A validation of the revised circadian type inventory in a working sample

https://doi.org/10.1016/j.paid.2005.04.012Get rights and content

Abstract

The measurement properties and construct validity of the two-factor 11-item Circadian Type Inventory (CTI) was examined in a working sample comprising shiftworkers and day workers. The two factors explained 50.04% of the variance (LV = 20.24% and FR = 29.80%). Inter-item correlations were all significant and the item-total correlations suggested the weakness of one item in the LV scale. Cronbach’s alpha was 0.72 (LV) and 0.79 (FR). Confirmatory factor analysis however provided marginal support for the posited structure. Vigorous types were significantly more alert than languid types across the day and suggested they are less influenced by sleep inertia on waking. Vigorous types also reported needing significantly less sleep. Significant differences in alertness between flexible and rigid types were found between 16:00 and 22:00 and suggested flexible types are less influenced by circadian and wakefulness factors during these periods. Flexible types also reported significantly less sleep need compared to rigid types. The findings suggested vigorous and flexible types would best cope with shiftwork. It is concluded that the CTI demonstrated good reliability and validity in a working sample. Recommendations for future studies are made.

Introduction

Human circadian rhythms have long been known to demonstrate individual differences in terms of the rhythm’s phase, amplitude and stability. These endogenous factors are considered to play a key role in promoting better adjustment to night work (Monk & Folkard, 1985). The disruption of circadian rhythms due to night work is most problematic since these rhythms are governed by an internal oscillator that is resistant to change (Klein et al., 1991, Wever, 1980). The recognition of individual differences in human biology and their role in adjustment to night work has resulted in a number of self report measures that seek to indirectly assess circadian rhythm characteristics. Such measures could assist in the selection of potential shiftworkers and in the counselling of existing shiftworkers (Di Milia et al., 2004, Natale and Cicogna, 2002).

The majority of research activity in the development and refinement of self report measures has concentrated on assessing differences in rhythm phase. These measures have linked a preference for activity earlier or later in the day (Kerkhof, 1985) with a phase difference of circadian timing. The morningness–eveningness questionnaire (MEQ, Horne & Ostberg, 1976) is one of the most widely used phase measures. A review of the MEQ and its relationship to a number of dependent variables can be found in Tankova, Adan, and Buela-Casal (1994). The MEQ has also been translated into a number of languages (Adan and Almirall, 1990, Mecacci and Zani, 1982) and has served as a base for the development of the Composite Scale of Morningness (CSM; Smith, Reilly, & Midkiff, 1989). The CSM has consistently been reported to have strong psychometric properties in samples drawn from a number of countries (Bohle et al., 2001, Caci et al., 2000). More recently, the CSM has received some criticism. For example, its reference to the timing of various activities is considered to account for differences in scores between American (Smith et al., 1989) and Australian students (Greenwood, 1994). To address this limitation, the Preferences scale (Smith et al., 2002) was developed and initial studies have reported strong psychometric properties across several countries (Bohle et al., 2001, Smith et al., 2002).

In comparison to phase measures, self report measures assessing a rhythm’s amplitude and stability has been limited to the circadian type questionnaire (CTQ; Folkard, Monk, & Lobban, 1979). In addition to measuring phase, the CTQ was developed with the ‘hypothesis that better adjustment might be shown by people with; (a) low amplitude rhythms and (b) ‘flexible’ or non-stable rhythms (p. 80)’. Rhythm stability is assessed via the dimension labelled flexibility/rigidity (FR) of sleeping habits. Rigid types were reported to be less able to; go to bed early, sleep in or sleep at unusual hours. They also reported a preference to sleep and eat at regular times. The second dimension defined as languid/vigorous (LV), represents the rhythm’s amplitude. Languid types were reported to find it difficult to overcome drowsiness and feel lethargic following reduced sleep. The authors reported that vigorous and flexible types showed ‘better’ circadian adjustment on a number of psychological and physiological.

The original study was largely exploratory and based on a small sample (n = 48). These factors may explain why subsequent evaluations reported the CTQ to have weak psychometric properties (Cocklin and Kirkland, 1982, Smith et al., 1989). In particular, the weakness of the phase measure (Greenwood, 1995) resulted in a revised measure known as the circadian type inventory (CTI; Folkard, 1987) which consisted of two 15-item scales. While the measurement properties of the CTI also received limited support (Smith, Brown, Di Milia, & Wragg, 1993), two longitudinal studies have suggested the relevance of the FR and LV constructs. Kaliterna, Vidacek, Prizmic, and Radosevic-Viadacek (1995) administered a battery of measures to a young workforce about to enter rotating shiftwork. At the one year follow up, small but significant correlations were obtained between rigidity of sleeping habits and impaired health and psychosomatic-digestive complaints. After three years of shift work experience, the CTQ was the best predictor of impaired health, respiratory and psychosomatic-digestive complaints. Test–retest reliability for FR/LV on the CTQ was 0.54 and 0.36 respectively, and FR/LV on the CTI was 0.35 and 0.51 respectively after nine years of shiftwork experience (Kaliterna, Vidacek, Radosevic-Viadacek, & Prizmic, 1993).

