Interrater Reliability in Finger Joint Goniometer Measurement in Dupuytren’s Disease

Abstract

We investigated interrater reliability of range of motion (ROM) measurement in the finger joints of people with Dupuytren’s disease. Eight raters measured flexion and extension of the three finger joints in one affected finger of each of 13 people with different levels of severity of Dupuytren’s disease, giving 104 measures of joints and motions. Reliability measures, represented by intraclass correlation coefficient (ICC), standard error of the mean (SEM), and differences between raters with the highest and lowest mean scores, were calculated. ICCs ranged from .832 to .973 depending on joint and motion. The SEM was ≤3° for all joints and motions. Differences in mean between highest and lowest raters were larger for flexion than for extension; the largest difference was in the distal interphalangeal joint. The results indicate that following these standardized guidelines, the interrater reliability of goniometer measurements is high for digital ROM in people with Dupuytren’s disease.

Keywords

arthrometry,articular Dupuytren contracture finger joint range of motion,articular reproducibility of results

Dupuytren’s disease (DD) is a common diagnosis in hand therapy in northern Europe; the prevalence in the Swedish population is 6% (Bergenudd, Lindgärde, & Nilsson, 1993). The disease causes an extension deficit in one or several fingers. When the extension deficit affects hand function, the treatment of the disease is surgical and involves removal of the pathological tissue (Thurston, 2003). Goniometer measurement is commonly used in hand therapy to follow progress or as an outcome measure. Digital extension is a common outcome measure used to evaluate results after surgery and treatment (Draviaraj & Chakrabarti, 2004; Roush & Stern, 2000; Sinha, Cresswell, Mason, & Chakrabarti, 2002). To draw conclusions and interpretations on the basis of these measurements, it is important to demonstrate that the measurements are reliable (Greenfield, Kuhn, & Wojtys, 1998) and reflect actual changes in range of motion (ROM) rather than measurement errors (Groth, VanDeven, Phillips, & Ehretsman, 2001). Measurement error may occur when a single observer performs repeated measurements or when several observers perform measurements on the same person (Greenfield et al., 1998).

Several studies on the accuracy of goniometer measurement have been published, but only a few (Ellis & Bruton, 2002; Groth et al., 2001; Hamilton & Lachenbruch, 1969; Kato et al., 2007; Lewis, Fors, & Tharion, 2010) have evaluated the reliability of goniometer measurement in the finger joints (Norkin & White, 2003). Hamilton and Lachenbruch (1969) noted that measurement of the joints in the hand is complicated by several factors, including the large number of joints in a relatively small area, the short axis of motion, complex patterns of movement, and the difficulty in stabilizing or controlling the forces affecting the joints.

Various factors may affect the reliability of measurements. Reliability increases if raters use the same method and instrument in every assessment, and it is important to define guidelines for assessment, such as measurement position, anatomical landmarks, and goniometric placement (Cambridge-Keeling, 2002). Although either dorsal or lateral placement of the goniometer is acceptable when measuring ROM in the finger joints (Norkin & White, 2003), most raters prefer dorsal placement (Groth et al., 2001), which has been recommended because it is easier when measuring finger joints (Cambridge-Keeling, 2002) and is more suitable when using a goniometer with short lever arms (American Society of Hand Therapists, 1992; Kato et al., 2007). Previous studies have shown differences in measurement among raters depending on which finger joint was measured. According to Ellis, Bruton, and Goddard (1997), flexion in the distal interphalangeal (DIP) joint showed the greatest variation (9.8°), whereas Lewis et al. (2010) found that flexion in the metacarpophalangeal (MCP) joint showed the greatest difference in mean between raters with the highest and lowest scores (8.1°).

Findings in the literature have supported the commonly accepted level of measurement error of 5° for goniometric measurement of joints in the hand. Goniometric measurement is considered a reliable and valid tool (Norkin & White, 2003), although most studies have measured reliability on healthy individuals; only one of the studies (Groth et al., 2001) used a person with disability of the hand. Nevertheless, the reliability of a measure is closely linked to the group of people on which the measurements are performed and on the situation and the instrument itself (Streiner & Norman, 2008). To be useful, the reliability of a measure should be investigated for application in a specific context or sample (Gajdosik & Bohannon, 1987; Streiner & Norman, 2008). Therefore, the aim of this study was to assess the interrater reliability of goniometer measurement of the finger joints in people with DD.

