Behat,. Res. Ther. Vol.34, No. 2, pp.

163-173, 1996
~ Pergamon 0005-7967(95)00035-6 Copyright© 1996ElsevierScienceLtd
Printed in Great Britain.All rights reserved
0005-7967/96$15.00+ 0.00

REVISION OF THE PADUA INVENTORY OF OBSESSIVE

COMPULSIVE DISORDER SYMPTOMS: DISTINCTIONS
BETWEEN WORRY, OBSESSIONS, AND COMPULSIONS

G. L E O N A R D BURNS, SUSAN G. K E O R T G E , G I N A M. F O R M E A and

LEE G. S T E R N B E R G E R
Psychology Department, Washington State University, Pullman, WA 99164-4820, U.S.A.

(Received 28 March 1995)

Summary---The Padua Inventory (PI), a self-report measure of obsessive and compulsive symptoms, is
increasingly used in obsessive compulsive disorder (OCD) research. Freeston, Ladouceur, Rheaume,
Letarte, Gagnon and Thibodeau (1994) [Behaviour Research and Therapy, 32, 29-36], however, recently
showed that the PI measures worry in addition to obsessions. In an attempt to solve this measurement
problem, this study used a content distinction between obsessions and worry to revisethe PI. The revision
was constructed to measure five content dimensions relevant to OCD i.e. (I) obsessional thoughts about
harm to oneselfor others; (2) obsessionalimpulsesto harm oneselfor others; (3) contamination obsessions
and washing compulsions; (4) checking compulsions; and (5) dressing/groomingcompulsions. A total of
5010 individuals participated in the study, 2970 individuals completing the PI and the Penn State Worry
Questionnaire (PSWQ) and an additional 2040 individuals completing only the PI. The results provided
support for the reliability and validity of the revision. In addition, the revision of the PI was more
independent of worry, as measured by the PSWQ, than the original PI. Support was thus found for the
validity of the content distinction between obsessions and worry. The importance of this content
distinction is also discussed for the evaluation of other hypothesizeddistinctions between obsessions and
worry.

INTRODUCTION

The Padua Inventory (PI) (Sanavio, 1988) was developed to provide a self-report measure of the
symptoms of obsessive compulsive disorder (OCD). While prior self-report scales of O C D provided
good measures of various checking and contamination compulsions (Steketee, 1994; Van Oppen,
1992), the PI had the advantage of providing two distinct obsessional scales in addition to
traditional checking and contamination scales. Studies in several countries have provided support
for the reliability and validity of the PI (e.g. Burns, Formea, Keortge & Sternberger, 1995; Sanavio,
1988; Sternberger & Burns, 1990, 1991; Van Oppen, 1992). The PI also appears to be one of the
more widely used self-report measures of O C D in research and clinical activities (Steketee, 1994).
Recently, however, a problem was noted with the obsessional subscales of the PI (Freeston,
Ladouceur, Rheaume, Letarte, Gagnon & Thibodeau, 1994). That is, the obsessional subscales,
rather than measuring only obsessions, also appear to measure worry (Freeston et al., 1994). This
problem has implications for the use of the PI, particularly in studies that utilize the PI to examine
the relation between obsessions and worry or the relation between O C D and generalized anxiety
disorder (GAD). In other words, if the PI measures worry in addition to the obsessional aspects
of OCD, then it becomes difficult to determine the relation of obsessions to worry because the
measure of obsessions is also measuring worry.
With the publication of the 4th edition of the Diagnostic and Statistical Manual of Mental
Disorders (DSM-IV) [American Psychiatric Association (APA), 1994], clear definitions were
provided for obsessions and worry. These definitions are relevant to this problem with the PI.
Obsessions are considered "persistent ideas, thoughts, impulses or images that are experienced as
intrusive and inappropriate and cause marked anxiety and distress" (APA, 1994, p. 418).
Obsessions are also considered as ego-dystonic, which means that the content of the obsession is
viewed as alien or inappropriate by the person. Typical content areas for obsessions include
thoughts of contamination, thoughts of harming oneself or others, impulses to harm oneself or
163
164 G. Leonard Burns et aL

