Semantic Examination of a Japanese Center for Epidemiologic Studies Depression: A Cautionary Analysis Using Mixed Methods

Abstract

Background

Cross-cultural research relies on the linguistic, conceptual, and semantic equivalence of instruments. Widely used translations of the Center for Epidemiologic Studies Depression (CESD) for cross-cultural samples should be analyzed to reaffirm conceptual and semantic equivalence.

Purpose

This methodological study aimed to discover and resolve problematic translations of a Japanese version of the CESD.

Design

Sequential explanatory mixed method design using spiraling integration.

Methods

Sample includes 34 first-generation Japanese women living in the US and 72 community-based women in Japan. Ethnographic analysis of the semantic meanings of items was followed by t tests to compare original and retranslated item means, as well as Cronbach’s reliability and corrected item-total correlations analyses.

Results

Six problematic items were retranslated: bothered, failure, hope, restless sleep, happiness, and “getting going.” Reliabilities for the CESD that included the new CESD item translations were the same; however, most item-scale correlations were higher for the revised translations across the two groups.

Conclusions

We conclude that both failure and “getting going” may be culturally bound items. Implications for cross-cultural and ethnographic nursing research include planning mini-ethnographic analysis when using translations to discover and reconcile cultural differences in connotations, motivations, and goals.

Background

In 2011, Canada had a foreign-born population of about 6.5 million, which is about 20.6% of the total population, making it the highest proportion of immigrants among the G8 countries. Asia is the largest source of immigrants (Statistics Canada, 2015). In the US, East Asian immigrant women have the highest distress of all immigrants (Takeuchi et al., 2007; Williams, 2002; Yeung et al., 2008), consistently reporting more depressive symptoms than Caucasians, with prevalence rates ranging from 18% to 40% (Cho, Nam, & Suh, 1998; Mui, Suk-Young, Chen, & Domanski, 2003; Saint Arnault, 2004a; Shibusawa & Mui, 2001). These women also have 50% to 80% higher risk for depression when compared with Caucasians (Yeung et al., 2002). Worldwide, suicide is the seventh leading cause of death among childbearing Asian women (World Health Organization, 2009). Despite high distress, East Asian immigrant women have the lowest overall service utilization rates of any cultural group in the US (Garland et al., 2005; Kimerling & Baumrind, 2005), and Asian immigrants to Canada are also much less likely to use any service, especially mental health services, compared with European Canadians (Tiwari & Wang, 2008). Some of the reasons for low service use include interacting cultural processes that may inhibit service utilization, such as group harmony values fostering indirect, nonverbal, and contextualized communication; somatic symptom distress patterns; stigmatizing values; and role expectations that restrict help-seeking options (Ro, 2002; Saint Arnault, 2009, 2014; True, 1990; Williams, 2002; Yeung et al., 2008).

As nursing research emerges as a leader in international and global research, investigators are turning their attention to best practices for instrument translation (Guo, Dixon, Whittemore, & He, 2013; Ljungberg, Fossum, Fürst, & Hagelin, 2015; Squires et al., 2013; Symon et al., 2013). This project seeks to add to this body of work by using mixed methods to examine the conceptual issues that surfaced during the conduct of an NIH-funded study on depression in first-generation Japanese women living in the US (RO1MH071307) (Saint Arnault & Fetters, 2011). This article uses ethnographic and quantitative methods to examine the semantic and connotative aspects of a widely used Japanese translation of the Center for Epidemiological Studies Depression (CESD) instrument (Shima, Shikano, Kitamura, & Asai, 1985). In our study, ethnographic methods are defined as the systematic study of the cultural dimensions of the experience and understanding of health, distress, and mental health through the use of methods that focus on the worldview (emic perspective) of the people under study. Ethnographic research is designed to capture the behavior, language, and health–culture interactions of individuals within a specific ethnic group (Mendlinger & Cwikel, 2008).

