Psychological Reports, 1989, 64, 115-123- © Psychological Reports 1989

BELIEFS AND INFERENCES: A TEST OF A RATIONAL-EMOTIVE

HYPOTHESIS: 2. ON THE PROSPECT OF SEEING A SPIDER1

WINDY DRYDEN, JULIA FERGUSON AND SEAN MC TEAGUE

Goldsmiths' College, University of London

Summary.—96 subjects were asked to imagine that they were about to enter a
room in which there may have been one or more spiders. They were also asked to
imagine that (a) they either held a rational or an irrational belief about spiders, (b)
they were about to enter the room either alone or with someone, and (c) that the
room was either dark or light. Having absorbed their assigned role, the subjects were
then asked to make inferences about various elements of their situation. The results
supported the hypothesis that holding an irrational belief leads to more negative infer­
ences. It was also found that the lighting conditions in the room and whether the
subject was alone or with someone affected the negativity of the inferences made. In
addition, there were several two-way and three-way interactions between the inde­
pendent variables which indicated that entering a light room or being with someone
else tended to moderate the negativity of inferences made by those holding a rational
belief rather than the opposite, amplifying the negativity of inferences made by those
holding an irrational belief. The results supported Ellis’s (1985) recent formulation
concerning the complex relationship between events and inferences (A), beliefs (B),
and emotional and behavioral consequences of beliefs (C).

The theory underpinning Rational-Emotive Therapy (RET) has its

foundations in, amongst other things, the Stoic Philosophers’ viewpoint on
the nature of causation of psychological disturbance. This stated that “peo­
ple are not disturbed by things but by their view of things” (Ellis &
Dryden, 1987). It espouses the view that humans are purposeful and goal-
directed, having free will and choice in their affairs. With this humanistic
perspective, RET emphasises the interaction of cognitions, emotions and
behaviors with perception of activating events to produce psychological
health or disturbance. Within this framework, rationality is assumed to be
relative; ‘rational” in RET theory means that which helps people to achieve
their basic goals and purposes, whereas ‘irrational’ means that which pre­
vents them from achieving these goals and purposes (Dryden, 1984). Thus
one of RET’s major contributions to the field of cognitive-behavior therapy
is this distinction between rational and irrational beliefs. The former are
nonabsolute or preferential, are relative and do not obstruct the attainment
of basic goals, whereas the latter are absolute or dogmatic, and are
expressed in the form of ‘musts,’ ‘shoulds,’ and ‘oughts,’ and as such tend
to lead to dysfunctional negative emotions.
The practice of RET is based on the ABC framework (Ellis, 1962)

‘Address correspondence to Windy Dryden, Ph.D., Department of Psychology, Goldsmiths’

College, University of London, New Cross, London SE14 6NW, England.
116 W. DRYDEN, ET AL.