others, impulses to perform embarrassing behaviors, images of sexual activity, and need for order
(APA, 1994, p. 418). Worry, as an aspect of GAD, is considered to involve apprehensive
expectation about real-life concerns such as finances, relations, work, school, household chores, and
so on (APA, 1994, p. 433). Obsessions and worry are thus considered to differ on content (Turner,
Beidel & Stanley, 1992; Wells & Morrison, 1994).
This content distinction between worry and obsessions is relevant to the problem of the PI
measuring both worry and obsessions. The content distinction also suggests a solution. For
example, an inspection of the PI reveals a group of items which could be answered on the basis
of worry or obsessions because these items lack any content (Freeston et al., 1994). Examples of
these PI items include: (11) When doubts and worries come to my mind, I cannot rest until I have
talked them over with a reassuring person; (26) I find it difficult to make decisions, even about
unimportant matters; (31) I invent doubts and problems about most of the things that I do; (32)
When I start thinking about certain things, I become obsessed with them; (33) Unpleasant thoughts
come into my mind against my will and I cannot get rid of them; and (44) When a thought or
doubt comes into my mind, I have to examine it from all points of view and cannot stop until I
have done so. These items could be answered on the basis of content relevant to worry (e.g. worry
about relations, finances, work) or on the basis of content relevant to obsessions (e.g. intrusive
thoughts about harm to self, intrusive thoughts about sexual behavior, obsessional
doubt/uncertainty about actions). Nearly all of these types of items occur on the "Impaired Control
over Mental Activities" obsessional subscale of the PI, thus making it unclear the extent to which
this subscale is measuring worry or obsessions (Freeston et al., 1994). Also, since this subscale is
the strongest one of the 4 PI subscales (i.e. accounts for more variance than the other subscales
in factor-analytic studies: Van Oppen, 1992, Table 2), the total score on the PI probably reflects
a significant amount of worry in addition to obsessions and compulsions.
A revision of the PI was needed to eliminate this problem. Since content, as noted above, is one
dimension on which obsessions and worry are hypothesized to differ (Turner et al., 1992), it was
possible to organize 39 of the 60 PI items into content categories relevant to obsessions and
compulsions. Table 1 shows our organization of these 39 PI items into 5 content categories. This
revision will be referred to as the Padua Inventory--Washington State University Revision
(PI--WSUR). The first content area involves "obsessional thoughts about harm to oneself or
others". The 7 items in this category involve intrusive thoughts about having harmed another
person or oneself (e.g. "When I hear about a disaster, I think it is somehow my fault"). The second
content area contains 9 items that measure "obsessional impulses to harm oneself or others" (e.g.
"While driving, I sometimes feel an impulse to drive the car into someone or something"). The
third content area contains 10 items and represents a mixture of obsessions and compulsions about
contamination. For example, some of the items measure contamination obsessions (e.g. Item 1),
others washing compulsions (e.g. Item 7), and some both (e.g. Item 10). The fourth content area
involves checking compulsions, though 2 of the 10 items also measure obsessional uncertainty
associated with checking (Items 27 and 42). Finally, the fifth content area, containing 3 items,
involves dressing/grooming compulsions.* While these 5 content areas do not represent all the
content dimensions for obsessions and compulsions, they do cover many of the dimensions noted
in the DSM-IV (APA, 1994, p. 418). As we will see, the content areas correspond to factorially
distinct dimensions.
Our purpose was thus to determine if this content-based revision of the PI provided a better
measure of obsessions and compulsions. A central aspect of this evaluation was to show that the
revision of the PI was more independent of worry than the original version of the PI. The most
widely used current measure of worry is the Penn State Worry Questionnaire (PSWQ) (Brown,
Antony & Barlow, 1992; Davey, 1993; Meyer, Miller, Metzger & Borkovec, 1990), and so this
measure was used to evaluate the independence of the revised PI from worry. In order for the

*Items were not included in the revision of the PI for the followingreasons: (1) If the item lacked content and thus could
be answered on the basis of either worry or obsessions (e.g. Items 11, 26, 28, 29, 30, 31, 32, 33, 35, 43, 44); (2) If the
item could be answered on the basis of different obsessional content areas and it was thus not clear what category to
place the item (e.g. Items 34, 53); (3) If the item was more central to another mental disorder (e.g. Item 58); and (4)
If the item did not fit in one of the standard obsessional or compulsivecontent categories(e.g. Item 13) or the content
category was represented by a single item (e.g. Item 40).
 Revision of the Padua Inventory 165

Table 1. Factors and items on the PI--WSUR

Obsessional thoughts of harm to self/others factor
36. I imagine catastrophic consequences as a result of absent-mindedness or minor errors which I make.
37. I think or worry at length about having hurt someone without knowing it.
38. When I hear about a disaster, I think it is somehow my fault.
39. I sometimes worry at length for no reason that I have hurt myself or have some disease.
51. I get upset and worried at the sight of knives, daggers, and other pointed objects.
59. When I hear about a suicide or a crime, I am upset for a long time and find it difficult to stop thinking about it.
60. I invent useless worries about germs and diseases.
Obsessional impulses to harm se(f/others factor
46. When I look down from a bridge or a very high window, I feel an impulse to throw myself into space.
47. When I see a train approaching, I sometimes think 1 could throw myself under its wheels.
48. At certain moments, I am tempted to tear off my clothes in public.
49. While driving, I sometimes feel an impulse to drive the car into someone or something.
50. Seeing weapons excites me and makes me think violent thoughts.
53. I sometimes feel the need to break or damage things for no reason.
54. I sometimes have an impulse to steal other people's belongings, even if they are of no use to me.
55. I am sometimes almost irresistibly tempted to steal something from the supermarket.
56. 1 sometimes have an impulse to hurt defenseless children or animals.
Contamination obsessions and washing compulsions factors
1. 1 feel my hands are dirty when I touch money.
2. 1 think even slight contact with bodily secretions (perspiration, saliva, urine, etc.) may contaminate my clothes or somehow
harm me.
3. I find it difficult to touch an object when I know it has been touched by strangers or by certain people.
4. 1 find it difficult to touch garbage or dirty things.
5. I avoid using public toilets because I am afraid of disease and contamination.
6. I avoid using public telephones because 1 am afraid of contagion and disease.
7. I wash my hands more often and longer than necessary.
8. I sometimes have to wash or clean myself simply because I think I may be dirty or 'contaminated'.
9. If I touch something I think is 'contaminated', I immediately have to wash or clean myself.
10. If an animal touches me, I feel dirty and immediately have to wash myself or change my clothing.
Checking compulsions factor
18. I have to do things several times before I think they are properly done.
19. I tend to keep on checking things more often than necessary.
20. I check and recheck gas and water taps and light switches after turning them off.
21. I return home to check doors, windows, drawers, etc. to make sure they are properly shut.
22. I keep on checking forms, documents, cheques, etc. in detail to make sure I have filled them in correctly.
23. I keep on going back to see that matches, cigarettes, etc. are properly extinguished.
24. When I handle money, I count and recount it several times.
25. I check letters carefully m a n y times before posting them.
27. Sometimes I am not sure I have done things which in fact I know I have done.
42. When I read, I have the impression I have missed something important and must go back and reread the passage at least two
or three times.
Dressing~grooming compulsions Jactor
14. 1 feel obliged to follow a particular order in dressing, undressing, and washing myself.
15. Before going to sleep, I have to do certain things in a certain order.
16. Before going to bed, I have to hang up or fold my clothes in a special way.
Note: P I - - W S U R = Padua I n v e n t o r y - - W a s h i n g t o n State University Revision. The item numbers correspond to the item numbers on
the original Padua Inventory.