Cross-cultural and ethnographic nursing research relies on the equivalence of instruments with lived experience (Banville, Desrosiers, & Genet-Volet, 2000; Behling & Law, 2000; Kristjansson, Desrochers, & Zumbo, 2003; Van de Vijver & Poortinga, 1997). Assessment of suffering and distress across cultural groups requires the assumption that there is phenomenological equivalence between the symptoms on translated self-assessment indicators and the experiences of the subject and that the translation used accurately captures that equivalence. The equivalence types are often hierarchically organized, from the more general or technical to exacting levels of precision at the phrase or word meaning level. Content equivalence is the assurance that the research instrument, as a whole, measures a phenomenon that is relevant to the culture under study, while conceding that there may be variations in the specifics within the culture. Technical equivalence is the assurance that the data collection methods are the same between cultural groups. Criterion equivalence is also sometimes called phenomenological equivalence, assuring that the symptoms or characteristics of the phenomenon are the same for the two cultures and that the measure in question assesses all of the relevant symptoms and indicators necessary to adequately and accurately capture the phenomenon of interest. Conceptual equivalence refers to a meaning-level equivalence, such that the terms used in the study have analogous meanings, and that those meanings are relevant to both cultures (Wang, Lee, & Fetzer, 2006). Semantic equivalence is attention to the meaning of the items within an instrument and focuses on the cultural nuances that are variant between cultures. Semantic equivalence refers to the meaning of each item and attends to both the connotations, cultural idioms, and use of the words or phrases in the speech of each culture (Squires et al., 2013). This study examines the semantic equivalence of the Japanese translation of a depression instrument used in ethnographic mixed-method study of depression in the Japanese.

The phrase “idioms of distress” is used in medical anthropology to describe the culturally specific experiences of distress and suffering. The concept of idioms of distress is used as a perspective that can capture the multitude of distress experiences across cultures and seek to understand the meanings of them, which may be the presence of physical diseases or disorders, an experience of mental illness, a symbolic representation of interpersonal conflicts, or may be culturally coded ways of expressing social discontent (Kirmayer, Dao, & Smith, 1998). From this perspective, cross-cultural research on depression presents several challenges (Interian, Lewis-Fernández, & Dixon, 2013), including accurate assessment of symptom expression, symptom variability, the meaning of the symptoms, and stigma associated with the symptoms impacting response patterns (Delisle et al., 2014; Joffres et al., 2013; H. Li et al., 2010; Z. Li & Hicks, 2010; Limon, Lamson, Hodgson, Bowler, & Saeed, 2016; Sung, Low, Fung, & Chan, 2013; Wong, Lam, & Poon, 2010; Young, Fang, & Zisook, 2010). Despite acknowledgement that depression symptoms may vary culturally, many recent studies have confirmed the reliability of the use of western-developed depression instruments in cross-cultural studies (which were probably translated at the content equivalence level) (Butucescu et al., 2013; De Silva, Ekanayake, & Hanwella, 2014; Kojima et al., 2002; Malakouti, Pachana, Naji, Kahani, & Saeedkhani, 2015; Mazlan & Ahmad, 2013; Tapia, Wagner, Heredia, & González-Forteza, 2015; Ying, 1988). Despite this, many researchers are calling for more careful analysis of construct validity and semantic equivalence, seeking to understand the meanings of the depressive experience as well as the impact of these on interpretation of findings (Delisle et al., 2014; Hui & Triandis, 1985; Kim, Landis, & Cain, 2013; Kwakkenbos et al., 2013; Limon et al., 2016; Sidani, Guruge, Miranda, Ford-Gilboe, & Varcoe, 2010; Smit, Van den Berg, Bekker, Seedat, & Stein, 2007).

As we have seen, cross-cultural studies of depression have often translated and validated the CESD. This article examines one such translation of the CESD into Japanese. Shima et al. (1985) translated and validated the Japanese version of the CESD (Radloff, 1977). The CESD is a 20-item screening instrument that asks participants to rate whether they had experienced the symptoms described in each item during the previous two weeks on a 4-point scale from 0 (never) to 3 (more than five days), with higher scores (above 16) indicative of clinically significant depression. The analysis presented here is a subproject of a larger research study that aimed to understand culture and depression in Japanese women. In that larger project, we used the Shima Japanese translation of the CESD. Because there was such wide use of this scale and because it was available in Japanese, we used the published Japanese version.

For the first 180 surveys in the Japanese women in the US, we were unaware of any semantic nuances in our CESD instrument. We selected a conservative initial item-total correlation of .6 for our initial analyses. Three items in the CESD had item-total correlations below this threshold (see Table 2), and the statistical team suggested that we consider carefully these items: CESD1 (bothered), CESD11 (restless sleep), and CESD12 (happiness). They also suggested that we explore dropping CESD8 (hope) and CESD20 (“get going”) to improve the reliability of the instrument. While we knew that there may be several reasons for low item-total correlations, this was an ethnographic study of the concept of depression in the Japanese population and these data signaled a problem with the instrument or the translation. The aim of this article is to describe our use of mixed ethnographic and quantitative methods to understand the possible conceptual and semantic issues that might have been hidden in our translation of the CES-D.

