Participants were eligible for the validation study if they were aged above 20 years, not pregnant, had not received an abnormal UCCS result in the last 6 months and had not a hysterectomy. The sample was recruited from four clinical sites: one gynecologic clinic and health screening center at one university hospital and two other out-patients based gynecologic clinics, in Gyeongju. The reason for the selection of sampling site was to collect the data of public and private UCCS. A total of 380 women were eligible to participate in this study and 339 (82.9%) women gave informed consent. Of these, 16 failed to complete the survey adequately and, as a result, the data from 323 women providing complete data was included in the final analysis.

In July 2008, the UCCSEQ was administered to 9 women who had a UCCS at a university health screening center and gynecological clinics. It took 10~15 min for them to complete the questionnaire. All women found the 29 items to be clear and easy to understand, and none had difficulty with the response categories.

Eligible women were invited to the questionnaire with assurance that there would be no disadvantages regarding the UCCS in case they declined to participate whenever they feel uncomfortable. After signing a written informed consent form, women completed a self-rating open-ended and closed questionnaire to assess demographic variables. Immediately after her UCCS, the participants were invited by research assistant into a separated place and left alone to complete the UCCSEQ and related psychometric measures. Participants were given a gift voucher valued of US$5 for completion of the survey. Data were collected during the period from August to September 2008.

All consenting participants completed a questionnaire package measuring UCCS embarrassment, general medical embarrassment, dispositional embarrassability, and Pap smear related negative emotion. A demographic questionnaire elicited information regarding self-reported age, education, household income, marital status and employment. Given the nature of UCCS experience, women's gynecological characteristics including reports of prior UCCS was also recorded.

The study was approved by the Dongguk University Gyeongju Hospital Institutional Review Board (DUGH 10-30). Women who met the criteria of this study were introduced by doctors or nurses to a research assistant, who explained the purpose of the study, the nature of participation, the ethical considerations of confidentiality and anonymity, as well as their voluntary nature of any involvement.

In the current analysis, Rasch model was applied to preliminary 29 items (Cho, 2003); (a) to improve the item-pool quality by calibrating item difficulty, Infit and Outfit Mean Square (MnSq), and point-measure correlation which enables detecting an inappropriate item that should be deleted, or replaced; (b) to examine the 7-point category functioning by calculating frequency of category use, observed person measure, threshold measure and outfit Mnsq for each category which estimate the rating scale structure and its utility (Bond & Fox, 2001). "Item difficulty" for the 29 items was estimated and each respondent's scores in logit unit were identified. In order to determine how well each item contribute to global embarrassment measurement, χ² fit statistics, known as "Infit" and "Outfit" MnSq were calculated. Item MnSq values≤1.2 are ideal (Linacre & Wright, 1991; Linacre, 2002). In this study, the predetermined fit criteria for item removal were Infit MnSq≤1.5 and Outfit MnSq≤2.0 (Ko et al., 2008). "Point-measure correlation" (equivalent to classical item-total correlation) was used to examine the discriminative power of each item (Linacre & Wright, 1991). The frequency of category use indicates how many persons have been rated in that particular category, estimating the relative locations of the person and items, 'targeting'. "Observed person measures" was examined to confirm whether response category 3 is expected to be higher than for category 2. "Threshold measure" is expected to increase with increasing category number. An "outfit MnSq" for each category was estimated to examine the consistency of use of the category.

Exploratory factor analysis on the 29 items was used to examine the potential sub-dimension of the UCCSEQ construct. Cronbach's α coefficients were calculated to establish internal reliability in the sub-scales and in the total 29-item scale. To assess the relative independence of the sub-scales, the Pearson correlation coefficients were calculated. Convergent and divergent validity was established by examining the correlations between the UCCSEQ, 100 mm VAS ratings, the MEQ, the SES, and the measure of Pap smear related negative emotions to determine. Known group validity was examined by comparing the CCCES scores of women who were high in SES score vs. those who had low score of SES. All statistical analyses were conducted using SPSS version 18.0 (SPSS Inc., Chicago, IL, USA) and WINSTEPS Rasch Analysis computer program +8.0.