revision of the PI to be considered a better measure of OCD, the revision was required to meet
several criteria. These criteria included:

1. Each P I - - W S U R item must correlate significantly higher with its own subscale and the total
score on the P I - - W S U R than with the PSWQ.
2. Each PSWQ item must correlate significantly higher with the PSWQ than with P I - - W S U R
subscales and total score on the P I - - W S U R .
3. Each PI---WSUR item must correlate significantly higher with its own subscale on the
P I - - W S U R than with the 4 other P I - - W S U R subscales.
4. The results from a principal components analysis with an a priori 6-factor solution on the
PSWQ and P I - - W S U R items must result in the items loading on the appropriate factors (i.e.
the PSWQ items loading on the worry factor and the P I - - W S U R items loading on the
appropriate factor of the 5 OCD factors--see Table 1).
5. The variance shared by the P I - - W S U R and the PSWQ must be less than the variance shared
by the original PI and the PSWQ.
6. The test-retest correlations for the 5 P I - - W S U R subscales must be significantly higher than
their respective cross-time correlations with the other 4 subscales (e.g. if a subscale has a
higher correlation with itself across time than with the other subscales across time, then such
provides some support for the discriminant validity of the subscale).
166 G. LeonardBurns et al.

7. Individuals with a diagnosis of OCD must have higher scores on the PI--WSUR than the
normative sample.
These outcomes would suggest that the revision of the PI has good psychometric properties. Such
results would provide validity for the dimensions of OCD measured by the PI--WSUR as well as
the independence of these content dimensions of OCD from worry. In addition, support for the
validity of the 5 OCD subscales would result in a measure with more clinical utility since the 5
subscales of the PI--WSUR are more closely tied to content dimensions relevant to OCD than
the original PI.

METHOD
Measures
Penn State Worry Questionnaire. The Penn State Worry Questionnaire (PSWQ) is a 16-item
self-report measure of worry (Meyer et al., 1990). Each item is rated on a 5-point scale (1 = "not
at all typical" to 5 = "very typical"). Several studies indicate that the PSWQ provides a reliable
and valid measure of general worry in patient and nonpatient samples (Brown et al., 1992; Brown,
Moras, Zinbarg & Barlow, 1993; Davey, 1993; Ladouceur, Freeston, Dumont, Letarte, Rheaume,
Gagnon & Thibodeau, 1992).
Padua Inventory--Washington State University Revision. The PI--WSUR is a 39-item self-report
measure of obsessions and compulsions (Burns, 1995). Each item is rated on a 5-point scale
according to the degree of disturbance caused by the thought or behavior (0 = "not at all" to
4 = "very much"). As indicated in Table 1, the PI--WSUR items were a priori organized to
measure 5 content areas relevant to OCD. These 5 areas are Obsessional Thoughts about Harm
to Self/Others (OTAHSO), Obsessional Impulses to Harm Self/Others (OITHSO), Contamination
Obsessions and Washing Compulsions (COWC), Checking Compulsions (CHKC), and Dressing/
Grooming Compulsions (DRGRC). Table 1 shows the items hypothesized to be associated with
each content area.

Subjects and Procedures

A total of 5010 native English speakers (2364 men and 2646 women) participated in the study.
Of these 5010 individuals, 2970 completed the PI and the PSWQ. The mean age of this group of
2970 individuals was 19.34 (SD = 2.29, range 17-52) with 1328 of these participants being men and
1642 being women. The ethnicity of this group varied as follows: 3.7% Asian-American, 2.2%
Black, 88.6% Caucasian, 0.1% Eskimo, 1.4% Hispanic, 1.9% Native American, and 2.2% Other.
A second group of 2040 individuals completed only the PI. The mean age of these 2040 individuals
was 19.06 (SD = 1.91, range 17-56) with 1036 being men and 1004 being women. The ethnicity
of this second group varied as follows: 3.3% Asian-American, 1.5% Black, 89.7% Caucasian, 0.1%
Eskimo, 1.5% Hispanic, 1.9% Native American, and 1.9% Other. The Ss were enrolled in
introductory psychology courses at Washington State University.

RESULTS
Discriminant validity of P I - - W S U R items with P S W Q
Table 2 shows the corrected item-total correlations between each PI--WSUR item and its own
subscale, the corrected item-total correlations between each PI--WSUR item and the total score
on the PI--WSUR, and the correlations between each PI--WSUR item and the PSWQ. As a first
step in establishing the discriminant validity of the revision of the PI, each PI--WSUR item was
required to have a significantly higher correlation with its own subscale and total score on the
PI--WSUR than with the PSWQ. Fisher r-to-Z scores for dependent correlations were used to test
these predictions (Meng, Rosenthal & Rubin, 1992). All of the PI--WSUR items met this criterion.
Each PI--WSUR item had a significantly higher correlation with its own subscale than with the
PSWQ. In addition, each PI--WSUR item had a significantly higher correlation with the total score
on the PI--WSUR than with the PSWQ (P < 0.0006, Bonferroni correction). These results thus
 Revision o f the P a d u a I n v e n t o r y 167