On-going refinements to the CTI resulted in an 18-item format which was included in the Standard Shiftwork Index (SSI; Barton et al., 1995). The SSI is a standard battery of measures used to study the relationship between a range of independent and dependent variables. Using a large nursing and industrial sample, Barton et al. reported the factor loadings for both scales to range from 0.33 to 0.80, the two factors explained approximately 26% of the variance in both samples and Cronbach’s alpha ranged from 0.73 to 0.79. Smith, Gibby, Zickar, Crossley, and Robie (2001) reported similar reliability of 0.78 for both scales but model fit indices using structural equation modelling suggested the structure was not optimal.

Recently, Di Milia et al. (2004) reported improvements in the psychometric properties of the CTI. Using principal components and reliability analyses, the authors produced a two-factor 11-item model in a student sample (n = 213). This model was supported by a confirmatory factor analysis (CFA) using structural equation modelling in a separate student sample (n = 367). The 11-item model was a marked improvement over the 18-item model on several incremental fit indices (>0.90) and was a more parsimonious model fit. Test–retest reliability (n = 178) after three months was 0.72 and 0.75 for the LV and FR scales respectively. In addition, they concluded the 11-item model applied for males and females. A student sample with no experience of night work is a necessary precursor in the development of self report measures seeking to predict adjustment to night work since work schedules can act as a powerful zeitgeber (Tankova et al., 1994).

The present study is the next step in the development of the 11-item CTI and has two main aims. The first is to replicate the factor and item structure of the CTI (Di Milia et al., 2004) and assess its psychometric properties in a working population. The second aim is to validate the CTI scales against self reported sleep need and self reported alertness ratings during waking hours. Based on Folkard et al.’s (1979) original work it is expected that flexible and vigorous types will show greater alertness and report less sleep need.

Section snippets

Participants

The survey was completed by two groups. In the rotating shiftworker (SW) group, surveys were completed by 116 male and four females working at a coal mine (response rate = 46%). Most of the workers were on a 12 h rapidly rotating shift schedule. Their mean age was 38 years (SD = 10.31) with a range from 19 to 65 years. The mean length of service was 12.4 years (SD = 9.60) with a range from one to 40 years. The SW reported a mean sleep of 6.98 h (SD = 1.09) per day.

The second group comprised day workers

Descriptive statistics, inter-item correlations and internal reliability

Descriptive statistics and internal reliability for the LV and FR scale are presented in Table 1 (N = 249). Mean scale scores are approximately at the midpoint for each scale. The standard deviations, skewness and kurtosis values suggested an approximately normal distribution of scores.

Item means, their standard deviations and inter-item correlations can be found in Table 2. Inter-item correlations for the LV scale ranged from 0.15 to 0.50 and all were significant at p < 0.01 except for one pair (p <

Discussion

Using a student sample, Di Milia et al. (2004) reported the 11-item CTI to be psychometrically superior to the former 18-item version. While student samples are necessary precursors in developing measures of individual differences, the instrument must subsequently demonstrate good measurement properties and construct validity in a working sample. The results provide some support toward both these aims.

An examination of the psychometric properties of the CTI suggested that Cronbach’s alpha was

Acknowledgements

The first author thanks Dr. Janet Barnes-Farrell for hosting his sabbatical and Dr. Robert Ho for his advice with SEM. The authors also appreciate the constructive comments of Emeritus Professor Don Tepas and the anonymous reviewers of earlier drafts.

References (32)

  • P. Bohle et al.

    Psychometric evaluation of the early/late preferences scale

    Ergonomics

    (2001)
  • B. Byrne

    Structural equation modelling with AMOS

    (2001)
  • B. Cocklin et al.

    Circadian-referenced predictive tests of adjustment: Some further explorations

    Perceptual and Motor Skills

    (1982)
  • Folkard, S. (1987). Circadian Type Inventory. Sheffield, United Kingdom: University of Sheffield, Department of...
  • S. Folkard et al.

    Beyond the three-process model of alertness: Estimating phase, time on shift and successive night effects

    Journal of Biological Rhythms

    (1999)
  • S. Folkard et al.

    Towards a predictive test of adjustment to shift work

    Ergonomics

    (1979)
  • Cited by (0)

    1

    Permanent address: School of Management and CRC for Integrated Engineering Asset Management, Central Queensland University, Rockhampton, Qld 4702, Australia.

    View full text