Method

Research Design

Interrater reliability was investigated with a repeated-measures design in which each rater measured each patient once. Patients provided informed consent, and the Research Ethics Committee of the Faculty of Health Science at Linkoping University, Linkoping, Sweden, approved the protocol.

Participants

Eight occupational therapists from different hospitals in the southeast region of Sweden participated in the study as raters. All raters had experience in hand therapy and goniometer measurement (Table 1). All patients with DD who had undergone surgery during the past year, were currently not in treatment, and were living within 70 km of the hospital were selected for the study. Of the 19 eligible patients, 6 declined participation, and 13 patients (11 men and 2 women) with different levels of severity of DD agreed to participate in the study. All patients were retired, and their mean age was 73 yr (standard deviation = 7.2). One patient had diabetes, 9 had hereditary DD, and 7 had unilateral DD. Affected fingers to be measured were digit V for 10 patients and digits II, III, and IV (one each) for the other 3 patients.

Table 1.

Characteristics of Raters (N = 8)

Characteristic	No. of Raters
Experience working as an occupational therapist, yr
0–5	0
6–10	3
11–15	2
>15	3
Experience working with hand injuries, yr
0–5	1
6–10	4
11–15	0
>15	3
Frequency of performing assessment of range of motion in daily practice
Daily	7
2–3 times a week	1
Level of care provided
Specialist hospital care	2
County hospital care	6
Primary health care	1 a

One rater worked at both the county and primary health care levels.

Procedure

We sent guidelines for measuring ROM to the raters 1 mo before the day of assessment. Raters read the guidelines individually and then reviewed them together during a meeting in which they had the opportunity to practice and discuss them. The guidelines included position of measurement, goniometer placement, and instructions to be given to the patient.

On the assessment day, the 13 patients with DD arrived in groups of four or five at three different times during the day. Each patient was placed at a separate table. The raters circulated among patients in a random order, with caution taken so that each rater could be the first rater only once. Each rater performed the measurement, wrote down the result for each patient on a piece of paper, put the paper upside down in a box, and continued to the next patient.

Guidelines for Measurement

All eight raters performed one trial of measurement for all 13 patients. Measurements of flexion and extension of the MCP, proximal interphalangeal (PIP), and DIP joints in the affected finger of all patients were performed, giving a total of 104 observations of joints and motions. Measurements were performed over the dorsal midline of the metacarpals or phalanges with a plastic goniometer (Figure 1). The scale of the goniometer used is graded in intervals of 2°. The position for measurement of flexion in the MCP and PIP joints was as follows: elbow placed on table and forearm in neutral position, wrist in 30° extension, and fingers in full flexion making a fist. For measurement of flexion in the DIP joint, raters instructed patients to extend the MCP joint as much as possible and keep the PIP and DIP joints in flexion, attempting to make a hook fist. The position for measurement of finger extension was as follows: elbow placed on table and forearm in neutral position, wrist in neutral position (0°), and fingers in full extension. Raters instructed patients to “make a fist,” “straighten your knuckles and keep your finger joints in flexion,” and “straighten the fingers all the way.”

Figure 1.

Universal finger goniometer used in the study, graded at intervals of 2°.

Statistical Analysis

We analyzed data using PASW Statistics 18 (SPSS Inc., Chicago) on individual joints and on total active extension (TAE) and total active flexion (TAF). We calculated TAE and TAF by adding the scores for extension and flexion in the MCP, PIP, and DIP joints. Descriptive statistics for each patient’s ROM as measured by the eight raters were calculated to illustrate the patient’s level of severity of DD. Descriptive statistics for each rater, joint, and motion were also calculated. The differences between the rater with the highest and the lowest mean scores were then calculated for each joint and motion. As an indication of the precision in measurements, the standard error of the mean (SEM) was calculated for each joint and motion. The SEM was chosen because it quantifies the precision of individual scores and indicates how much difference in a measurement is needed for that measurement to be considered a true change (Weir, 2005). The SEM differs from the standard deviation (SD), which indicates the variability in the population the sample comes from (Altman & Bland, 2005). Calculation of SEM was performed as described by Altman and Bland (2005).