Table 2. Corrected item-total correlations of P I - - W S U R

items with the item's subscale, corrected item-total
correlations of P I - - W S U R items with P I - - W S U R total
score, and correlations of P I - - W S U R items with PSWQ
PI--WSUR PI--WSUR PI--WSUR
items subscales total PSWQ
OTAHSO subscale
36. 0.53 0.54 0.28
37. 0.53 0.49 0.29
38. 0.47 0.40 0.17
39. 0.53 0.48 0.23
51. 0.38 0.36 0.19
59. 0.45 0.42 0.24
60. 0.48 0.48 0.21
OITHSO subscale
46. 0.54 0.40 0.09
47. 0.59 0.41 0.11
48. 0.41 0.25 0.00
49. 0.64 0.40 0.07
50. 0.52 0.28 -0.02
53. 0.58 0.42 0.08
54. 0.52 0.32 0.02
55. 0.47 0.32 0.04
56. 0.42 0.25 0.01
CO WC subscale
1. 0.44 0.33 0.14
2. 0.52 0.46 0.11
3. 0.62 0.51 0.17
4. 0.57 0.44 0.14
5. 0.54 0.41 0.12
6. 0.55 0.43 0.14
7. 0.60 0.51 0.17
8. 0.59 0.51 0.15
9. 0.55 0.44 0.10
10. 0.51 0.42 0.16
CHKC subscale
18. 0.60 0.58 0.26
19. 0.69 0.62 0.33
20. 0.65 0.56 0.23
21. 0.62 0.54 0.21
22. 0.64 0.51 0.20
23. 0.56 0.50 0.17
24. 0.60 0.55 0.18
25. 0.64 0.55 0.24
27. 0.56 0.56 0.28
42. 0.47 0.48 0.24
DRGRC subscale
14. 0.61 0.50 0.19
15. 0.72 0.49 0.22
16. 0.56 0.43 0.18
Note: N = 2970. Each PI--WSUR item correlated
significantly higher with its own subscale as well as with
the P I - - W S U R total score than with the PSWQ
(P < 0.0006, Bonferroni correction). OTAHSO =
Obsessional Thoughts about Harm to Self/Others;
OITHSO = Obsessional Impulses to Harm Self/Others;
COWC = Contamination Obsessions and Washing
Compulsions; CHKC = Checking Compulsions;
DRGRC = Dressing/Grooming Compulsions; P I - -
WSUR = Padua Inventory--Washington State
University Revision; PSWQ = Penn State Worry
Questionnaire.

indicated that each P I - - W S U R item was more strongly related to its particular aspect of OCD
(i.e. its own subscale) as well as to OCD in general than to the measure of worry.

Discriminant validity of PSWQ items with PI--WSUR scales

The next step in establishing the discriminant validity between worry and OCD as measured by
these instruments was to show that each PSWQ item had a significantly higher corrected item-total
correlation with the PSWQ than with the P I - - W S U R subscales and P I - - W S U R total score. The
results supported this requirement with each PSWQ item having a significantly higher correlation
with the PSWQ than with the 5 subscales of the P I - - W S U R and the total score on the P I - - W S U R
(P < 0.0005, Bonferroni correction). Table 3 shows these results.
168 G . L e o n a r d B u r n s et al.

Table 3. Corrected item-total correlation of PSWQ items with PSWQ and correlations of PSWQ items
with P I - - W S U R
P I - - W S U R subscales
PSWQ PI--WSUR
items PSWQ OTAHSO OITHSO COWC CHKC DRGRC total
I. 0.42 0.11 -0.06 0.04 0.12 0.10 0.09
2. 0.68 0.31 0.11 0.20 0.26 0.17 0.29
3. 0.59 0.18 0.01 0.10 0.17 0.12 0.16
4. 0.67 0.27 0.07 0.17 0.26 0.17 0.26
5. 0.74 0.29 0.06 0.15 0.27 0.16 0.26
6. 0.65 0.24 0.01 0.15 0.22 0.15 0.22
7. 0.72 0.33 0.13 0.21 0.31 0.23 0.33
8. 0.58 0.16 -0.01 0.07 0.16 0.09 0.14
9. 0.62 0.27 0. I 0 0.20 0.28 0.22 0.30
10. 0.56 0.15 -0.02 0.08 0.13 0.10 0.12
1 I. 0.43 0.16 0.02 0.07 0.13 0.08 0.13
12. 0.68 0.27 0.05 0.14 0.25 0.16 0.24
13. 0.70 0.27 0.06 0.13 0.25 0.15 0.24
14. 0.69 0.35 0.13 0.20 0.30 0.20 0.33
15. 0.73 0.35 0.13 0.20 0.32 0.21 0.33
16. 0.63 0.26 0.05 0.21 0.29 0.22 0.29
Note: N = 2970. Each PSWQ item correlated significantly higher with the PSWQ than with the 5 P I - - W S U R
subscales and the P I - - W S U R total score (P < 0.0005, Bonferroni correction). PSWQ = Penn State
Worry Questionnaire; OTAHSO = Obsessional Thoughts about Harm to Self/Others;
OITHSO = Obsessional Impulses to Harm Self/Others; COWC = Contamination Obsessions and
Washing Compulsions; C H K C = Checking Compulsions; D R G R C = Dressing/Grooming Compulsions;
P I - - W S U R = Padua Inventory--Washington State University Revision.

Discriminant validity of P I - - W S U R items with P I - - W S U R subscales

While the results from Tables 2 and 3 indicated that each OCD item was more strongly
related to a measure of OCD than to worry and each worry item was more strongly related to
the measure of worry than to the measures of OCD, it was also important to show that each
OCD item had a stronger relation with its own aspect of OCD (i.e. its particular subscale)
than with the other aspects of OCD (i.e. the other subscales). The results supported this
requirement, with each P I - - W S U R item having a significantly higher correlation with its own
subscale than with the other 4 P I - - W S U R subscales (P < 0.0003, Bonferroni correction). Table 4
shows these results. At an item level, therefore, each P I - - W S U R item showed discriminant validity
with the construct of worry as measured by the PSWQ as well as with the different OCD content
areas.