Overall, the mean age of the sample (n=323) was 42.7 (SD=8.38), ranged 26~64 years. The majority of the respondents (86%) had a high school education or higher, only 8% were not married, and 52% had full-time jobs. 43.4% reported household's monthly income over 3.0 million won ($3,200). The majority had, on average, been pregnant 2.7 times (SD=1.41) and majority (81.7%) had at least two children. Most participants (70.6%) had no history of gynecological disease. More than half of the participants (53%) had their first UCCS during their thirties and the mean frequency of UCCS women reported in the preceding 3 years was 1.98 (SD=0.82), a frequency range of 1-3.

The item difficulty of the 29 items ranged from -.50 to .59 logits (Mean=0, SD=0.31) for the most difficult and least difficult items respectively. The person ability of the items ranged from -3.90 to 3.50 logits (Mean=-.43, SD=1.28). As shown in Table 1, all the 29 items adhered to acceptable fitting criteria for item inclusion (infit and outfit MnSq≤1.5 and 2.0, respectively) in range 1.83~.78, verifying as the unidimensional construct. The item-total correlations of the 29 items are adequate (all value≥0.30).

Across all the 29 items the frequency distribution of 7 category use was not skewed towards the lowest or highest categories, indicating good targeting. Summary statistics for the 7 rating-scale categories use are shown in Table 2. The "Frequency of Use" of all categories was high, indicating that no category was underused and also that there were sufficient observations of each category for a stable measurement. The frequency of category use from 1 to 7 was fairly even. Category 1 - 'not at all' agree to the statements of UCCSEQ items was used approximately 3 times more often than any of the other 6 categories (count=2258). This implied that the majority of subjects were hardly embarrassed on many statements of UCCSEQ items, supporting the result that the mean person ability (-.43) was lower than the mean item difficulty (0). The "observed person measures" increased from a low category representing low embarrassment to a high category, demonstrating no need for collapsing on rating categories. The "threshold measures" increased with the rating category value, indicating that the raters are most likely to choose from 1 up to 7 as their embarrassment increases. The total range of the 7 category thresholds was 1.75 logits, suggesting that the rating function could be improved with the category number lesser than 7. The "outfit MnSq" for all categories was ≤ 1.50, indicating no extreme misuse in any of the categories.

Principal Components Factor Analysis with Varimax rotation on the preliminary 29 items was carried out to determine the construct of UCCS embarrassment. Although the fit of the hypothesized four-factor model analysis was moderately high accounting for 73.7% of the total variance, there were several items misfitting to the underlying conceptual model with loadings ≤ .40. Inspection of the scree plot suggested that a three-factor solution (70.2% variance explained) was more readily interpretable. Consequently, a preliminary solution of three correlated factors were selected as the best fit to the data base d on eigen values ≥1 in which all communalities were above .30 and all items loaded at least .40 on only one factor. In a subsequent three-factor model analysis process, it was revealed that all items describing emotional-cognitive aspects of embarrassment loaded on only one factor, while items on verbal disturbances, behavioral predicaments and physiological responses tended to be double-loaded. Having considered that these three attributes can be broadly conceptualized as bodily embarrassment, two-factor model analysis was ultimately conducted, resulting in 66.2% of the total variance. Having deletion of four items (I feel like running away, firmly grabbed my hands, I winced my body, and my body is getting rigid and tense), all items related distinctly into one of the two factors with correlation coefficients≥.5. For the purpose of psychometric testing, these four items were dropped from the instrument because these items did not load properly on any factors, due to by either the factor-loading≤.40 or a low rotated factor score≤.1. Therefore, two factors containing 25 items were retained for further structural model fitting. The final 25 items loaded on two factors explained 67.3% of the total variance (Table 3). These two factors were labeled according to their contents as well as prior conceptual frameworks: "bodily embarrassment" displaying a various range of verbal disturbances, behavioral predicament and physiological impediment" (F1: 17 items) and "apprehension of unwanted or unintended social face" (F2: 8 items) and accounted for 36.0% and 31.3%, respectively (Table 3).