Table 2.

CESD and NCESD mean and t test for Japanese in US and Japan.

Item: concept	US Japanese (n = 34)				Japan Japanese (n = 72)
	Mean (SD)				Mean (SD)
	CESD	NCESD	t	p	CESD	NCESD	t	p
CESD total	9.5 (10.9)	8.8 (10.5)	2.19	.04	8.5 (10.6)	7.9 (10.4)	2.43	.02
Item 1: bothered	.53 (.86)	.50 (.78)	.23	.83	.57 (.75)	.46 (.81)	1.473	.15
Item 8: hope	.63 (1.1)	.37 (.76)	2.11	.04	.69 (.89)	.58 (.81)	1.841	.70
Item 9: failure	.60 (1.0)	.27 (.69)	2.28	.03	.56 (.77)	.42 (.72)	1.637	.11
Item 11: sleep	.50 (.94)	.53 (1.0)	−.22	.83	.52 (.84)	.60 (.92)	−.798	.43
Item 12: happiness	.27 (.65)	.10 (.40)	1.31	.20	.69 (.93)	.37 (.74)	3.001	.00
Item 20: “get going”	.27 (.79)	.60 (.93)	−2.28	.03	.27 (.70)	.53 (.85)	−2.792	.01

CESD: Center for Epidemiologic Studies Depression.

Methods

Design

This research is a subproject of a larger ethnographic mixed-method study that examined culture, distress, and help seeking for Japanese women in the US and in Japan. The parent project was a concurrent mixed-method design aimed at examining the relationship between distress experiences, cultural interpretations, social structures, and help seeking for a random sample of 250 first-generation Japanese women (Saint Arnault & Fetters, 2011). However, this research is part of a broader program of research that includes 30 years of fieldwork with the Japanese in Japan and in the US (Saint Arnault, 1998,Saint Arnault, 2004, 2014). Details of these field studies and the parent study are reported elsewhere; however, in summary, we gathered both ethnographic and quantitative data concurrently and integrated the data at the end of the project. We used quantitative measures to examine the relationships among known variables, which included symptoms, causal beliefs, stigma attitudes, values, and help seeking. We also used ethnographic interview methods to expand our knowledge about aspects of those same variables that were unknown, including how these processes operated in the women’s lives.

In the subproject reported here, we employed the sequential explanatory design, using a spiraling integration. Spiraling, as a mixed-method strategy, is the strategic oscillation or alternation between one type of data and another for the purpose of deeper understanding. In this method, the researcher integrates understanding gleaned from the analysis of one method toward the analysis of the subsequent methods, providing a sequential “building up” of understanding (Mendlinger & Cwikel, 2008) (see Figure 1). This spiraling for us was the sequential analysis that began with the identification of the three items that had lower item-total correlations, followed by a mini-ethnographic study of the semantic nuances in the concepts in the CESD, and the resulting retranslation of six new CESD (NCESD) items. Then, we gathered quantitative data using both the old and newly translated items, and finally, integrated both sets of findings with an ethnographic interpretation, which is presented in our discussion.

Figure 1.

Spiraling analysis.

Participants

The full sample for the parent project was 209 Japanese women in the US who were immigrant women who came to the US for reasons other than their education, were part of the population of families working in the Japanese Automotive industry, and were sampled from primary care sites and the general community. We focused on Japanese women because they are at risk for mental health problems due to separations from extended families, intergenerational conflict, and family system and role relationships changes. Men were excluded because this study aimed to examine social and cultural factors specific to the women.

The sample for this subproject was the 34 Japanese women in our U.S. sample who could still receive the revised CES-D items in their survey after we made any necessary changes. In addition, because we wanted to test the new translations, we also sought a nonimmigrant sample of Japanese women in Japan. This second sample was 72 Japanese women living in Japan recruited from publically available names from a voter registry in Shiga prefecture. All research staff used random start interval sampling to select from names on their lists and mailed survey packets to the selected women. Because this subproject was exploratory, statistical power for these analyses was not conducted.