Factorial composition of P I - - WSUR

To evaluate the hypothesized factorial composition of the PI--WSUR, a principal components
analysis with varimax rotation was first performed on the sample of 2970 individuals. Six factors
were a priori retained for rotation given the expectation that 1 worry factor and 5 0 C D factors
would be found in the data. Each PSWQ item loaded on the first factor with none of these 16 items
having a loading greater than 0.20 on the 5 0 C D factors. In addition, each of the P I - - W S U R items
loaded on its hypothesized factor as predicted in Table 1. Finally, only 3 of the P I - - W S U R items
had loadings greater than 0.30 on a factor other than the primary or hypothesized factor (i.e. Item
27 had a loading of 0.31 on the OTAHSO factor, Item 46 had a loading of 0.31 on the OTAHSO,
and Item 60 had a loading of 0.31 on the COWC factor).
A second principal components analysis with an a priori 5-factor solution was next performed
on the second sample of 2040 individuals, the individuals who had completed only the PI--WSUR,
to attempt to replicate the results from the first principal components analysis for the PI--WSUR.
In this analysis each item again loaded on its hypothesized factor with loadings very similar to those
in the first analysis. There were 4 items, however, that had loadings greater than 0.30 on a
nonprimary factor (i.e. Item 27 had a loading of 0.33 on the OTAHSO factor, Item 42 a loading
of 0.31 on the OTAHSO factor, Item 36 a loading of 0.33 on the C H K C factor, and Item 60 a
loading of 0.37 on the COWC factor). While all the P I - - W S U R items in the first and second
principal components analysis loaded on their hypothesized factors, 2 of the items loaded slightly
greater than 0.30 on secondary factors in both samples (Items 27 and 60). This result makes sense
given the content of these items, especially Item 60 (see Table 1). In summary, however, the two
principal component analyses provided initial support for the proposed factorial structure of the
 R e v i s i o n o f the P a d u a I n v e n t o r y 169

Table 4. Corrected item-total correlations of each P I - - W S U R item with its own subscale and
correlations of each P I - - W S U R item with the other P I - - W S U R subscales
P I - - W S U R items P I - - W S U R subscales
O T A H S O items OTAHSO OITHSO COWC CHKC DRGRC
36, 0.53 0.39 0.27 0.46 0.30
37, 0.53 0.30 0.26 0.44 0.25
38, 0.47 0.28 0.26 0.30 0.18
39, 0.53 0.32 0.33 0.36 0.25
51. 0.38 0.17 0.28 0.28 0.21
59. 0.45 0.24 0.28 0.34 0.21
60. 0.48 0.25 0.39 0.37 0.25
OITHSO items OITHSO OTAHSO COWC CHKC DRGRC
46. 0.54 0.36 0.15 0.28 0.18
47. 0.59 0.35 0.18 0.26 0.17
48. 0.41 0.21 0.10 0.14 0.08
49. 0.64 0.33 0.14 0.27 0.14
50. 0.52 0.17 0.09 0.17 0.13
53. 0.58 0.33 0.17 0.32 0.16
54. 0.52 0.26 0.11 0.19 0.15
55. 0.47 0.26 0.14 0.19 0.16
56. 0.42 0.18 0.10 0.13 0.11
COWC items COWC OTAHSO O1THSO CHKC DRGRC
1. 0.44 0.23 0.05 0.21 0.23
2. 0.52 0.33 0.17 0.33 0.24
3. 0.62 0.35 0.18 0.35 0.27
4. 0.57 0.27 0.11 0.30 0.29
5. 0.54 0.26 0.12 0.29 0.23
6. 0.55 0.29 0.15 0.28 0.24
7. 0.60 0.35 0.15 0.39 0.30
8. 0.59 0.36 0.21 0.36 0.29
9. 0.55 0.27 0.14 0.34 0.24
10. 0.51 0.28 0.10 0.30 0.29
C H K C items CHCK OTAHSO OITHSO COWC DRGRC
18. 0.60 0.41 0.23 0.36 0.42
19. 0.69 0.43 0.27 0.38 0.4 I
20. 0.65 0.38 0.21 0.35 0.34
21. 0.62 0.38 0.21 0.36 0.29
22. 0.64 0.33 0.19 0.30 0.32
23. 0.56 0.38 0.20 0.31 0.25
24. 0.60 0.35 0.29 0.34 0.30
25. 0.64 0.38 0.23 0,33 0.34
27. 0.56 0.48 0.33 0,30 0.32
42. 0.47 0.39 0.25 0,31 0.29
D R G R C items DRGRC OTAHSO OITHSO COWC CHKC
14. 0.61 0.33 0.24 0,35 0.41
15. 0.72 0.32 0.18 0~34 0.41
16. 0.56 0.27 0.13 0,31 0.37
Note: N = 2970. Each P I - - W S U R item correlated significantly higher with its own subscale than
with the other P I - - W S U R subscales (P < 0.0003, Bonferroni correction). PSWQ = Penn State
Worry Questionnaire; O T A H S O = O b s e s s i o n a l Thoughts about Harm to Self/Others;
OITHSO = Obsessional Impulses to H a r m Self/Others; COWC = Contamination Obsessions
and Washing Compulsions; C H K C = Checking Compulsions; D R G R C = Dressing/Grooming
Compulsions.

P I - - W S U R . Subsequent studies should use confirmatory factor analysis to further test the
adequacy of this proposed model for the factorial structure of the P I - - W S U R . *

Independence of worry from obsessions and compulsions

The results so far have provided support for the discriminant validity of the items in the 5
P I - - W S U R subscales as well as support for the hypothesized factorial composition of the
P I - - W S U R . The next step was to evaluate the independence of worry, as measured by the PSWQ,
from obsessions and compulsions, as measured by the P I - - W S U R . The expectation was that the
revision would result in a measure of obsessions and compulsions which was more independent
of worry than the original PI.
While the PSWQ was significantly correlated with each of the P I - - W S U R subscales and total
score (P < 0.001, see Table 5), the more critical issues was the amount of variance the measure
of worry shared with the measures of OCD. Earlier research (Freeston et al., 1994, Table 1) found
that the PSWQ and the original PI had 34% of their variance in common. As indicated in Table 5,
the current study found that PSWQ and the PI---WSUR had 12% of their variance in common.

*The results f r o m these t w o p r i n c i p a l c o m p o n e n t s a n a l y s e s a r e a v a i l a b l e f r o m the first a u t h o r .

BRT 34/2--F
170 G. Leonard Burns et al.