A two-way repeated-measures analysis of variance (ANOVA) for each ROM measure was performed to determine whether there were differences in means between raters. ANOVA is a complement to the intraclass correlation coefficient (ICC), which measures relative interrater reliability. ANOVA is an additional way to show whether there are any systematic errors (Streiner & Norman, 2008). Relative interrater reliability was assessed with the ICC two-way mixed model and an absolute-agreement definition (McGraw & Wong, 1996). This model of ICC treats raters as a fixed effect and patients as random effects. We chose the absolute-agreement definition because systematic variability among raters was relevant to the analysis. The strength of ICC was interpreted using Munro’s (2005) classification, in which 0–0.25 = little if any, 0.26–0.49 = low, 0.50–0.69 = moderate, 0.70–0.89 = high, and 0.9–1.0 = very high correlation.

Results

ROM varied in the patients with DD according to the severity of the disease (Table 2). The widest range of joint motion was seen in digital extension, and the largest range was for the PIP joint; the difference between the patients with the best and worst digital extension (Patients 3 and 8) was 98.8°. For the DIP and the MCP joints, the differences were 67.9° and 59.7°, respectively. Seven patients had a total active extension of <30°, and 2 had a total active extension >80°.

Table 2.

Mean Range of Motion for Each Individual Digit, Joint, and Motion Measured by the Eight Raters

	MCP ext		PIP ext		DIP ext		MCP flex		PIP flex		DIP flex
Digit	M	SD	M	SD	M	SD	M	SD	M	SD	M	SD
II	30.6	6.3	28.8	3.3	27.6	3.6	86.5	2.0	93.1	2.6	45.0	4.7
III	27.6	3.0	62.5	2.0	−0.5	1.4	87.1	4.0	102.0	1.6	59.3	3.8
IV	37.8	2.7	2.6	2.1	−1.5	3.5	91.5	3.7	63.5	10.4	27.4	7.1
V	9.5	2.9	26.3	2.7	19.1	2.7	89.3	3.5	84.5	2.6	75.9	4.1
V	37.3	9.2	−36.3	4.9	52.8	7.3	101.5	3.9	58.0	3.3	73.6	4.9
V	0.4	2.5	19.9	3.4	−13.3	4.1	68.8	4.4	80.0	2.2	35.8	2.4
V	−16.9	3.4	34.8	3.4	1.9	2.7	77.9	3.0	93.3	3.5	65.6	6.8
V	1.8	4.2	18.6	4.6	−0.1	5.0	85.6	5.2	91.5	2.4	69.9	3.6
V	2.2	3.6	10.3	5.2	−15.1	6.7	88.4	2.7	94.0	3.0	67.5	6.0
V	18.8	4.2	0.9	2.1	0.4	1.1	94.3	2.6	77.0	2.5	51.1	2.7
V	−21.9	4.7	42.3	5.3	30.8	3.3	90.3	2.3	72.5	1.9	54.3	2.7
V	−6.9	2.9	29.5	5.2	−2.6	3.2	92.6	2.5	83.9	2.3	53.1	5.0
V	−14.4	5.8	19.0	3.9	−0.5	0.9	98.1	6.1	90.1	1.2	71.0	4.7

Note. Minus signs indicate hyperextension in a joint. DIP = distal interphalangeal joint; ext = extension; flex = flexion; M = mean; MCP = metacarpophalangeal joint; PIP = proximal interphalangeal joint; SD = standard deviation; TAE = total active extension; TAF = total active flexion.

Descriptive statistics for all raters and each joint and motion are shown in Table 3. A difference in mean existed among raters for all measurements except MCP extension, PIP flexion, and DIP extension (p < .05). Differences in mean between the highest and lowest raters were larger for measurement of flexion (4°–18°) than of extension (3°–8°). For isolated joints, DIP flexion had the largest difference (11°). The SEM was <3° for all joints and motions.

Table 3.