Procedures and data collection

All procedures and materials for this study were approved by the university institutional review boards in the US and Japan. All communication with women was in Japanese, including all written contact, consents, surveys, and all research materials. Translation of most of the research instruments in the parent study was carried out with a team of three graduate-level bilingual staff from three different regions of Japan, with the assistance of two other bilingual Japanese research assistants. One graduate level staff was bicultural, and the others were either international graduate students or immigrants. One of the three graduate level staff had training in mental health. Our procedure was that two translators translated each instrument independently while making systematic notations of difficulties with translations of concepts. Next, we rotated instruments to another translator for back translation into English, and the staff again made careful notes of translation difficulties and nuances. Then, we met as a team and began detailed reconciliation of the differences in the translations, back-translations, and notations. Items of concern received a second round of translations. Our overall translation aim was typical daily speech, without clinical jargon and local dialect differences, as well as conceptual and semantic equivalence.

Our study examined depression, and we selected the translation by Shima et al. (1985) for our depression instrument because the scale had been used in numerous studies and had been identified in publications as one of the only depression scales that had received rigorous back translation for semantic equivalence (Furukawa, Hirai, Kitamura, & Takahashi, 1997). However, research has shown that some East Asian groups may have cultural biases against endorsing positive affect, possibly affecting the response pattern on the CESD (Cho, Kim, Cho, & Shin, 2007; Iwata & Roberts, 1996; Iwata, Roberts, & Kawakami, 1995; Noh & Avison, 1996; Noh, Avison, & Kaspar, 1992). Therefore, we reversed the wording for the four positive items of the Japanese version of the CESD as suggested by Cho, Noh, and Iwata. In order to make these reversals, we used our usual protocol involving independent translations, independent back translations and reconciliation meetings. However, we translated the items for those four positive items only. All instruments in the entire final survey were field tested with a panel of eight Japanese women, each of whom took the survey in total, then received a cognitive interview about semantic or cultural issues, as well as readability and clarity. They did not find problems in any of our instruments. It should be noted, however, that the cognitive interviewing did not ask women about the cultural relevance of the questions in the survey or the cultural appropriateness of them but only addressed natural speech, readability, and clarity.

Mini-ethnography

After our discovery of the six problematic items, we used mini-ethnography to examine the concepts within the CESD. A mini-ethnography is a focused interview technique aimed at gathering data about cultural beliefs, concepts, and practices related to a clinical problem (Nastasi et al., 2015). We used the mini-ethnography within the research team to understand the cultural aspects that may be related to the semantic and connotative concepts we identified as problematic. However, our ability to use this approach was because the lead author has over 30 years of fieldwork in Japanese culture. These ethnographic analyses focused on concepts contained within each of the problematic items, and the author acted as an ethnographic investigator, carrying out group interviews with the five research staff, who acted as “key informants” or “culture brokers.” We used a “compare and contrast” framework that sometimes included a review of literature about a concept in Japanese or Western culture. After these meetings, we finally selected our final new translations. We used two criteria to ultimately employ a new revised translation: the independent translations and back-translations revealed that there was a concept in the translation that differed from what our ethnographic discussion deemed best to use, and that it was possible and appropriate to translate the original English CESD concept into Japanese.

Quantitative analysis

We used quantitative data gathering of both the original 20 items (CESD) and the additional six retranslated New CESD (NCESD) items for remaining 34 women in the U.S. Japanese group, as well as for 72 women in Japan. All analyses reported here are only for those women who took the original and the revised Japanese CESD items (N = 106). We used quantitative psychometric analysis of the old and revised scales including alpha reliability analyses, as well as a review of item-total correlations. Our quantitative analysis also included t tests to examine whether there were within-group and between-group differences in CESD total mean scores for the original and revised scales for both the US and the Japan sample. We examined whether there were within-group differences between Cronbach’s alpha reliability scores for the original and revised scales, differences between item-total correlations, and differences between item means. Finally, we integrated both the ethnographic and quantitative findings in the discussion of the overall findings, adding literature to explain our findings.

Results

Mini-Ethnography Results and Revised Translations

Table 1 shows the original English CESD, the published Shima Japanese, our back-translation of their Japanese, our reversal, our new CESD (NCESD) Japanese and English, and the means and inter-item correlation ranges for the Japanese in the US and Japan. As noted earlier, the final NCESD items are the result of our mini-ethnographic analyses of these concepts.

Table 1.

Shima CESD and revised NCESD Japanese, with reversals and back translations.

