R
E S E A R C HA
R T I C L EPsychometric Characteristics of the Parsimonious Chinese Version of the
Smoking Self-Efficacy Survey (CSSES-20) ∗
HUEY-SHYSCHEN, PhD, MSN, RN, CHESa
JIUNN-JYESHEU, PhD, MSPH, CHESb CHING-SUNGHO, PhDc
ABSTRACT
BACKGROUND: Cigarette smoking is a health-risk behavior of global proportions.
Self-efficacy plays an important role in both smoking acquisition and smoking resistance. Reliability and validity of an instrument is fundamental to research results, particularly in its simplified form on a different population. The purpose of this study was to conduct psychometric testing on the parsimonious Chinese version of the Smoking Self-efficacy Survey (CSSES-20).
METHODS: The randomized cluster sample was drawn from 61 middle schools in Taipei City, Taiwan. Following a pilot test of the CSSES-20, the CSSES was administered to 571 adolescents. Construct validity was tested by the exploratory factor procedures and the contrasted group approach. Pearson correlation coefficients were used to examine the criterion validity and test-retest reliability for the stability of the scale. Cronbach’s alpha coefficients were used to determine the internal consistency of the scale.
RESULTS: The exploratory factor analysis yields three components, ‘‘opportunity to smoke,’’ ‘‘emotional stress,’’ and ‘‘influence of friends,’’ accounting for 80.4% of the total variances. The criterion validity was also supported by the study results. The contrasted group approaches affirmed the construct validity of the CSSES-20. Stability of scales was supported by test-retest reliability. Cronbach’s alphas for 3 subscales ranged from .90 to .93.
CONCLUSIONS: A concise instrument can alleviate response burden for adolescent study participants and increase their recruitment and retention rates. The CSSES-20 demonstrated satisfactory construct validity, criterion validity, stability, and internal consistency reliability. These findings can be used to provide school teachers and nurses with information about the relationship between self-efficacy to resist smoking and adolescent smoking behavior.
Keywords: reliability; validity; smoking and tobacco; child and adolescent health.
Citation: Chen H-S, Sheu J-J, Ho C-S. Psychometric characteristics of the parsimonious Chinese version of the Smoking Self-efficacy Survey (CSSES-20). J Sch Health. 2010; 80:
259-264.
Received April 9, 2009 Accepted November 19, 2009
aAssociate Professor, (chenhu@umdnj.edu), School of Nursing, University of Medicine and Dentistry of New Jersey, 65 Bergen Street, PO Box 1709, Newark, NJ 07101-1709.
bAssistant Professor, (jjsheu@hhp.ufl.edu), Department of Health Education and Behavior, University of Florida, FLG 5, Stadium Road, PO Box 118210, Gainesville, FL 32611-8210.
cAssociate Professor and Chair, (csho@asia.edu.tw), Department of Healthcare Administration, Asia University, 500 Lioufeng Rd., Wu-Feng, Tai-Chung County 41354, Taiwan.
Address correspondence to: Ching-Sung Ho, Associate Professor and Chair, (csho@asia.edu.tw), Department of Healthcare Administration, Asia University, 500 Lioufeng Rd, Wu-Feng, Tai-Chung County 41354, Taiwan.
