-xviii) do not attempt a definition but offer “guidelines” to
features which commonly occur. It follows almost inevitably that there is no agreed
systematic classification of Contemporary Legends (CL). Indeed, Bennett and
Smith in this work organise their texts quite differently from the classificatory
system proposed by Brunvand in The Baby Train (1992) and revised by him in his
Encyclopaedia of Urban Legends (2012). We set out our ideas relating to issues of
CL classification in our presentation at the Perspectives on Contemporary Legend
conference held in Dublin (Main and Hobbs, 2008).
2. Given the lack of a consistent classification of CL, we have set out to tackle the
problem of definition by an indirect method. This does not seek to be all-inclusive
but to explore an aspect which seems to be common.
3. One particular subset of CL texts which have a distinct structure we term a
Substitute Personal Experience Narrative (SPEN). Its key components are as
follows: the story has a protagonist (P) or protagonists. Story information is given
to the audience in the same order as it was given to P. At or near the end of the
narrative, P and the audience receive some new crucial information which leads the
audience to experience a cognitive shift in understanding the story, i.e. a reappraisal
of how facts or events contained within the story are to be interpreted. As a result of
the cognitive shift in story comprehension, there is also a high likelihood of an
emotional response to the reappraised material by the audience which might
consequently manifest itself behaviourally in the audience’s facial expressions
and/or non-verbal responses for emotions such as disgust, fear or laughter. We
initially reported our ideas on SPEN at the ISCLR conference in Dublin, 2008 and a
more detailed examination of the ideas is provided in Main and Hobbs (2007). (NB
the journal publication date is nominal, as the journal article was written and
published some years after the presentation of the conference paper).
4. We suggested that this structure appears to typically apply to the contemporary
legends generally known as the “Boyfriend’s Death” and the “Surpriser Surprised”,
but it is absent in CL such as “Alligators in the Sewers”. In our more recent paper
(Main and Hobbs, 2012) we also argued that the SPEN structure typically applied to
the Vanishing Hitchhiker. A fuller review of contemporary legends would establish
how common the SPEN structure is and what alternative structures or devices are
typically present in CL narratives that might be collectively termed “non-SPEN”.
26
5. We do not argue that any given legend will always consistently be told with a
SPEN structure. Using a SPEN story structure provides a storyteller with an
effective rhetorical technique, irrespective of whether or not the teller employs it
consciously. The creation of a surprise ending which involves a reappraisal of
previous information is not dissimilar to a punchline in a joke, as the alternative
perspective in what or why something has occurred in the story now takes on a new
light. The verbal “cleverness” of the change in viewpoint and the empathy an
audience can feel for the P in the story as they undergo a cognitive shift in thinking,
can be particularly effective in creating a vivid story. The potential to evoke
emotional responses such as fear, disgust or contempt for a character in the story
may also unconsciously or consciously “tag” the story as being worthy of
memorising for later use by audience members, thus aiding the survival value of
SPEN structured legends.
6. The presence of Non-SPEN versions of CL which are (we believe) normally told
using a SPEN format is probably best explained by two sorts of factor. First, there
are individual differences in audience members’ memory ability (both at the
encoding and recalling stages of memory usage) and to the ability of each
individual to retell a story previously heard. Good storytellers will retain a good
structure but might embellish or revise story information to suit their own style or to
localise the events. A poor storyteller may try to recall many of the key facts of the
story they have heard but during reproduction, fail to give the story a structure that
will recreate the cognitive shift characteristic of a SPEN story. This process is a
verbal performance attribute many people will have encountered when hearing
someone, who is not a good joke teller, failing to properly structure information
they have in memory when retelling a joke they have heard. Secondly, there are
differences in circumstances. In a conversation, an individual may deem it
appropriate to refer to a story they had heard without necessarily thinking it
appropriate to relate the whole story in detail.
7. In itself SPEN obviously cannot account for the survival of legends which are not
generally told with that structure. (We are at present seeking ways of characterising
such non-SPEN legends.) Nor can SPEN be regarded as the sole reason for the
spread of those legends which lend themselves to such a mode of telling. However,
we consider that evaluating basic structural differences within the corpus of CL
narratives is an important facet in building a workable classification system to aid
in future CL research.