	English CESD	Shima Japanese original	Shima back-translation	Revised Japanese	Revised back-translation
1	I was bothered by things that usually do not bother me.		I feel tired and get annoyed by things that do not usually make me feel that way.		Things that I am not usually irritated by irritate me.
8	I felt hopeful about the future.		I think positively.	Used reversal
	Reversal: I felt hopeless about the future.		My thinking is pessimistic.		I do not have hope for the future.
9	I thought my life had been a failure.		I obsessively think (ruminate) about my past.		I think my life was a failure.
11	My sleep was restless.		I cannot fall into a sleep; It takes time to fall asleep.		I cannot have a good and quiet sleep.
12	I am happy.		I am without complaints or unhappiness in my life.	Used reversal
	Reversal: I am unhappy.		I have complaints in my life.		I feel unhappy.
20	I could not get “going.”		I cannot focus on my work; I feel distracted so that I cannot get my work done.		I cannot get myself to move to do things; I do not have the energy to do things.

CESD: Center for Epidemiologic Studies Depression.

CESD1 captures the concept of being bothered. The original Shima translation used the concept “annoyed” and also included the concept of “tiredness.” In our ethnographic meetings, we noted that being irritated or bothered is a significant part of Japanese distress and depression. We discussed that while being tired or fatigued is an also important symptom, it is possible to feel bothered or irritated without being tired. Therefore, in our NCESD1 revision, we used the concept of “irritated” as the Japanese equivalent to “bothered,” and our NCESD1 was “Things that I am not usually irritated by irritate me.”

CESD8 was one of our reversal items and captures the concept of hope (or hopelessness in our reversal). The original Shima translation captured the concept of “thinking or viewing things positively.” Our original reversal was “My thinking is pessimistic.” Our ethnographic discussions centered on why we had originally used the attitude of “optimism” and the reversal concept of “pessimism.” Our discussion revealed a strong stigma about the concept of hopelessness for the Japanese. This aversion to using the terms hope and hopelessness was related to a Japanese cultural sense that as long as one is attached to their group, there was hope. Our discussion also distinguished the Japanese concept of hope from an American ideal of autonomy, understanding that the American concept of hope relates to personal control, while the Japanese concept of hope references an embeddedness with in a group. We ascertained that translation using optimism and pessimism might have been part of a desire to de-stigmatize the concept, softening it so that women would not be offended. Next, we distinguished pessimism from hopelessness, deciding that optimism and pessimism were attitudes or dispositions, while hopelessness was a feeling of insecurity, a lack of confidence, or negative expectations when one imagines their future. Therefore, in our NCESD8 revised reversal, we used the concept of “not feeling hope,” and our NCESD8 was “I do have hope for the future.”

CESD9 captures the concept of feeling that one's life was a failure. In the original Shima translation, the concept tapped was obsessive thinking about the past (or rumination). In our ethnographic meetings, we discussed the Japanese tendency to “worry” about others opinions and noted that worry, and even rumination, was very common for average Japanese people. We also discussed the way that the Japanese understand mistakes or failure. In Japanese culture, the informants noted that past mistakes are often recalled as a way to work towards being “a better Japanese person.” The team thought that worry and rumination were related to anxiety about social relationships. In the end, we decided that it is possible to think about failure without necessarily ruminating about it and that the concept of failure distinguishes depressive thinking from a more typical Japanese “worrying about past mistakes.” Therefore, we used the concept of “failure in life,” and our NCESD9 was “I think my life is a failure.”

CESD11 captures the concept of restless sleep. In the original Shima translation, they used the concept of “being unable to fall asleep.” In our ethnographic discussions, we noted that while sleep quality is an important aspect of functioning, the team did not believe that this concept revealed any important cultural differences. We decided that falling asleep was not equivalent to restlessness while sleeping, and therefore used the concept of “good and quiet” sleep, and our NCESD11 item was “I do not have a good and quiet sleep.”