*Indicates CHES and Nursing continuing education hours are available. Also available at: http://www.ashaweb.org/continuing education.html
C
igarette smoking, a health-risk behavior of global proportions, is highly prevalent among Taiwanese adolescents. A previous national health survey of Taiwan revealed that smoking prevalence rates increased proportionally with age from 12 to 40 years, among both males and females.1A nationwide survey of high school students (grades 10 to 12) indicated that nearly 25% of boys currently smoked.2 Furthermore, several studies conducted in Taiwan demonstrated that the majority of students began smoking in grades 5 and 6 and became regular smokers by grades 7 or 8.3,4 Early onset of cigarette smoking leads to a longer period of exposure to cigarette toxins, which in turn may contribute to an increased risk for a range of severe health consequences.5 Self-efficacy plays an important role in addictive behaviors, including smoking acquisition and smoking resistance. Research has shown that self-efficacy to resist smoking is related to efforts by Taiwanese adolescents to resist smoking.6,7 Reliability and validity of an instrument is funda- mental to research results.8 Although an instrument may have established sound psychometric properties, it is necessary to reevaluate the reliability and valid- ity with different populations.9 According to previous research experiences, the simplicity of instruments is a determining factor in their usefulness for assessing smoking attitudes among youth. The purpose of this study was to conduct psychometric testing on the parsi- monious Chinese version of the Smoking Self-efficacy Survey (CSSES-20) with Taiwanese adolescents.The Smoking Self-Efficacy Survey
Self-efficacy is a central concept of Bandura’s Social Cognitive Theory,10,11 which attempts to predict and explain behavior change. Perceived self-efficacy is one’s perceived capabilities to execute or restrain from a particular behavior.10,11 The Smoking Self- efficacy Survey (SSES) developed by Lawrance and Rubinson12 is based on Bandura’s self-efficacy theory and measures an adolescent’s perceived ability to resist smoking in each setting. The SSES consists of 3 subscales, the influence of friends, emotional stress, and opportunity to smoke, each containing 12 items, for a total of 36 items. Each item uses a 6-point Likert- type scale, ranging from 1 for ‘‘I am very sure I would smoke’’ to 6 for ‘‘I am very sure I would not smoke.’’
The 12 items in each subscale were summed for the subtotal score with a possible score ranging from 12 to 72. Higher scores on each subscale indicated increased perceived ability to resist smoking.
The psychometric characteristics of the SSES were established with American young adolescents.12,13 Reliability was supported by the test-retest reliability12 and internal consistency.12,13 Validity was supported by expert validity,12 item-to-total correlation coeffi- cients; concurrent validity, and construct validity.13
The Chinese Version of the Smoking Self-Efficacy Survey Chen et al6 used the SSES in a study of smoking behaviors of 11- to 14-year-old Taiwanese children to establish the equivalence of meaning and cultural congruence between the original instrument and its Chinese version. The SSES was translated into Chinese then back-translated into English. Three items were removed from the CSSES after they were deemed to be inappropriate for Taiwanese adolescents based on inputs from focus group interviews with children, review by a panel of experts, and laypersons.6 Therefore, the CSSES had 33 items with 11 items for each subscale, the influence of friends, emotional stress, and opportunity to smoke, respectively. Cronbach’s alpha coefficients for the 3 subscales, the influence of friends, emotional stress, and opportunity to smoke were .93, .90, and .96, respectively.6Most recently, Chen et al7administered the CSSES to Taiwanese children aged 11 to 14 years to examine the relationships between stages of change in smoking acquisition and levels of self-efficacy; the reliability results supported the internal consistency of the CSSES.
Results from item reduction procedures (principal component factor analyses) suggested that the original 33-item CSSES could be shortened to a 20-item CSSES (CSSES-20) in the previous study with Taiwanese children aged 11 to 14 years.6 The CSSES-20 had 3 components accounting for 86.3% of the total variances. The first component, the influence of friends, accounted for 37.3% of the total variance and had 7 items. The second component, emotional stress, accounted for 25.0% of the total variance and had 6 items. The third component, opportunity to smoke, accounted for 24% of the variance and had 7 items.6 The preliminary psychometric tests supported that the CSSES-20 was equivalent to the 33-item CSSES (r= .99) when used in studies with children.6Because the CSSES-20 has not yet been tested with Taiwanese adolescents, it is necessary to establish reliability and validity of the CSSES-20 for use in an adolescent population.
METHODS
Sample and Setting
Random cluster sampling was used in this study. The sample was drawn from 61 middle schools consisting of 101,278 students located in 12 school districts in the city of Taipei, Taiwan. Two school districts were randomly selected both of which contained 6 middle schools. Researchers randomly selected 1 school from each school district, so that the final sample consisted of 2 of 12 schools. The selected schools had 2219 and 2244 students, respectively. The unit sampled was the class, and all students from the selected classes were invited to participate in the study. The number
of students in each class ranged from 30 to 33. Nine classes, 3 from each grade level (grades 7-9) were randomly selected from each junior high school. A total of 571 students (96%) provided complete data.