Table 5. Correlations between PSWQ

and P I - - W S U R subscales and total
score
P I - - W S U R subscales PSWQ
OTAHSO 0.37a
CHCK 0.34a
DRGRC 0.23 b
COWC 0.21 b
OITHSO 0.08¢
P I - - W S U R total 0.34
Note: N = 2970. All correlations are sig-
nificant (P < 0,001). Correlations
with different superscripts differ sig-
nificantly (P < 0.005, Bonferroni
correction). PSWQ = Penn State
Worry Questionnaire; O T A H $ O =
Obsessional Thoughts about
Harm to Self/Others; OITHSO =
Obsessional Impulses to Harm Self/
Others; COWC = Contamination
Obsessions and Washing Compul-
sions; C H K C = Checking Compul-
sions; D R G R C = Dressing/
Grooming Compulsions; PI--
WSUR = Padua Inventory--Wash-
ington State University Revision.

Our revision of the PI thus reduced the amount of overlap between worry as measured by the
PSWQ and OCD as measured by the PI--WSUR. Our revision of the PI also reduced the overlap
between the different aspects of OCD and worry. For example, the PI--WSUR subseale which
measures OTAHSO shared 14% of its variance with the PSWQ. In the original version of the PI,
the Impaired Control of Mental Activities subscale shared 45% of its variance with the PSWQ
(Freeston et al., 1994, Table 1). The removal of the PI items which could be answered on the basis
of worry or obsessions thus improved the distinctiveness of worry as measured by the PSWQ and
intrusive thoughts of an ego-dystonic nature as measured by this new scale of the PI--WSUR. In
addition, the PI--WSUR subscale which measures OITHSO shared less than 1% of its variance
with the PSWQ. The original version of this subseale shared 12% of its variance with the PSWQ
(Freeston et al., 1994, Table 3). The revision of the PI thus resulted in a measure that was more
independent of worry than the original version of the PI.
Table 5 also shows that the OTAHSO and CHKC subscales were more strongly related to the
PSWQ than were the DRGRC, COWC, and the OITHSO subscales. In turn, the DRGRC and
the COWC subscales were more strongly related to the PSWQ than the OITHSO subscale
(P < 0.005, Bonferroni correction). Obsessional thoughts about harm to self/others and checking
compulsions were thus the two content areas of OCD that were most strongly related to worry
(see also Freeston et al., 1994; Tallis & de Silva, 1992). Obsessional impulses to harm oneself or
others, in turn, shared virtually no variance with the measure of worry.

Test-retest stability and discriminant validity of P I - - W S U R subscales

Table 6 shows the test-retest correlations for the 5 PI--WSUR subscales for a 6 to 7-month
interval. Of 700 individuals who completed the PI in the spring semester, 200 were randomly
selected to complete the PI a second time the subsequent fall semester. We were able to locate 176
of these individuals. The test-retest correlation for the total PI--WSUR was 0.76 with the values
for the 5 subscales varying from 0.61 for the OTAHSO subscale and 0.84 for the OITHSO subscale.
Van Oppen (1992) reported a test-retest correlation for an 8 week interval of 0.79 for the original
PI. The 5 subscales and total score of the PI--WSUR thus showed adequate levels of stability,
especially given that the interval was 6 to 7 months.
Additional evidence for the discriminant validity of the 5 subscales is obtained by a comparison
of the test-retest correlations for the 5 subscales with their cross-time correlations with the other
4 subscales (e.g. a subscale was expected to have a higher correlation with itself across time than
with the other 4 subscales across time). Each test-retest correlation was significantly higher than
the 8 across time correlations involving the particular test-retest correlation with one exception
(P < 0.006, Bonferroni correction). The test-retest correlation of 0.61 for the OTAHSO subscale
 Revision o f the P a d u a I n v e n t o r y 171

Table 6. Test-retest and cross time correlations for the PI--WSUR subscales for a
6- to 7-month interval
Time 2

Time 1 OTAHSO OITHSO COWC CHKC DRGRC

OTAHSO 0.61 0.31 0,30 0.50 0.37
O1THSO 0.37 0.84 0.27 0.41 0.26
COWC 0.17 0.14 0,72 0.31 0.23
CHCK 0.46 0.45 0.47 0.74 0.5 l
DRGRC 0.33 0.17 0.34 0.41 0.67
Note: N = 176. Bold values are the test-retest coefficients. Each test-retest corre-
lation was significantly higher than the eight across year correlations involving
the particular test-retest correlation (P < 0.006, Bonferroni correction) with one
exception. The test-retest correlation for OTAHSO (0.61) was not significantly
higher than the correlation of OTAHSO (Time 1) with CHCK (Time 2) (0.50,
P <0.01). OTAHSO=Obsessional Thoughts about Harm to Self/Others;
OITHSO = Obsessional Impulses to Harm Self/Others; COWC =
Contamination Obsessions and Washing Compulsions; C H K C = C h e c k i n g
Compulsions; D R G R C = Dressing/Grooming Compulsions. The test-retest cor-
relation for the total scale was 0.76.

was not significantly higher than the correlation of 0.50 between OTAHSO at Time 1 and the
CHKC subscale at Time 2 (P = 0.01). With this one exception, however, these results provided
some evidence for the discriminant validity of the subscales.