CESD12 was another one of our reversal items and captures the concept of “happiness” (or unhappiness in our reversal). In the original Shima translation, they used the concept of “no complaints in life,” and our reversal was “I have complaints in life.” In our ethnographic meetings, we discussed the concept of happiness in Japanese culture. The team described that, while they thought that Japanese people certainly want to be happy, the feeling of happiness in Japanese culture might be understood as a transient experience and that a more typical desire was life satisfaction, peace, or contentment. The team discussed several movies, haiku poetry, and even Buddhist concepts that promote acceptance of sadness as a state in life and impermanence as inevitable. The team explained the concept of the impermanence of things ( mono no aware), saying it could also be loosely translated as “understanding the gentle sadness about things.” In the end, the team agreed that the concept of happiness and unhappiness is a natural and directly translatable concept in Japanese, despite the fact that the cultural emphasis on the feeling might not be equivalent for the Japanese. Therefore, in our revised NCESD12 translation, we used the concept of “unhappy,” and our NCESD12 item was “I am unhappy.”

CESD20 captures the concept of “getting going.” The original Shima translation used the concepts of “cognitive focus and distraction.” The concept of “getting going” generated an interesting discussion among the team. First, we talked about the Japanese concept of “ki” or vital energy. The idea of ki being “blocked ( , ki ga omoi)”, or sluggish ( , ki ga meiru), are idioms for depression, and one might even ask “ (O genki desu, ka)” which loosely translated as “Is the source (gen) of your vital energy (ki) good, solid, robust or healthy? Are you well?” We also discovered that the Japanese have herbal remedies that focus on mobilizing or energizing the ki, and may be used for treatment of depression. We noted that “getting going” is an American idiom about “forward movement,” “feeling an ability to move,” and suggesting carrying out activities of daily living. We thought the “getting going” idiom connoted motivation, energy, and drive. Since ki is a familiar concept in Japanese and was similar enough to the idea of motivation, will, or energy, we decided that we could use the concept of ki in the translation. We also decided that distraction or attention span were not the same as having energy or “getting going.” Therefore, in our NCESD20 revision, we used the concept of “energy to move and do things,” and our NCESD20 was “I cannot get myself to move to do things; I don’t have the energy to do things.”

Quantitative analysis of revised items

Sample characteristics

The mean age for the women in Japan was higher (M = 49.8, SD = 14.5) than the women living in the US (M = 40.0, SD = 8.2), p = .00. In our U.S. subsample, 44% had a college education, compared with 23% in the Japan sample. The women in the U.S. sample generally could not work because of visa restrictions, and 83% were unemployed or volunteers, while only 50% of the Japan sample was unemployed.

Between-group comparisons

Between-group comparisons of CESD scores using independent samples t-test revealed no statistical differences between women in the US and Japan for either the CESD or the NCESD means. Cronbach’s alpha reliability coefficients for the original and the revised scale for both the Japanese women in the US and Japan were the same (α = .94).

Within-group Japanese in the US

Findings from the quantitative analysis are included in Table 2. The original CESD total mean (M = 9.5, SD = 10.9) was significantly higher than the revised NCESD total mean (M = 8.8, SD = 10.5); t(32) = 2.19, p = .04. Cronbach’s alpha reliability coefficients for the original scale and the revised scale were the same (α = .94).

Pairwise t tests of the original and revised translation item means revealed that four of the item means were significantly different. The original Item 9 failure was significantly higher (M = .60, SD = 1.0) than the revised NCESD9 (M = .27, SD = .69); t(29) = 2.28, p = .03. The original Item 20 “get going” was significantly lower (M = .27, SD = .79) than the revised NCESD20 (M = .60, SD = .93); t(29) = _2.28, p = .03. The original Item 8 hope was also significantly lower (M = .63, SD = 1.1) than the revised NCESD8 (M = .37, SD = .76); t(29) = 2.11, p = .04.

The item-total correlations for four of the six revised items were higher for our revised version: Item 1 bothered (CESD1 r = .59; NCESD1 r = .68); Item 8 hopelessness (CESD8 r = .84; NCESD8 r = .89); Item 11 restless sleep (CESD11 r = .24; NCESD11 r = .59); and Item 20 “get going” (CESD20 r = .71; NCESD20 r = .85). Two of the item-scale correlations were lower for our revised items: Item 9 failure (CESD9 r = .75; NCESD9 = .53); and Item 12 unhappiness (CESD12 r = .60; NCESD12 r = .59) (Table 3).

Table 3.

CESD and NCESD Item-total correlations for Japanese in US and Japan.

Item: concept	US Japanese (n = 34)		Japan Japanese (n = 72)
	Item-total r		Item-total r
	CESD	NCESD	CESD	NCESD
Item 1: bothered	.59	.68	.72	.76
Item 8: hope	.84	.89	.86	.83
Item 9: failure	.75	.53	.74	.79
Item 11: sleep	.24	.59	.57	.62
Item 12:happiness	.60	.59	.68	.78
Item 20: “get going”	.71	.85	.75	.81

CESD: Center for Epidemiologic Studies Depression; NCESD: New CESD.