which conceptualises clients’ problems according to activating events and

inferences made about these events (As), beliefs (Bs) about these activating
events, and emotional and behavioral consequences (Cs) of beliefs. An
important aspect of rational-emotive theory concerns the relationship
between irrational beliefs and negatively distorted inferences. Ellis (1977)
has argued that (a) “for the most part . . . you tend, as a human, to make
antiempirical overgeneralisations because you have a hidden MUSTurbatory
agenda in your thinking” (p. 9) and (b) “If you really stayed with desires
and preferences and virtually never escalated them into needs and necessi­
ties, you would relatively rarely make antiempirical statements to yourself
and others. But just as soon as you make your desires into dire needs, such
unrealistic statements almost invariably follow, and follow, frequently in
great numbers” (p. 9). Stated simply, RET theory hypothesises that, if a cli­
ent’s beliefs about certain events are irrational, then these are likely to lead
to negatively distorted inferences which impede the attainment of goals or
purposes and result in psychological disturbance.
Although RET is widely used and discussed, the central hypotheses
have not attracted a great deal of empirical investigation. One study
(Tobacyk & Milford, 1982) has indirectly tested the hypothesis that nega­
tively distorted inferences stem from irrational beliefs by asking subjects to
complete Newmark, Frerking, Cook, and Newmark’s (1979) Irrational
Beliefs Scale and Haney’s (1954) Uncritical Inferences Test. They found
that endorsement of irrational beliefs was negatively associated with scores
on the Uncritical Inference Test, indicating less critical inference making;
however, the inferences under investigation in that study were not directly
related to specific irrational beliefs.
One reason for the lack of direct testing in this area may be that to
elucidate causality it would be necessary to induce beliefs and assess the
consequences; however, in view of the likelihood of negative emotional out­
comes, this is an unethical step to take. To overcome this problem involved
in studying the effect of beliefs in real-life situations, the present study
employed a role-playing paradigm in which subjects were asked to function
‘as if’ they held a rational or an irrational belief. They were asked to imag­
ine doing certain things but were not, in fact, required to do them.
The context for this study involved the possibility of encountering spi­
ders and constitutes a more direct test of the irrational belief-negatively
distorted inference hypothesis in that first, inferences are directly linked to
type of belief and second, that detailed discriminations are made between
rational and irrational beliefs. On this latter point, both irrational and ratio­
nal beliefs given to subjects contained a premise and a derivative from the
premise, e.g., irrational belief: I want you to imagine that you truly believe
that you absolutely must not see a spider (irrational premise) and that it
 BELIEFS AND INFERENCES: TEST OF AN RET HYPOTHESIS 117

would be really terrible if you did (irrational derivative); rational belief: I

want you to imagine that you truly believe that you do not want to see a
spider but that you may (rational premise). It would be bad if you did, but
not the end of the world (rational derivative).
Two situational variables were introduced because persons hold and act
upon beliefs within an environmental context. These were (a) the lighting
conditions in the room and (b) whether or not the subject was alone. For
the former variable, subjects were asked to imagine either that they were
about to enter a darkened room or a lighted room, and for the latter vari­
able, subjects were asked to imagine either that they were about to enter a
room alone or with someone else. Sex of subject was also examined as it is a
commonly held stereotype that women are more afraid of spiders than men.
It was hypothesised that subjects asked to hold irrational beliefs would
make more negative inferences about elements of the situation than subjects
who were asked to hold a rational belief, and although RET does not specif­
ically make predictions about the effect of other types of variables (such as
those described) or predict how they will interact with belief variables, it
was hypothesised here that such variables would have a moderating or
amplifying effect on the negativity of inferences made. For example, if sub­
jects were asked to hold an irrational belief about spiders, entering a lighted
room should moderate the negativity of their inferences, while entering a
dark room should amplify the negativity of their inferences.
Method
Subjects
The subjects were 48 male and 48 female undergraduate students be­
tween the ages of 18 and 21 yr. They were drawn from a wide range of sub­
ject disciplines.
Design
The design was a2x2x2x2 mixed arrangement whereby sub­
jects were randomly assigned to one of eight conditions which were
comprised of all possible combinations of three dichotomous independent
variables. These were (a) whether the subject held a rational or an irrational
belief, (b) whether it was dark or light in the room, (c) whether the subject
was alone or with someone, and the fourth independent variable was sex of
subject.
There were six dependent variables which were subjects’ predictions of
(1) the likelihood of there being a spider in the room, (2) the size of the spi­
der (in inches), (3) the likelihood of them seeing a spider, (4) the distance
between themselves and the spider (in inches), (5) the likelihood of the spi­
der moving towards them, and (6) the number of spiders in the room.
118 W. DRYDEN, ET AL.