Norms and subscale correlations for the P I - - W S U R

Table 7 provides the normative data on the PI--WSUR for the sample of 5010 individuals. Each
of the 5 subscales had a high level of internal consistency as measured by Cronbach's 0c(0.77-0.88)
with the ct for the total scale being 0.92. These values, along with the evidence for the discriminant
validity and stability of the subscales, suggest that the 5 subscales can be used for clinical purposes.
The 2 subscales which shared the greatest amount of variance, 32%, were the OTAHSO and the
CHKC scales. The content dimensions of obsessional thoughts about harm to self/others and
checking compulsions thus had the most variance in common. The OITHSO subscale, in turn,
shared the least amount of the variance with the other 4 subscales, especially the COWC (5%),
CHKC (13%), and DRGRC (5%) scales (see Table 7).
The 5 subscales and total score on the PI--WSUR were also examined for gender differences.
Men and women did not differ significantly on the total score and the checking subscale. Women
scored significantly higher than men on the OTAHSO (3%), COWC ( < 1%), the DRGRC ( < 1%)
subscales with men scoring higher than women on the OITHSO (3%) subscales. Since the
statistically significant gender differences accounted for such a small percentage of the variance in
these 4 subscales (i.e. the number in parentheses after each scale), norms are presented collapsed
across gender. In addition, the correlations between the 5 scales for men and women were very
similar (i.e. no difference in correlation value was greater than 0.05 with the average difference being
0.03 for the 10 comparisons).

P I - - W S U R scores and OCD diagnosis

Table 7 also shows the PI--WSUR means for 15 individuals who met the DSM-III-R diagnostic
criteria for OCD on the basis of the Anxiety Disorders Interview Schedule--Revised (Dinardo &

Table 7. Alphas, means, standard deviations, and intercorrelations for the PI--WSUR for the
normative sample (N = 5010) and means and standard deviations for an OCD group (N = 15)
Normative sample OCD group

PI--WSUR ct M SD 2. 3. 4. 5. M SD
1. OTAHSO 0.77 2.92 3.51 0.42 0,46 0.57 0.37 10,00 5.01
2. OITHSO 0.82 2.83 4.14 0,23 0.36 0.22 6,00 3.87
3. COWC 0.85 6.54 5.53 0.51 0.41 13,87 7.96
4. CHCK 0.88 7.48 6.39 0.48 19,87 9.69
5. DRGRC 0.78 2.02 2.54 5,20 4.61
6. Total 0.92 21.78 16.33 54,93 16.72
Note: All correlations significant at P < 0.001. OTAHSO = Obsessional Thoughts about Harm to
Self/Othcrs; OITHSO = Obsessional Impulses to Harm Self/Others; COWC = Contamination
Obsessions and Washing Compulsions; CHKC = Checking Compulsions; DRGRC = Dressing/
Grooming Compulsions; OCD = Obsessive Compulsive Disorder.
172 G. Leonard Burns et al.

Barlow, 1988). In two earlier studies (Burns et al., 1995; Sternberger & Burns, 1991), individuals
were selected who scored in the top 3% on the original PI [2287 individuals were screened in the
Burns et al. (1995) study and 678 in the Sternberger & Burns (1991) study]. Approximately 6
months after the screening, 31 individuals who scored in the top 3 % and 31 individuals who scored
in the normal range on the PI were administered the diagnostic interview. In the high scoring group,
15 of the 31 individuals met the criteria for OCD with none of the 31 individuals in the comparison
group meeting criteria. The P I - - W S U R means in Table 7 for these 15 individuals were from the
completion of the PI at the time of the diagnostic interview (the 6-month earlier screening means
were higher).* The total score mean and the OTAHSO and C H K C subscale means for these 15
individuals were approximately 2 standard deviations above the respective normative means. In
addition, the scores on the COWC and D R G R C subscales were approximately 1.25 standard
deviations above the respective normative means.

DISCUSSION
Freeston et al. (1994) indicated that a problem with the PI, especially the 2 obsessional subscales,
was that the instrument was measuring worry in addition to obsessions. It was possible, however,
to revise the PI by the use of a recent theoretical distinction between obsessions and worry (Turner
et al., 1992; Wells & Morrison, 1994). As noted by Turner et al. (1992), obsessions and worry are
considered to differ in content with the "content of worry or rumination typically related to normal
experiences of everyday living (e.g. family, money, work), whereas the content of obsessions
frequently include themes of dirt/contamination, religion, sex, and aggression" (p. 266).
This content distinction between obsessions and worry allowed us to organize 39 of the original
60 PI items into 5 content categories relevant to obsessions and compulsions. This process also
resulted in the elimination of PI items that lacked content, and thus could be responded to on the
basis of either obsessions or worry (see footnote on p. 164). The 5 content categories included:
(1) obsessional thoughts about harm to oneself or others; (2) obsessional impulses to harm oneself
or others; (3) contamination obsessions and washing compulsions; (4) checking compulsions; and
(5) dressing/grooming compulsions. The items in each content category showed discriminant
validity from a measure of worry as well as from the other content categories of OCD. Principal
component analyses also provided support for the hypothesized 5-factor structure of the revised
PI. In addition, each of the 5 subscales showed high levels of internal consistency and stability over
a 6 to 7-month interval as well as discriminant validity from the other subscales across time. Finally,
individuals with a diagnosis of OCD had higher scores on the P I - - W S U R than a normative sample.
The results also showed that the revised PI, compared to the original version, was more
independent from worry. The current study found that the revised measure had 12% of its variance
in common with the PSWQ compared with 34% for the original PI (Freeston et al., 1994). Similar
results also occurred for the 2 content-based obsessional subscales in the revision. Here the
O T A H S O subscale shared 14% of its variance with the PSWQ and the OITHSO subscale shared
less than 1% of its variance with the PSWQ. The values for the 2 original obsessional subscales
of the PI were 45 and 12%, respectively (Freeston et al., 1984). These results thus indicate that
the theoretically-based revision of the PI provides a better self-report measure of obsessions and
compulsions. In addition, since the 5 subscales now measure content dimensions specifically
relevant to OCD, the subscales will hopefully have more clinical utility for assessment and
treatment outcome evaluation. Finally, the large normative sample of 5010 individuals will be
useful in this regard as well.
To further evaluate the usefulness of the content distinction between obsessions and compul-
sions, the current study should be replicated with a content-based measure of worry such as the
Worry Domains Questionnaire (Tallis, Eysenck & Mathews, 1992) or the Worry Questionnaire
(Keortge & Burns, 1994), because the PSWQ is not a content-based measure of worry (Davey,
1993). For example, the Worry Questionnaire was designed to measure 20 worry content domains
such as relations, finances, work and school performance, and world events, among others. These