The age of participants ranged from 13 to 15 years, with a mean of 14.3 years. Of the research participants, 311 (54.5%) were boys and 260 (45.5%) were girls. In all, 201 students (35.2%), 190 students (33.3%), and 180 students (31.5%) were from the seventh, eighth, and ninth grades, respectively. Inclusion criteria in this study were as follows: (a) enrollment in the seventh, eighth, or ninth grade students in Taipei City public schools; (b) parental consent for the child to participate; (c) student assent after the parental consent has been obtained; and (d) ability to read, write, and speak Chinese.
Ferketich14 suggested that at least 200-300 study participants are needed to achieve a higher reliability of newly translated instruments. In addition, the appropriate ratio of research participants to items is 5 to 10 for the factor analysis method.15,16The numbers of items for the CSSES and the CSSES-20 are 33 and 20, respectively. Therefore, at least 330 participants were needed in this study. The sample of 571 in this study satisfied the minimum sample size at a power of .80 with an alpha of .05 by a factor analysis for the newly translated instrument.
Instruments
Two instruments were used to gather data. A demographic form with questions about self-report of smoking status (never smoked, experimentally smoke, and regularly smoke) and the CSSES were used in this research.
The Chinese Version of the Smoking Self-Efficacy Survey. The CSSES was translated by Chen et al6 to measure Taiwanese children’s perceived ability to resist smoking in each setting. The CSSES consists of 3 subscales, the influence of friends, emotional stress, and opportunity to smoke, with a total of 33 items using a 6-point Likert-type scale, ranging from 1 for ‘‘I am very sure I would smoke’’ to 6 for ‘‘I am very sure I would not smoke.’’ Each subscale had 11 items which were summed for the subtotal score with a possible score ranging from 11 to 66.
Twenty items from the 33-item CSSES were chosen to create the CSSES-20, the parsimonious version of the CSSES with a 6-point Likert-type scale ranging from 1 for ‘‘I am very sure I would smoke’’ to 6 for ‘‘I am very sure I would not smoke.’’ The CSSES-20 also had 3 subscales, the influence of friends, emotional stress, and opportunity to smoke, with 7, 6, and 7 items, respectively. Higher scores on each subscale indicated increased perceived ability to resist smoking.
Procedure
A pilot test of the CSSES-20 with 50 students aged 13 to 15 years was conducted before the formal study.
Students took 10-15 minutes to complete the CSSES- 20. The alpha coefficients of the 3 subscales, the influence of friends, emotional stress, and opportunity to smoke, were .91, .90, and .93, respectively, in the pilot test, and the preliminary findings supported the internal consistency of the CSSES-20.
The CSSES was used to collect data in the formal study. Before administering the questionnaires, the researcher explained the purpose of the study, described procedures for completing the instrument, and answered questions. The CSSES was completed by students in 20-30 minutes during the Health Education class. Twenty items from the 33-item CSSES produced the CSSES-20 and were used in analyses.
DATA ANALYSIS
Data were entered and analyzed by the Statistical Package for Social Sciences (SPSS Inc, Chicago, IL) version 16.0. Data were examined for accuracy, missing data, and outliers after entry. Construct validity was used to determine how well the instrument measured what it is supposed to measure.
Construct validity can be approached in several ways, such as examination of theoretical constructs and contrasted group approach.17 In addition, exploratory factor analysis procedures were found to be most appropriate for determining the identifiable attributes of a construct in the early stages of validation of a newly developed instrument18such as the CSSES-20.
Furthermore, because the CSSES-20 has not yet been tested in any studies involving Taiwanese adolescents, exploratory factor procedures with varimax rotation were used to validate the CSSES-20 scale. Pedhazur and Schmelkin19 suggest that only the items with factor loadings exceeding .40 are meaningful in applications of factor analysis. In this study, items were retained only if the factor loading was above .50. In addition, only components with Eigen values greater than 1.00 were retained for interpretation.
Further construct validity of the CSSES-20 was examined by using contrasted group approach.
Analyses of variance (ANOVAs) and Scheffes’ post hoc procedures were used to compare the mean scores of the ‘‘opportunity to smoking’’ subscale, the ‘‘emotional stress’’ subscale, and the ‘‘influence of friends’’ subscale among adolescent’s self-reported smoking status (never smoked, experimentally smoke, and regularly smoke). In addition, the Bonferroni adjustment procedures were applied to protect against inflated type I error. Therefore, the level of significance associated with the post hoc test for differences between the 3 smoking groups was set at .02 (.05/3).