Procedure
Subjects were first presented with a role outline corresponding to one
of the conditions, for example, for subjects holding an irrational belief and
entering a dark room alone, the instructions were:
“Thank you for agreeing to take part in this experiment. The purpose
is an investigation of role-playing behavior, and I appreciate you volunteer­
ing to help with it.
“I would like you to imagine that you believe that you must not see a
spider and that it would be terrible if you did. You cannot bear to see a spi­
der of any kind. With this belief you are just about to enter an unfamiliar,
dark and dingy room alone. Can you really try to think yourself into this sit­
uation and that you are about to enter an unfamiliar, dark and dingy room
alone and you can’t bear to see a spider of any kind?”
If the subject answered “No” to this question, they proceeded no fur­
ther with the experiment, and if they answered “Yes,” then they were given
a card with this role outlined on it and asked to spend a little time absorb­
ing it. The subject then completed a questionnaire which consisted of seven
items to be answered using nine-point rating scales. The last item of the
questionnaire asked the subject to what extent they felt they had been able
to adopt their assigned role, and if the subject’s response was below 7 (9, to
a great extent; 0, not at all), then their data were excluded from the analy­
sis. Eighteen subjects did not meet this criterion and were subsequently
replaced. Subjects kept the role card in front of them throughout for refer­
ence purposes.
Results
The data were analysed using SPSS* programs for analysis of variance.
Responses were analysed by item, each pertaining to one of the six depend­
ent variables. 2 x 2 x 2 x 2 analyses of variance were employed to
examine response differences to these variables in relation to the independ­
ent variables—rational or irrational belief, dark or light room, alone or with
someone, and sex of subject.
Belief
There was a significant main effect of belief on all the dependent vari­
ables. Subjects holding an irrational belief (1) predicted a greater likelihood
of there being a spider in the room (M = 8.54, SD = 0.50) than those
holding a rational belief (M = 4.31, SD = 3.00; Flgg = 789.44, p < .001),
(2) estimated a larger size of spider (Al = 4.06, SD = 1.12 irrational;
M = 2.13, SD = 1.36 rational; Fl88 = 6532, p < .001), (3) thought it
more likely that they would see the spider (M = 8.10, SD = 0.01 irratio­
nal; M = 4.13, SD = 2.92 rational; Fl8g = 935.41, p < .001), (4)
predicted less distance between themselves and the spider (Af = 22.15, SD
 BELIEFS AND INFERENCES: TEST OF AN RET HYPOTHESIS 119

= 15.92 irrational; M = 30.88, SD = 18.59; FI88 = 14.69, p < .001), (5)