Measures of All Raters Across All Joints and Motions

Joint and Motion	M	SD	CI (95%)	Difference Between Raters a	SEM	ANOVA (p)	ICC	CI (95%)
MCP ext	8°	20	4°–12°	3°	2°	.745	.952	.904–.982
PIP ext	20°	23	15°–24°	7°	2°	.000*	.973	.944–.991
DIP ext	8°	19	4°–11°	3°	2°	.409	.960	.919–.985
MCP flex	89°	9	87°–90°	4°	1°	.023*	.832	.696–.934
PIP flex	83°	13	81°–86°	4°	1°	.063	.920	.844–.970
DIP flex	58°	15	55°–61°	11°	1°	.000*	.909	.809–.967
TAE	36°	30	30°–41°	8°	3°	.031*	.949	.898–.981
TAF	230°	24	225°–234°	18°	2°	.000*	.898	.873–.976

Note. ANOVA = two-way repeated measures analysis of variance; CI = confidence interval; DIP = distal interphalangeal joint; ext = extension; flex = flexion; ICC = intraclass correlation, two-way mixed effects model, absolute agreement definition, single measures; M = mean; MCP = metacarpophalangeal joint; PIP = proximal interphalangeal joint; SD = standard deviation; SEM = standard error of the mean; TAE = total active extension; TAF = total active flexion.

Difference in scores between the raters with the highest and lowest mean scores.

p < .05.

The ICC values across all raters for each joint and motion are shown in Table 3. The ICC values were higher for measurement of digital extension than of flexion, with the PIP joint showing the highest value (.973). The lowest value was seen in MCP flexion (.832).

Discussion

The results show that interrater reliability in goniometric measurement of the finger joints in people with DD is high or very high, as defined by Munro’s (2005) classification scheme. When looking at isolated joint measures in this study, the differences between the raters with the highest and lowest mean scores averaged 11°, depending on joint and motion. For all isolated joints and motions, except for DIP flexion, the differences averaged ≤7°. These differences in mean between highest and lowest raters were higher than the 5° reported by Norkin and White (2003), which is the commonly accepted measurement error for goniometric measurement of joints in the hand. However, the SEM in this study, representing the measurement error, was ≤3° independent of joint or motion and thus lower than Norkin and White’s 5° margin of error. When adding the TAE or TAF, the differences between raters with the highest and lowest mean score increased to 8° and 18°, respectively. According to Hopkins (2000), it is common for participants with large value variables or subgroups of participants to have larger differences in typical errors. Larger differences in mean between raters could be expected because the patients in this study had different levels of severity of DD.

Patient characteristics in this study were consistent with those of the DD population in regard to unilateral versus bilateral disease (Bergenudd et al., 1993), presence of diabetes (Hindocha, Stanley, Watson, & Bayat, 2006), age (Gudmundsson, Arngrímsson, Sigfússon, Björnsson, & Jónsson, 2000), and affected digit (Mikkelsen, 1976). Heredity for DD was positive for 70% of our sample, higher than previously reported data on heredity of 38%–44% among the DD population (Hindocha, John, Stanley, Watson, & Bayat, 2006; Hindocha, Stanley, et al., 2006). Our sample was heterogeneous in digital ROM and had both hyperextension and extension deficits in different finger joints. This heterogeneity should be seen as strength of this study because the interrater reliability of the measurements performed captured a wide spectrum of ROM deficits typical in DD. However, the variation may lead to a greater likelihood that the typical error was not the same for all patients.

The findings in this study, including the larger differences between raters when measuring digital flexion of the DIP joint (11°), are in agreement with Ellis et al.’s (1997) earlier results. The larger differences when measuring digital flexion of the DIP joint may reflect difficulties associated with the measurement position and placement of the goniometer. The goniometer used in this study was a universal type with short levers developed for measurement of finger joints. Nevertheless, measurement of flexion in the DIP joint was difficult to perform in the full-fist position. Raters therefore performed this measurement in an attempted hook-fist position: Patients kept the PIP and DIP joint in flexion while extending the MCP joint to the furthest extent the extension deficits allowed. Measurement of the DIP joint in either of these positions (hook fist or full fist) should not matter theoretically because differences in DIP flexion are more likely to occur when the PIP joint is in extension (Norkin & White, 2003). The use of a goniometer with one very short lever arm would favor measurement of the DIP joint in the full-fist position but could lead to difficulties when measuring the MCP or PIP joints.

We attempted to standardize the measurement procedure by giving the raters the opportunity to coordinate instructions and practice measurement guidelines on each other. Differences in the amount of pressure raters used to apply the goniometer against the phalanges became obvious. This possible source of error has been mentioned previously (Ellis et al., 1997; Lewis et al., 2010), and the practice session led to some raters becoming aware of the need to apply the goniometer more firmly against the phalanges. Factors related to the patients, such as the effects of repeating the same motion several times, could also affect the result. Effects such as learning, motivation, or fatigue can lead to a systematic change in the mean, and such changes need to be eliminated from estimates of within-individual variation (Hopkins, 2000). To avoid systematic errors in the current study, the raters performing the measurements circulated between the patients in a random order, with caution taken so that each rater was the first rater only once.