Pearson correlation coefficients were used to examine
Table 1. The Chinese Version of the Smoking Self-Efficacy Survey (CSSES-20): Results of the Varimax Rotation and Item Analysis Component Loading (% Variance)
Item
Opportunity to Smoke
(27.1%)
Emotional Stress (26.9%)
Influence of Friends (26.4%)
Item- Subtotal Correlation
Shopping .69 .77
Watching TV .68 .81
Doing homework .83 .83
Riding bike .78 .86
Recess or after school .74 .83
Waiting for movies .77 .83
On the way home from school .83 .82
Angry .70 .83
Feel frustrated .78 .88
Worried .83 .89
Feel upset .83 .91
Feel down .86 .90
Feel sad .81 .88
At friend’s house .69 .80
See others smoking .63 .82
Someone offers a cigarette .73 .88
Look cool .72 .80
Feel grown up .71 .78
Best friend smoking .79 .87
Friends smoking .77 .86
the criterion validity of the CSSES-20. Cronbach’s alpha coefficients were used to determine the internal consistency of the scale.
RESULTS
Construct Validity of the CSSES-20
After exploratory factor analysis was conducted, 20 items loaded on 3 components accounted for 80.4% of total variance on the CSSES-20. The first component accounted for 27.1% of the total variance and had 7 items with the minimum item loading of .68.
Theoretically, these items represent adolescents who tend to smoke under certain circumstances. These findings are consistent with the previous study by Chen et al,6 which showed the same items loaded together on 1 component. The component is labeled as ‘‘opportunity to smoke.’’ The item-to-subtotal correlation ranged from .77 to .86 for the opportunity to smoking subscale. The second component accounted for 26.9% of the total variance and had 6 items with item loading ranging from .70 to .86. These items theoretically represent adolescents who intend to smoke under emotional stress. This finding is also consistent with findings by Chen et al.6 This component is labeled as ‘‘emotional stress.’’ The item- to-subtotal correlation ranged from .83 to .91 for the emotional stress subscale; the third component accounted for 26.4% of the total variance and had 7 items, which theoretically represent adolescents who tend to smoke when their friends smoke or
ask them to try smoking. The items loaded at .63 or above on the component. Again, consistent with previous studies,6 these items loaded on the same component and are labeled ‘‘influence of friends’’
(see Table 1). The item-to-subtotal correlation ranged from .78 to .88 for the influence of friends subscale.
The findings suggested that the CSSES-20 had good construct validity.
Contrasted group approaches were used to assess whether the construct measured by the instrument could be distinguished between groups with extremely different characteristics.9,17 Of the research partici- pants, 495 (86.7%) had never smoked, 60 (10.5%) had tried smoking, and 16 (2.8%) smoked regularly.
Table 2 presents the mean, standard deviation, and eta square of the opportunity to smoking subscale, the emotional stress subscale, and the influence of friends subscale for the 3 different smoking groups.
Analysis of variances of the opportunity to smoke subscale score, the emotional stress subscale score, and the influence of friends subscale score by smoking groups revealed significant differences among groups (p < .001) and also produced large effect size. Effect size was defined as the magnitude of the difference of independent variables with respect to dependent variables.20 According to Polit and Beck,9 eta square (η2) could be treated as an effect size index for the ANOVA tests. A large effect size is about .14 or above, a medium effect size is about .06, and a small effect size is about .01.20 All effect sizes for the 3 subscales were large. Scheffes’ post hoc analyses
Table 2. Mean, Standard Deviations, and Eta Square for the Subscales of Chinese Version of the Smoking Self-Efficacy Survey (CSSES-20) Across Different Smoking Groups
CSSES-20 Subscale Smoking Group n Mean SD F (df) Eta Square (η2) p
Opportunity to smoke Never smoked 495 40.3 4.4 208.9 (2568) .42 <.001
Experimentally smoke 60 30.4 8.1
Regularly smoke 16 21.4 5.3
Emotional stress Never smoked 495 34.5 4.0 165.3 (2568) .36 <.001
Experimentally smoke 60 26.1 9.0
Regularly smoke 16 17.4 7.4
Influence of friends Never smoked 495 41.1 3.0 113.0 (2568) .28 <.001
Experimentally smoke 60 37.1 6.9
Regularly smoke 16 27.8 9.7
showed that nonsmokers scored significantly higher on each subscale than those in either experimental or regular smoker groups. The above findings supported the construct validity of the opportunity to smoking subscale, the emotional stress subscale, and the influence of friends subscale.