8. Data presented in our previous paper (Main and Hobbs, 2012) suggests that
Vanishing Hitchhiker texts are typically, but not invariably, told in the SPEN
format. Vanishing Hitchhiker texts have been widely collected and there are many
possible explanations for its persistence. We suggest that most texts may be
regarded as a storyteller’s attempt at providing “evidence” for the reality of some
supernatural force (the storyteller does not have to believe in such forces, but might
be acting on the conviction that their audience does believe). Specific subcategories
27
of texts may have other more precise uses. For example, “Corpse in the Car”
legends predicting Hitler’s death or the end of the war would have been highly
attractive to audiences in pre-war Europe, or in allied and occupied territories
during the Second World War.
9. Texts where the “hitchhiker” is the ghost of a woman who died young may be
effective because they allow the expression of the poignancy of a life cut short, as
proposed by Bennett (2011). Other VH subcategories provide evidence for specific
religious belief systems, such as Christian, Mormon or Hawaiian folk religion. This
list is not intended to be exhaustive.
We believe that Gillian Bennett has made many valuable contributions to the
understanding of contemporary folklore. She employed an insightful simile in one of her
books where she compared her approach to studying CL to Paul Klee’s consideration of
drawing, as taking a line for a walk (Bennett, 2005, pp. xiv-xv). We are sorry if our “walk”
with SPEN seemed to her to be an invasion of her garden. We consider ourselves as simply
having been digging in the garden next door.
References
Baughman, E. W., Type and Motif-Index of the Folktales of England and North America, The Hague,
Mouton, 1966.
Beardsley, R. K., and Hankey, R., “The Vanishing Hitchhiker”, California Folklore Quarterly, 1
(1942), 303-335.
Bennett, G., Bodies: Sex, Violence, Disease, and Death in Contemporary Legends, Jackson,
University Press of Mississippi, 2005.
Bennett, G., “Phantom Hitchhikers and Bad Deaths”, Tradition Today, 1 (2011), 3-18.
Bennett, G., “Phantoms, Cars and Corpses: A Response to Main and Hobbs”, Tradition Today, 3
(2013), 1-10.
Bennett, G., and P. Smith, Urban Legends: A Collection of International Tall Tales and Terrors,
Westport, Connecticut, Greenwood Press, 2007.
Bonaparte, M., Myths of War, London, Imago, 1947.
Breakwell, G. M., S. Hammond, and C. Fife-Schaw, eds, Research Methods in Psychology, 2nd edn,
London, Sage, 2000.
Brunvand, J. H., The Baby Train, New York, Norton, 1992.
Brunvand, J. H., Encyclopedia of Urban Legends, 2nd edn, Santa Barbara, California, ABC-Clio,
2012.
Glazer, M., “Continuity and Change in Legendry: Two Mexican-American Examples”, in P. Smith,
ed., Perspectives on Contemporary Legend, Sheffield, Centre for English Cultural Tradition
and Language, University of Sheffield, 1984, pp. 108-127.
Hobbs, S., and D. Cornwell, “A Behavior Analysis Model of Contemporary Legend”, Contemporary
Legend, 1 (1992), 93-106.
Jones, L. O., “Hitchhiking Ghosts in New York”, California Folklore Quarterly, 3 (1944), 284-293.
Main, D., and S. Hobbs, “The Substitute Personal Experience Narrative in Contemporary Legend”,
Contemporary Legend, new series, 10 (2007), 38-51.
Main, D., and S. Hobbs, “Classifying Contemporary Legends by their Psychological Function: A New
Look”, paper presented at the 26th Perspectives on Contemporary International Legend
Conference, Dublin, 2008.
28
Main, D., and S. Hobbs, “The Phantom Hitchhiker: An Alternative Approach”, Tradition Today, 2
(2012), 24-32.
Wilson, S., and R. McLean, Research Methods and Data Analysis for Psychology, London, McGrawHill, 2011.
David Main and Sandy Hobbs
Psychology section
University of the West of Scotland, Paisley, Scotland
29