Several studies have shown that differences in intrarater reliability are lower than differences in interrater reliability (Ellis & Bruton, 2002; Ellis et al., 1997; Lewis et al., 2010); consequently, we did not rate intrarater reliability in this study. This strategy may constitute an advantage of the current study; the results reflect the clinical situation in a realistic way because the raters conducted only one measure per joint when the patients were making an active motion. The raters were experienced occupational therapists, and most practiced hand therapy and ROM measurement on a daily basis. We performed no analysis of reliability in relation to experience because this concern was beyond the scope of our study. Previous research has shown that more experienced therapists tend to be more reliable (Bovens, van Baak, Vrencken, Wijnen, & Verstappen, 1990).

A key reason for undertaking this study was to enable the development and use of guidelines for measurement. To achieve more reliable measurements in clinical settings, practice and discussion of measurement guidelines should be used as a natural part of the introduction of new colleagues and as a way to create consensus among more experienced hand therapists.

Implications for Practice

This study included only people with DD, and the results may not be applicable to people with other diseases or injuries of the hand. Future research should investigate the interrater reliability in goniometer measurement of the finger joints in other diagnoses in hand therapy following the guidelines used in the current study.

The guidelines developed in the present study can be recommended in a clinical setting as a result of the high or very high interrater reliability.

Standardized guidelines should cover measurement position, goniometer placement, type of goniometer, and instructions to the patient.

Discussion and practice of measurement guidelines should be a natural part of the clinical setting.

To reach consensus about goniometer measurement in a clinical setting, both new and more experienced colleagues should participate in the discussion and practice of the guidelines.

Conclusion

In this study, interrater reliability was high or very high for goniometer measurement of ROM in finger joints of people with DD; our raters were experienced and performed this measurement following standardized procedures and guidelines. The difference in mean between raters with the highest and lowest scores was higher than in studies performed on healthy people. The standard error of measurement, however, was less than the commonly accepted measurement error. Clinicians require this information to draw accurate conclusions about the true change in ROM for people with DD and to avoid the risk of measurement error.

Footnotes

Acknowledgments

We thank the Department of Hand Surgery, Plastic Surgery and Burns, Linkoping University Hospital, for providing practical support and Henrik Magnusson, Linkoping University, for statistical support. The study was supported by grants from the Medical Research Council of Southeast Sweden.

References

Altman

D. G.

, & Bland

J. M.

(2005). Education and debate—Statistics notes: Standard deviations and standard errors. British Medical Journal, 331, 903. doi: doi: 10.1136/bmj.331.7521.903

American Society of Hand Therapists. (1992). Clinical assessment recommendations (2nd ed.). Chicago: Author.

Bergenudd

Lindgärde

, & Nilsson

B. E.

(1993). Prevalence of Dupuytren’s contracture and its correlation with degenerative changes of the hands and feet and with criteria of general health. Journal of Hand Surgery, 18, 254–257.

Bovens

A. M.

van Baak

M. A.

Vrencken

J. G.

Wijnen

J. A.

, & Verstappen

F. T.

(1990). Variability and reliability of joint measurements. American Journal of Sports Medicine, 18, 58–63. doi: doi: 10.1177/036354659001800110

Cambridge-Keeling

(2002). Range of motion measurement of the hand. In Hunter

J. M.

Mackin

E. J.

, & Callahan

A. D.

(Eds.), Rehabilitation of the hand (pp. 169–182). St. Louis, MO: Mosby

Draviaraj

K. P.

, & Chakrabarti

(2004). Functional outcome after surgery for Dupuytren’s contracture: A prospective study. Journal of Hand Surgery, 29, 804–808.