As shown in Table 4, both subscales (F1, F2) and total the UCCSEQ Cronbach's α demonstrated adequate reliability (Cronbach's α=.96, .97, .97, respectively).

Subscale-to-subscale and subscale-to-Overall UCCSEQ correlations were in the expected direction and of general magnitude (F1-F2: r=.71, p<.001; F1 & Overall UCCSEQ: r=.96, p<.001 and F2 & Overall UCCSEQ: r=.87, p<.001), indicating that the UCCSEQ had good construct validity.

Table 4 shows that the scores of both the two UCCSEQ subscales and overall UCCSEQ were positively associated with other closely related convergent validation measure-the 100 mm VAS of embarrassment, the two subscales of MEQ, the SES, and Pap smear related negative emotion. As would be expected (see Consedine et al., 2011), the UCCSEQ subscales (F1: z=4.95, p<.001; F2: z=2.68, p<.05) and overall UCCSEQ (z=3.89, p<.001) were more closely correlated with the MEQ's bodily embarrassment factor than with the judgment concern score. These findings provide good evidence of the discriminant validity of the subscale scores.

It was hypothesized that women showing high dispositional embarrassability, in comparison to those with a low score, would feel more embarrassment during UCCS. As expected (Table 5), women reporting higher trait embarrassability (with mean score ≥ 79 on SES) scored higher than those with lower trait embarrassability (with mean score <79 on SES), indicating that the UCCSEQ had good known-group validity.

As previous studies suggest that women may be embarrassed by either bodily predicaments (or exposures) (Keltner, 1997) or the apprehension of an unintended public-face (Ha, 2000), the 25-item UCCSEQ can also be constructed into two sub-domains. Bodily UCCS embarrassment can be assessed by verbal, non-verbal, and physiological indicators of embarrassment. For example, women tended to report verbal difficulties such as'voice trembled', 'voice is altered', 'become a stammered', 'become speechless', and 'awkward laughing'. In addition, women who were embarrassed reported a number of behavioral embarrassment manifestations such as'grip the sidebar of the examination bed tightly', 'bottoms keep moving upward from the examination table', 'keep touching face with hands', 'thighs get turned inward with a full strength', 'clench teeth', 'keep trying to pull skirt downward', 'tend to close eyes', and'tend to avoid eye contact'. Women also reported a number of reactive physiological changes such as'heart keeps beating', 'face gets blushed', and 'get thirsty'. Meanwhile, UCCS embarrassment by the apprehension of an unintended public-face can be manifested through aversive emotions such as, shyness, self-consciousness, and feeling uncomfortable, awkward, ashamed or uneasy arising prior to the screening.

In current study, the 29 items which were originally developed as the preliminary UCCSEQ in the previous study (Cho, 2003) was found to be as a valid measure within Rasch model. Meanwhile, given that the preliminary UCCSEQ was composed of four-domain; verbal, non-verbal, physiological and cognitive-affective embarrassment, it seems reasonable to assume that the 29 items (Cronbach's α=.96) were readily reduced into the 25-items (Cronbach's α=.97) under two distinct dimensions.

As expected, the two factors of the UCCSEQ were related to one another and showed a pattern of relations with other measures indicating convergent as well as discriminant validity. Our expectation, that the two factors of the UCCSEQ would be positively associated with the 100 mmVAS, general medical embarrassment, Pap smear related negative emotion and, to lesser degrees, dispositional embarrassment was confirmed (Table 4), with associations being stronger for conceptually more similar constructs. For example, correlations with other related measures of embarrassment ranged from .57 to .61, while those with dispositional embarrassment were between .36 and .32. These comparatively weak associations between the UCCSEQ and dispositional embarrassment suggests that although the UCCSEQ two factors are related to dispositional embarrassment, they are distinct from the dispositional measure and appear to capture the response to specific elicitors in the UCCS context.