Japanese women in Japan

The mean for the original CESD total scores (M = 8.5, SD = 10.6) were statistically higher than the revised NCESD (M = 7.9, SD = 10.4); t(71) = 2.43, p = .02 for women in Japan. Cronbach’s alpha reliability coefficients for the original and the revised scale were the same (α = .94).

A pairwise t-test of the original and revised item means revealed that two-item means were significantly different. The original Item 12 happiness was significantly higher (M = .69, SD = .93) than the revised NCESD12 (M = .37, SD = .74); t(64) = 3.00, p = .00. The original Item 20 “get going” was significantly lower (M = .27, SD = .70) than the revised NCESD20 (M = .53, SD = .85); t(64) = 2.79, p = .01.

The corrected item-total correlations for five of the six revised items were higher for our revised translations: Item 1 bothered (CESD1 r = .72; NCESD1 = .76); Item 9 failure (CESD9 r = .74; NCESD9 r = .79); Item 11 restless sleep (CESD11 r = .57; NCESD11 r = .62); Item 12 unhappiness (CESD12 r = .68; NCESD12 r = .78); and Item 20 “get going”(CESD20 r = .75; NCESD20 r = .81). The corrected item-total correlation for revised Item 8 hopelessness was lower than the original (CESD8 r = .86; NCESD8 r = .83).

Integration and discussion

This research was an extension of decades of ethnographic fieldwork and interviewing by the lead author aimed at understanding Japanese culture, gender, distress, and help seeking. Despite the fact that our research was focused on using ethnographic approaches to understanding the meaning of depression in a Japanese population, even we took the translated CESD items at face value. This kind of conceptual blindness, cultural bias, or ethnocentrism is a familiar trap in cross-cultural research. Researchers and translators will always hold biases and assumptions, and it is human nature to be blind to them. Discovering biases requires rigorous personal scrutiny of assumptions, as well as practice level policies and procedures. For example, one of our practice level policies was our translation procedure, which involved having bilingual translators make notes about their cognitive processes as they translated and back translated. This served us well throughout all aspects of the project, especially when we discovered the semantic differences in the CESD instrument. The research presented here reminds us that nursing and health research might be well served to incorporate a rigorous translation protocol, even when the research uses translated instruments.

Item-scale correlations were higher for most of the revised translations across the two groups for the CESD concepts, including feeling bothered, restless sleep, and unhappiness. However, we had mixed results for feelings of failure and “getting going.” Moreover, the reliability of the scales remained the same. While these differences were small, this study gives some evidence that careful attention to culturally based semantics can make a difference. Moreover, this process greatly informed our overall understanding of depression, as well as how culture affects the items within a scale. These findings reinforce for researchers that attention to cultural aspects of symptomology is critical for cross-cultural research, suggesting that attention to semantics can be an important part of the translation and interpretation process. It is likely that the concepts that had mixed results (feelings of failure and “getting going”) are culturally understood phenomena, making them likely to involve connotative or idiomatic dimensions. For example, there is a body of psychological research illustrating cultural differences in perceptions of success and achievement, as well as general concepts of self-esteem, within Japanese culture (Falk & Heine, 2015; Heine & Lehman, 1999; Heine, Takata, & Lehman, 2000; Kitayama, Markus, & Kurokawa, 2000; Markus, 1991; Markus, Mullally, & Kitayama, 1997). This may help explain the mixed results in the revised failure translation. However, the presence of idioms within either the English or the Japanese CESD was not evident to us as we embarked on this project. The best example of this is the concept of “getting going.” From an ethnographic point of view, the use of this concept in the American developed CESD instrument may reference a historical, perhaps Protestant, American value that it is good to be clear, focused, and on a mission (Kang, 2009). Along a similar line, other research on western values embedded in western biomedicine has documented the use of idioms that reference the human body as being “like a machine” (Martin, 2001). It is possible that the concept of “getting going” may reference humans as being “like a car,” and the concept of not getting going is like a car that will not start. Finally, it is entirely possible that the Shima team selected the concepts of optimism, obsessive thinking, being unable to fall asleep, and not having complaints in life because they were important dimensions in Japanese depression. This research shows that investigators need to have a knowledge of the cultural aspects that are under study, as well as how their instrument’s concepts relate to those cultural beliefs and practices. In addition, the field of cross-cultural nursing research could benefit from more emphasis on the way that western derived instruments do or do not match cultural concepts of health.