predicted it to be more likely that the spider would be moving towards
them (Al = 7.73, SD = 0.77 irrational; M = 3.40, SD = 2.35 rational;
Fl88 = 832.00, p < .001), and lastly, (6) if there was more than one spider
in the room, tended to predict greater numbers (M = 28.83, SD = 20.99
irrational; M = 8.04, SD - 8.61 rational; Figg = 51.17, p < .001).
There were also significant two-way interactions between belief and
whether the room was dark or light for four of the dependent variables, (a)
Subjects holding an irrational belief were more likely to predict that there
was a spider in the room regardless of whether it was dark or light (Af =
8.79, SD = 0.41 dark; Af = 8.29, SD = 0.62 light) than subjects holding a
rational belief, who tended to be much less likely to predict a spider’s pres­
ence in a light room (M = 1,50, SD = 0.51) than in a dark room
(Al = 7.13, SD = 1.23; F]88 = 289.83, p < .001). (b) Concerning predic­
tions about seeing the spider, for the subjects holding a rational belief, being
in a lighted room (M = 1.38, SD = 0.58) greatly reduced predictions of the
likelihood of seeing the spider than for those in a dark room (Af = 6.88,
SD - 1.15), whereas for subjects holding an irrational belief, predictions
were relatively unaffected by this variable (Af = 7.50, SD - 0.72 light;
Af = 8.71, SD = 0.55 dark; Fl88 = 272.03, p < .001). (c) Subjects holding
an irrational belief were more likely to predict that the spider would be
moving towards them, regardless of the lighting conditions in the room
(M = 8.00, SD = 0.72 dark; M = 7.46, SD = 0.72 light) than those in
the rational belief condition, for whom entering a lighted room (Af = 1.25,
SD - 0.42) rather than a dark room (Af = 5.54, SD = 1.21) greatly atten­
uated the prediction of this likelihood (FIgg = 155.77, p < .001). (d)
Where predictions about the number of spiders were concerned, subjects
who held an irrational belief on entering a lighted room (Af = 9.08, SD =
3.84) predicted far fewer spiders than those entering a dark room (Af =
48.58, SD = 29.82). For subjects in the rational belief condition, the effect
was similar but not nearly as dramatic (Af = 2.50, SD = 0.78 light; Af =
13.58, SD = 9.31 dark; Fl88 = 23.90, p < .001).
The type of belief held also interacted with sex of subject on three of
the dependent variables; (a) women in the irrational belief condition (Af =
4.33, SD = 1.05) predicted a larger size spider than men in this condition
(Af = 3.79, SD - 1.14), whereas women in the rational belief condition
predicted a smaller size spider (Af = 1.88, SD = 1.23) than did their male
peers (Af = 2.38, SD = 1.44; F(gg = 4.57, p < .05). (b) When holding an
irrational belief, both men (Af = 7.75, SD = 0.94) and women (Af = 8.46,
SD = 0.66) made stronger predictions about the likelihood of seeing a spi­
der in the room than those who held a rational belief, with the men in this
group making slightly stronger predictions (Af = 4.29, SD = 3.01) th an the
120 W. DRYDEN, ET AL.

women (M = 3.96, SD = 2.76; Figg = 16.03, p < .001). (c) Concerning

predictions about the likelihood of the spider moving towards the subject,
those in the irrational belief condition thought it more likely that the spider
would be moving towards them than did the subjects in the rational belief
condition, with women in the former group (M = 8.08, SD = 0.58) think­
ing such movement more likely than the men in that group (M = 7.38, SD
= 0.77), whereas in the latter condition, men thought it slightly more likely
that the spider would be moving towards them (M = 3.54, SD = 2.43)
than did the women (M = 3.25, SD = 2.22; FL88 = 11.08, p < .001). The
final two-way interaction was between belief and whether the subject was
alone or with someone else. In this case, subjects in the irrational belief con­
dition were more likely to predict seeing a spider in the room, regardless of
whether they were alone (Al = 8.00, SD = 0.93) or with someone
(M = 8.21, SD = 0.82) than those in the rational belief condition, for
whom being with someone (Al = 3.79, SD = 2.62) rather than being alone
(Al = 4.46, SD = 3.20) decreased the strength of their predictions (Flgg =
11.31, p < .001).
There was a three-way interaction among belief, sex, and whether or
not the subject was alone with respect to the dependent variable of likeli­
hood of seeing a spider (Flgg = 7.41, p < .01). When alone, subjects in the
irrational belief condition tended to make stronger predictions about the
likelihood of seeing a spider, especially women (M = 8.58, SD = 0.52
women; M = 7.42, SD = 0.90 men), than subjects in the rational belief
condition (Al = 4.17, SD = 2.85 women; M = 4.75, SD = 3.65 men), for
whom there was little difference by sex. When subjects were with someone,
those in the irrational belief condition still made stronger predictions (M =
8.33, SD = 0.78 women; M = 8.08, SD = 0.90 men) than those in the ra­
tional belief condition (M = 3.75, SD = 2.67 women; M = 3.83, SD =
2.69 men), but the sex differences were negligible.
A second three-way interaction was found between belief, whether the
room was dark or light and whether or not the subject was alone and con­
cerned the likelihood of seeing the spider (Flgg = 9.26, p < .01). When the
subjects were alone, those holding an irrational belief and who entered a
dark room made only slightly stronger predictions about seeing the spider
(M = 8.50, SD = 0.67) than those holding a rational belief (M = 7.50, SD
= 1.09); however, being in a lighted room greatly reduced the predicted
likelihood of seeing a spider for those holding a rational belief (M = 1.42,
SD = 0.51) but had little effect on those with an irrational belief (M =
7.50, SD = 0.90). A similar pattern of interaction was seen when the sub­
ject was with someone. In the dark, subjects in the rational condition
predicted that it was less likely that they would see a spider (M = 6.25, SD
= 0.87) than those in the irrational belief condition (M - 8.92, SD =
 BELIEFS AND INFERENCES: TEST OF AN RET HYPOTHESIS 121