Criterion Validity of the CSSES-20
Criterion validity is established by use of a criterion measure and may be quantified by the correlation coefficient between the 2 sets of measurements.17 Chen et al6suggested that the CSSES was a valid and reliable instrument with Taiwanese children, so the CSSES was used as a criterion in this study to examine the criterion validity between the original version (CSSES) and the parsimonious version (CSSES-20).
The correlation coefficient between the CSSES and CSSES-20 was .99, which supported the use of the parsimonious version (CSSES-20) as an equivalence scale to the original CSSES.
Reliability of the CSSES-20
Regarding test-retest reliability, 50 students from the pilot study agreed to take the same instrument 2 weeks later. The test-retest correlation coefficients for the 3 subscales were above .90. These results suggested that the CSSES-20 had good stability over time.
Cronbach’s alpha estimates of internal consistency for the opportunity to smoke subscale, the emotional stress subscale, and the influence of friends subscale were .90, .93, and .90, respectively. Results demon- strated acceptable levels of internal consistency.
DISCUSSION
Although an instrument may possess sound psy- chometric properties in the original version, it is necessary to reestablish the reliability and validity of the instrument when adapted for use with new populations. Based on the results of this study, the CSSES-20 demonstrated satisfactory construct valid- ity, criterion validity, stability, and internal consistency
reliability when used with a population of Taiwanese adolescents.
Consistent with results of previous studies,6 the factor analysis indicate that the CSSES-20 is a brief, validated, and reliable instrument composed of 3 subscales, the opportunity to smoke, the emotional stress, and the influence of friends. This structure was further confirmed in the sample of Taiwanese adolescents aged 13 to 15 years and also was found to be applicable to evaluate attitudes about smoking initiation. In addition to reliability and validity, response burden is a vital element to consider when conducting studies with young adolescents. A brief instrument can alleviate response burden for young adolescents and increase the recruitment and retention of study participants.
The results from the contrasted group procedures showed that there were significant differences in the mean scores of 3 subscales of the CSSES-20 across 3 different smoking groups (p < .001). Test results revealed that adolescents’ perceptions of their efficacy to resist smoking were associated with whether they would start smoking. Thus, this finding was consistent with those of previous studies.12,13,21 The study provides further evidence in support of Bandura’s self- efficacy theory,10 which predicts that an individual who perceives stronger self-efficacy for a behavior is more likely to perform tasks successfully, such as the task of resisting smoking under varying circumstances.
Limitations
Several limitations may restrict the broader appli- cation of the study. The sample of this study was randomly selected, but it only reflected 1 city in Taiwan. Therefore, the findings of this study should be limited to middle school students in Taipei. Caution must be used when generalizing the findings to young adolescents in other cities or countries. Furthermore, to validate the theoretical structures of the self-efficacy construct and support the construct validity of the CSSES-20, confirmatory factor analysis procedures are recommended for future studies.
Conclusions
The findings of this study suggest that the CSSES-20 is valid and internally consistent. It is a useful measure- ment tool for studying self-efficacy to resist smoking, as well as assessing and evaluating antismoking programs among Taiwanese adolescents. For future instrument development, the CSSES-20 needs to be tested with different populations and settings.
IMPLICATIONS FOR SCHOOL HEALTH
These findings have significant implications for providing school teachers and nurses with information about the relationship between self-efficacy to resist smoking and adolescents’ smoking behavior. By using the CSSES-20, school teachers and nurses can identify those adolescents who are at high risk for exhibiting low self-efficacy to resist the temptation to smoke.
Specifically, school teachers and nurses can utilize the subscales in the opportunity to smoke, the emotional stress, and the influence of friends to identify the proper intervention strategies for students who exhibit low self-efficacy to resist the temptation to smoke. The CSSES-20 can also be used for intervention purposes.