Other findings provided evidence that the two factors of the UCCSEQ are separable and capable of assessing slightly distinct aspects of the general construct. First, dispositional embarrassment was more strongly correlated with 'bodily predicament' factor than to the 'apprehension of undesirable social face' factor (Table 4). In addition, the analysis of known-group validly showed that both bodily embarrassment and apprehensive embarrassment for those high in trait embarrassment were greater than those low in trait embarrassment (Table 5). It may be that bodily embarrassment in UCCS are more closely linked to dispositional response patterns and less contextually sensitive than embarrassments caused by the apprehension of unintended public face. On the other hand, items from the apprehension of unintended public face are closely linked to the interactional situations between health care provider and women during procedures. Above all, this study suggests that the bodily predicament factor should be more directly to UCCS behavior, perhaps particularly so among women high in trait embarrassment.

Rating scale analysis in this study showed that no collapse or addition of rating is necessary in CCES. However, given that the respondents tend to response to relatively easy rating categories and the total range of 7-category thresholds was smaller than expected, the category number might be reduced to something lesser than 7 in future study.

Having considered a careful and objective assessment of specific domains of embarrassment can guide systematic interventions to address possible sources of non-attendance in UCCS, the current study provides some preliminary guidance - it is the bodily predicament that seems to prevent screening attendance. Therefore, the specific bodily predicaments should be given special consideration as possible deterrents to UCCS non-attendance. Further work might continue this differentiation in other types of cancer screening such as breast self-examination or colorectal cancer screening (Consedine et al., 2011). In addition, given the strong preference of same-gender health personnel for intimate health care such as gynecological care (Lodge et al., 1997) or colorectal screening (Consedine, Reddig, Ladwig, & Broadbent, 2011), the subscale of bodily predicaments could be applied to address an unmet need for intimate bodily care in which embarrassing care is delivered by the opposite-gender health care provider.

It is assumed that bodily embarrassment, especially for those high in trait embarrassment, more likely to deter UCCS attendance, thus interventions for enhancing dramaturgical bodily skills regarding the UCCS-procedures (Cho, 2003; Henslin & Biggs, 1971) should be targeted at this risk-group. Even though we have concentrated on bodily embarrassment, it seems likely that discrepancies in the social image projected between health personnel and women screened, such as apprehension of immodest or stigmatization could deter screening attendance (Waller et al., 2009). Therefore, public consensus should be progressively established by mass media campaigns focusing on UCCS attendance as being a normative social behavior.

Although the above contributions presented, the results of the study should be interpreted with caution due to a number of potential limitations. Given that we used a self-rating questionnaire retrospectively in a cross-sectional design, the components of the UCCSEQ would not be guarantee for a real time self-reporting data or distinctive physiological and behavioral responses reported on other studies (Leary & Kowalsaki, 1995).

Another limitation of this study may be a lack of representativeness of the samples. The feeling of 'embarrassment' as a mechanism of self-control might reflect the extent of the civilization of the society (Edelmann, 1990). Women, therefore, belonging to an urban society may perceive the UCCS embarrassment differently from the current rural samples. The fact that the views of women screened by female doctors and were not or never screened in preceding three years were not reflected should also be considered. Given that the UCCS embarrassment seems to have universal/common attributes (Henslin & Biggs, 1971) while related factors could be varied (Edelmann, 1990), a psychometric validation study with a larger and more diverse sample in different cultural settings is needed.

Furthermore, the UCCSEQ merely measures the degree of cognitive-affective and bodily reaction which women instantly feel or react in embarrassing moments. Therefore, it is necessary to expand the instrument to cover comprehensively other related factors, such as antecedents or consequences, which were not specifically described in UCCSEQ items.