0.29), whereas the effect of a lighted room was much greater on those in the
rational belief condition (Af = 1.33, SD = 0.65) than those in the irrational
belief condition (M = 7.50, SD = 0.52), whose strength of predictions did
not greatly decrease.

There were significant main effects of lighting on all the dependent

variables. Subjects who imagined that they entered the room in the dark (a)
predicted a greater likelihood of there being a spider present (Af = 7.96, SD
= 1.24) than those who entered a lighted room (M = 4.90, SD = 3.48;
Fl88 = 413.97, p < .001), (b) predicted a greater size of spider (M = 3.48,
SD = 1.56 dark; M = 2.71, SD = 1.51 light; Flgg = 10.02, p < .01), (c)
predicted a greater likelihood of seeing the spider (M = 7.79, SD = 1.29
dark; M = 4.44, SD = 3.16 light; Fiag = 664.64, p < .001), (d) predicted
a smaller distance between them and the spider (Af = 13.23, SD = 9.85
dark; M = 39.79, SD = 13.45 light; Flgg = 136.04, p < .001), (e) esti­
mated a greater likelihood of the spider moving towards them (Al = 6.77,
SD = 1.59 dark; M = 4.35, SD = 3.19 light; Figg = 258.77, p < .001),
and lastly, (f) they also predicted more spiders in the room (Af = 31.08, SD
= 28.11 dark; M = 5.79, SD = 4.31 light; Flgg = 75.74, p < .001).
There was a two-way interaction between this variable and sex of sub­
ject with respect to the likelihood of seeing a spider (Flgg = 4.33, p < .05).
In the dark, men thought it slightly more likely that they would see a spider
(Af = 7.83, SD = 1.17) than women (Af = 7.75, SD = 1.42). In a lighted
room, however, although both sexes made less strong predictions, the situa­
tion was reversed with women (Af = 4.67, SD = 3.37) who thought it
slightly more likely that they would see a spider than men (Af = 4.21, SD
= 2.99). A second two-way interaction, concerning predictions about the
number of spiders in the room, was found between the lighting conditions
and whether the subject was alone or with someone (Figg = 6.74, p < .05).
Here, entering a lighted room produced predictions of far fewer spiders
both for those alone (Af = 6.21, SD = 4.97) and those who were with
someone (Af = 5.38, SD = 3.50), than those subjects who entered a dark
room alone (Af = 39.04, SD = 34.94) or with someone (Af = 23.13, SD =
16.19). For both cases, predictions were slightly greater if subjects were
alone.
Alone or With Someone
Whether subjects were alone or not had a significant effect on three of
the independent variables. If the subjects imagined that they were entering
the room on their own, they tended to predict a greater likelihood of there
being a spider there (Af = 6.58, SD = 3.11) than if they were with some­
one (Af = 6.27, SD = 2.94; FJ8g = 4.31, p < .05). They were also more
122 W. DRYDEN, ET AL.