School teachers and nurses can design specific programs to increase students’ self-efficacy to resist smoking and resist social influences, to teach skills to refuse smoking, to enhance decision-making ability, to emphasize the health consequences of smoking and the benefits of smoking resistance, or to teach effective stress management skills and conflict resolution techniques to these high-risk students. Furthermore, it is important to consider social influences on children’s smoking behavior and extend antismoking programs to include siblings, peers, and parents.
The simplicity of the CSSES-20, with psychometri- cally sound validity and reliability, can help researchers increase the recruitment and retention of study par- ticipants. The instrument validation experience in Chinese may encourage researchers to consider using and assessing the merits of a revised and shortened instrument in English, as SSES has 36 items in its original form, and other languages.
Human Subjects Approval Statement
This study was approved by the Asia University institutional review board.
REFERENCES
1. Wen CP, Levy DT, Cheng TY, Hsu CC, Tsai SP. Smoking behaviour in Taiwan, 2001. Tob Control. 2005;14(suppl 1):
i51-i55.
2. Wen CP, Tsai SP, Cheng TY, Hsu CC, Chen T, Lin HS. Role of parents and peers in influencing the smoking status of high school students in Taiwan. Tob Control. 2005;14(suppl 1):
i10-i15.
3. Chen HS, Percy MS, Horner SD. Cigarettes: a growing prob- lem for Taiwanese adolescents. J Community Health Nurs.
2001;18(3):167-175.
4. Huang L, Tso Z, Yin J, Young R, Huang T. Tobacco smok- ing behavior among junior high school students. J Nurs.
1996;43(1):35-46.
5. US Department of Health and Human Services. Preventing Tobacco Use among Young People: A Report of the Surgeon General.
Washington, DC: US Government Printing Office; 1994.
6. Chen HS, Horner SD, Percy MS. Validation of the smoking self-efficacy survey for Taiwanese children. J Nurs Scholarsh.
2002;34(1):33-37.
7. Chen HS, Horner SD, Percy MS, Sheu JJ. Stages of smoking acquisition of young Taiwanese adolescents: self-efficacy and decisional balance. Res Nurs Health. 2008;31(2):119-129.
8. Patrick DL, Beery WL. Measurement issues: reliability and validity. Am J Health Promot. 1991;5(4):305-310.
9. Polit DE, Beck CT. Nursing Research: Principle and Methods. 7th ed. Philadelphia, PA: Lippincott Williams & Wilkins; 2004.
10. Bandura A. Self-efficacy mechanism in human agency. Am Psychol. 1982;37(2):122-147.
11. Bandura A. Health promotion by social cognitive means. Health Educ Behav. 2004;31(2):143-164.
12. Lawrance L, Rubinson L. Self-efficacy as a predictor of smoking behavior in young adolescents. Addict Behav. 1986;11(4):
367-382.
13. Lawrance L. Validation of a self-efficacy scale to predict adolescent smoking. Health Educ Res. 1989;4(3):351-360.
14. Ferketich S. Focus on psychometrics. Aspects of item analysis.
Res Nurs Health. 1991;14(2):165-168.
15. Ferketich S. Focus on psychometrics. Internal consistency estimates of reliability. Res Nurs Health. 1990:437-440.
16. Mishel M. Methodological studies: instrument development.
In: Brink PJ, Wood MJ, eds. Advanced Design in Nursing Research.
Newbury Park, CA: Sage; 1998:235-282.
17. Waltz CF, Strickland OL, Lenz ER. Measurement in Nursing and Health Research. 3rd ed. New York: Springer; 2004.
18. Dixon J. Grouping techniques. In: Munro BH, ed. Statistical Methods for Health Care Research. 3rd ed. Philadelphia, PA:
Lippincott Williams & Wilkins, 1997:310-340.
19. Pedhazur EJ, Schmelkin LP. Measurement, Design, and Analysis:
An Integrated Approach. Hillsdale, NJ: Lawrence Erlbaum Associates; 1991.
20. Cohen J. A power primer. Psychol Bull. 1992;112(1):155-159.
21. Chammah L. The Self-efficacy of Sixth Grade Students to Resist Smoking. [Unpublished thesis]. The University of Texas at Austin; 1995.