*Thecompleteset of PI--WSUR means and standard deviationsfor the screeningand interviewsfor the two clinicalgroups
and the comparison group are available from the first author.
 Revision of the Padua Inventory 173

domains were selected to measure the content dimensions of worry as noted by Turner et al. (1992)
and DSM-IV. If the results of the current study were replicated with a content-based measure of
worry, then such would provide even more support for the usefulness of this distinction between
obsessions and worry. Confirmatory factor analysis would also be useful in determining the
adequacy of this conceptualization of obsessions and worry.
The validity of the content distinction between obsessions and worry is also important for the
evaluation of the other dimensions on which obsessions and worry are hypothesized to differ
(Keortge & Burns, 1994; Rachman, 1993; Rheaume, Ladouceur, Freeston & Letarte, 1995; Turner
et al., 1992; Wells & Morrison, 1994). For example, worry and obsessions are also considered to
differ in terms of distress, controllability, degree of realism, responsibility, types of cause, and time
consuming nature. In order to evaluate the validity of these other dimensions, however, it is
necessary to have measures of obsessions and worry which are theoretically-based and valid. For
example, if the measure of obsessions is also measuring worry, then attempts to evaluate how
obsessions and worry differ on the other hypothesized dimensions are confounded by the
measurement problem. Our hope is that this content-based revision of the PI and its independence
from worry, at least as measured by the PSWQ, will provide a means to seek better answers to
these issues.

REFERENCES
American Psychiatric Association (1994). Diagnostic and Statistical Manual of Mental Disorders (4th edn). Washington, DC:
Author.
Burns, G. L. (1995). Padua Inventory--Washington State University Revision. Pullman, WA: Author. (Available from G.
Leonard Burns.)
Burns, G. L., Formea, G. M., Keortge, S. & Sternberger, L. G. (1995). The utilization of nonpatient samples in the study
of obsessive compulsive disorder. Behaviour Research and Therapy, 33, 133-144.
Brown, T. A., Antony, M. M. & Barlow, D. H. (1992). Psychometric properties of the Penn State Worry Questionnaire
in a clinical anxiety disorders sample. Behaviour Research and Therapy, 30, 33-37.
Brown, T. A., Moras, K., Zinbarg, R. E. & Barlow, D. H. (1993). Diagnostic and symptom distinguishability of generalized
anxiety disorder and obsessive-compulsive disorder. Behavior Therapy, 24, 227-240.
Davey, G. C. L. (1993). A comparison of three worry questionnaires. Behaviour Research and Therapy, 31, 51-56.
DiNardo, P. A. & Barlow, D. (1988). Anxiety Disorders blterv&w Schedule--Revised (ADIS-R). Albany, NY: Authors,
Freeston, M. H., Ladouceur, R., Rheaume, J., Letarte, H., Gagnon, F. & Thibodeau, N. (1994). Self-report of obsessions
and worry. Behaviour Research and Therapy, 32, 29 36.
Keortge, S. & Burns, G. L. (1994). Worry and obsessions: Are they distinct cognitive phenomena? Paper presented at the
28th Annual Meeting of the Association for the Advancement of Behavior Therapy, San Diego, CA.
Ladouceur, R., Freeston, M. H., Dumont, J., Letarte, H., Rheaume, J., Gagnon, F. & Thibodeau, N. (1992). Penn State
Worry Questionnaire: Validity and reliability of a French translation. Canadian Psychology, 33, 236.
Meng, X. L,, Rosenthal, R. & Rubin, D. B. (1992). Comparing correlated correlation coefficients. Psychological Bulletin,
I11, 172-175.
Meyer, T. J., Miller, M. L., Metzger, R. L. & Borkovec, T. D. (1990). Development and validation of the Penn State Worry
Questionnaire. BehaL'iour Research and Therapy, 28, 487~,95.
Rachman, S. (1993). Obsessions, responsibility, and guilt. Behaviour Research and Therapy, 31, 149-154.
Rheaume, J,, Ladouceur, R., Freeston, M. H. & Letarte, H. (1995). Inflated responsibility in obsessive compulsive disorder:
Validation of an operational definition. Behaviour Research and Therapy, 33, 159-169.
Sanavio, E. (1988). Obsessions and compulsions: The Padua Inventory. Behaviour Research and Therapy, 26, 169-177.
Steketee, G. (1994). Behavioral assessment and treatment planning with obsessive compulsive disorder: A review
emphasizing clinical application. Behavior Therapy, 25, 613-633.
Sternberger, L. G. & Burns, G. L. (1990). Obsessions and compulsions: Psychometric properties of the Padua Inventory
with an American college population. Behaviour Research and Therapy, 28, 341-345.
Sternberger, L. G. & Burns, G. L. (1991). Obsessive Compulsive Disorder: Symptoms and diagnosis in a college sample.
Behavior Therapy, 22, 569-576.
Tallis, F. & de Silva, P. (1992). Worry and obsessional symptoms: A correlational analysis. Behaviour Research and Therapy,
30, 103-105.
Tallis, F., Eysenck, M. & Mathews, A. (1992). A questionnaire for the measurement of nonpathological worry. Behaviour
Research and Therapy, 13, 161-168.
Turner, S. M., Beidel, D. C. & Stanley, M. A. (1992). Are obsessional thoughts and worry different cognitive phenomena?
Clinical Psychology Rev&w, 12, 257-270.
Van Oppen, P. (1992). Obsessions and compulsions: Dimensional structure, reliability, convergent and divergent validity
of the Padua Inventory. Behaviour Research and Therapy, 30, 631-637.
Wells, A. & Morrison, A. P. (1994). Qualitative dimensions of normal worry and normal obsessions: A comparative study.
Behaviour Research and Therapy, 32, 867-870.