likely to think that the spider was moving towards them in the former case
(A4 = 5.79, SD = 2.84) than in the latter case (A4 = 5.33, SD - 2.74; F1S8
= 9.31, p < .01), and they tended to predict more spiders when they were
alone (Al = 22.63, SD = 29.74) than when they were with someone (M =
14.25, SD = 14.66; Fig8 = 8.30, p < .01).
Sex
There were no main effects for sex.
Discussion
The results of this study support the central hypothesis that holding an
irrational belief leads to more negatively distorted inferences than holding a
rational belief. However, two factors must be borne in mind when interpret­
ing these results. First, for ethical reasons previously discussed, the
experimental design employed a role-playing paradigm, so subjects did not in
reality hold the beliefs which they were asked to adopt and may, in fact,
have responded according to their implicit theories about how people in
their position would have acted. To generalise from these results it would be
necessary to repeat the experiment with a clinical population who endorse
real rational and irrational beliefs, and it might be expected in this case that
inferences would be even further negatively distorted. Second, only those
subjects who were clearly able to adopt the role assigned to them were in­
cluded in the analysis. This is also a limiting factor.
There were main effects for the type of belief on all the dependent
variables; in each case subjects holding an irrational belief made more nega­
tive inferences than those holding a rational belief. The results further
showed that either entering a dark room or entering a room alone also led to
more negative inferences being made. There were no main effects for sex of
subject which is surprising considering the common belief that women tend
to have more extreme phobic reactions to spiders than men. However, sex
did interact with belief on three of the dependent variables, although results
were not always in the direction which might have been predicted. It may
be concluded from this that either the sample in this study is atypical of the
population as a whole or that this particular stereotype is a myth.
The other main hypothesis was that the interaction of the belief vari­
able with the other variables would either amplify or moderate the
negativity of inferences. Inspection of the interactions tends to support this
with respect to most of the dependent variables, although the effect tended
to show moderation of the inferences of those holding a rational belief
rather than amplification of the inferences of those holding an irrational be­
lief.
It can be concluded from this study, then, that role playing at holding
an irrational belief does lead to more negatively distorted inferences than
 BELIEFS AND INFERENCES: TEST OF AN RET HYPOTHESIS 123

role playing a rational belief. It may also be concluded that other factors of
a person’s situation affect the negativity of inferences. In this experiment
being on their own and entering a dark room led subjects to make more neg­
ative inferences than being accompanied and entering a lighted room, both
as main effects and in interaction with the belief variable. It should be
noted here that the latter variable, that is, the lighting conditions, had more
effect on the inferences made by subjects than did the former variable.
These results support Ellis’s (1985) recent reformulation of RET’s ABC
framework, where complex interactions between (A) events (and inferences
about those events), (B) beliefs, and (C) emotions and behaviors are put for­
ward.
REFERENCES
Dryden, W. (1984) Rational-emotive therapy: fundamentals and innovations. London: Croom-
Helm.
Ellis, A. (1962) Reason and emotion in psychotherapy. Secaucus, NJ: Lyle Stuart.
Ellis, A. (1977) The basic clinical theory of ration al-emotive therapy. In A. Ellis & R.
Grieger (Eds.), Handbook of rational-emotive therapy. New York: bpringer. Pp. 3-34.
Ellis, A. (1985) Expanding the ABC’s of RET. In M. J. Mahoney & A. Freeman (Eds.),
Cogfftôo# and psychotherapy. New York: Plenum. Pp. 313-323.
Ellis, A., & Dryden, W. (1987) The practice of rational-emotive therapy. New York: Springer.
Haney, W. (1954) Measurement of the ability to discriminate between inferential and descrip­
tive statements. Dissertation Abstracts International, 14, 405-406.
Newmark, C., Frerking, R., Cook, L., & Newmark, C. (1979) Endorsement of Ellis’ irratio­
nal beliefs as a function of psychopathology. Journal of Clinical Psychology, 35, 754-756.
Tobacyk, J., & Milford, G. (1982) Criterion validity for Ellis’ irrational beliefs: dogmatism
and uncritical inferences. Journal of Clinical Psychology, 38, 605-607.

Accepted December 12, 1988.