Ellis

, & Bruton

(2002). A study to compare the reliability of composite finger flexion with goniometry for measurement of range of motion in the hand. Clinical Rehabilitation, 16, 562–570. doi: doi: 10.1191/0269215502cr513oa

Ellis

Bruton

, & Goddard

J. R.

(1997). Joint angle measurement: A comparative study of the reliability of goniometry and wire tracing for the hand. Clinical Rehabilitation, 11, 314–320. doi: doi: 10.1177/026921559701100408

Gajdosik

R. L.

, & Bohannon

R. W.

(1987). Clinical measurement of range of motion: Review of goniometry emphasizing reliability and validity. Physical Therapy, 67, 1867–1872.

10.

Greenfield

M. L.

Kuhn

J. E.

, & Wojtys

E. M.

(1998). A statistics primer: Validity and reliability. American Journal of Sports Medicine, 26, 483–485.

11.

Groth

G. N.

VanDeven

K. M.

Phillips

E. C.

, & Ehretsman

R. L.

(2001). Goniometry of the proximal and distal interphalangeal joints: Part II. Placement preferences, interrater reliability, and concurrent validity. Journal of Hand Therapy, 14, 23–29.

12.

Gudmundsson

K. G.

Arngrímsson

Sigfússon

Björnsson

, & Jónsson

(2000). Epidemiology of Dupuytren’s disease: Clinical, serological, and social assessment: The Reykjavik Study. Journal of Clinical Epidemiology, 53, 291–296. doi: doi: 10.1016/S0895-4356(99)00145-6

13.

Hamilton

G. F.

, & Lachenbruch

P. A.

(1969). Reliability of goniometers in assessing finger joint angle. Physical Therapy, 49, 465–469.

14.

Hindocha

John

Stanley

J. K.

Watson

S. J.

, & Bayat

(2006). The heritability of Dupuytren’s disease: Familial aggregation and its clinical significance. Journal of Hand Surgery, 31, 204–210. doi: doi: 10.1016/j.jhsa.2005.09.018

15.

Hindocha

Stanley

J. K.

Watson

, & Bayat

(2006). Dupuytren’s diathesis revised: Evaluation of prognostic indicators for risk of disease recurrence. Journal of Hand Surgery, 31, 1626–1634.

16.

Hopkins

W. G.

(2000). Measures of reliability in sports medicine and science. Sports Medicine, 30, 1–15. doi: doi: 10.2165/00007256-200030010-00001

17.

Kato

Echigo

Ohta

Ishiai

Aoki

Tsubota

, & Uchiyama

(2007). The accuracy of goniometric measurements of proximal interphalangeal joints in fresh cadavers: Comparison between methods of measurement, types of goniometers, and fingers. Journal of Hand Therapy, 20, 12–18, quiz 19. doi: doi: 10.1197/j.jht.2006.11.015

18.

Lewis

Fors

, & Tharion

W. J.

(2010). Interrater and intrarater reliability of finger goniometric measurements. American Journal of Occupational Therapy, 64, 555–561. doi: doi: 10.5014/ajot.2010.09028

19.

McGraw

, & Wong

S. P.

(1996). Forming inferences about some intraclass correlation coefficients. Psychological Methods, 1, 30–46. doi: doi: 10.1037/1082-989X.1.1.30

20.

Mikkelsen

O. A.

(1976). Dupuytren’s disease—A study of the pattern of distribution and stage of contracture in the hand. Hand, 8, 265–271.

21.

Munro

B. H.

(2005). Statistical methods for health care research (5th ed.). Philadelphia: Lippincott Williams & Wilkins.

22.

Norkin

, & White

(2003). Measurement of joint motion: A guide to goniometry (3rd ed.). Philadelphia: F. A. Davis.

23.

Roush

T. F.

, & Stern

P. J.

(2000). Results following surgery for recurrent Dupuytren’s disease. Journal of Hand Surgery, 25, 291–296.

24.

Sinha

Cresswell

Mason

, & Chakrabarti

(2002). Functional benefit of Dupuytren’s surgery. Journal of Hand Surgery, 27, 378–381.

25.

Streiner

, & Norman

(2008). Health measurement scales: A practical guide to their development and use (4th ed.). Oxford, England: Oxford University Press.

26.

Thurston

A. J.

(2003). Dupuytren’s disease. Journal of Bone and Joint Surgery, 85, 469–477. doi: doi: 10.1302/0301-620X.85B4.14215

27.

Weir

J. P.

(2005). Quantifying test–retest reliability using the intraclass correlation coefficient and the SEM. Journal of Strength and Conditioning Research, 19, 231–240.