Psychological research on health, psychological well-being, and life satisfaction has recognized that there are cultural differences in the meaning of happiness (Tov & Diener, 2009; Uchida & Ogihara, 2012). Like evaluations of success and failure, evaluations of good feelings are based on broader evaluations of cultural goals, motivations, and ideals (Kitayama et al., 2000; Markus & Kitayama, 1991). The Japanese have been described as a group-oriented culture (Kitayama et al., 2000; Kitayama, Markus, Matsumoto, & Norasakkunkit, 1997; Markus & Kitayama, 1991). To the extent that happiness is a value for the Japanese, feeling happy is rooted in the operations of the group and one’s contentment with their place within that group. Other studies have explored the interaction between cultural values on the one hand and response bias to surveys about happiness and psychological wellbeing on the other (Veenhoven, 2012). Indeed, we knew this going into this project, which was why we used the culturally revised version of the CESD, which purports to eliminate the tendency to under-report positive emotions. We recommend that investigators consider the biases and assumptions about what constitutes health, as well as illness, from a cultural lens.

Cross-cultural nursing research is uniquely situated to examine health and functioning. However, our research reminds us to be aware that concepts of normal or preferred patterns of sleep, energy, motivation, or feeling “bothered” or irritated are rooted in cultural expectations about what constitutes health and optimum functioning (Behling & Law, 2000). For example, sleep quality can be alternatively interpreted as falling asleep, staying asleep, waking up too early, feeling rested after sleeping, and/or sleeping restfully. Similarly, what qualifies as motivation depends on normative expectations about temporal orientation and productivity. Health-related emphasis on motivation implies values for vitality, zest for life, and general enthusiasm. Of course, all of these nuances cannot be captured in a 20-item instrument. However, understanding implicit assumptions about health and functioning is critical when carrying out cross-cultural studies of health, quality of life, and functioning.

The translation of hopelessness generated the most vigorous discussion within our translation team. First, we had a discussion of the Japanese idiom of shikata ga nai ( ), which refers to a disposition, attitude, or philosophical position toward difficult or tragic circumstances (Long, 1999). It translates as “it can’t be helped,” or “nothing can be done about it.”¹ However, while there may be an element of hope within that phrase, we discussed that hope within an American cultural value system was related to notions of individual will, personal power, and social autonomy (Del Vecchio Good, Good, Schaffer, & Lind, 1990). In group-oriented cultures, confidence about future security rests on connection with, and optimal functioning of, one’s group (Kitayama et al., 1997, 2000; Markus & Kitayama, 1991). Therefore, endorsements of feeling hopeful for the Japanese may relate to both personal will, but also may reference a sense of group-level security. This complex conceptualization of hope may make answering this question difficult, or at least multifaceted, from a Japanese perspective. We recommend that investigators carry out focused or mini-ethnographic studies about the concepts they want to understand when carrying out cross-cultural research. These findings can inform translation and interpretation of findings.

We recognize many limitations of this study. First, the sample sizes were small and the US and Japan samples were not equivalent for age, education, or employment. The larger study from which this data were drawn aimed to understand women’s immigration experience because gender greatly influences stress, coping, and engagement with social networks. Therefore, we did not gather data from men in this research, although future research should compare and contrast findings in these populations. Our US sample characteristics were consistent with migration patterns for immigrants coming to the US as part of the Japanese automobile industry (Black, 1994; Briody & Chrisman, 1991; Flory, 1989; White, 1988), who report that these migrant women average around 35 years old and stays an average of five years, suggesting very limited acculturation on the whole. Women who come to the US are also typically more educated than the general Japanese population. Secondly, we had only 34 women in the US and 72 women in Japan who received our revised NCESD items. Therefore, we may not have had sufficient power to draw generalizations.

Because we had not carried out appropriate analysis of the semantic equivalence of the CESD items for the reasons explained above, this work represents an internal critique and strategy to appropriately examine our depression instrument, albeit late in our study. Even with our due diligence, this effort did not yield major changes in item-total correlations. What this work did do was to elevate our confidence that we could understand the patterns we were seeing, and that these patterns related to the concepts under study. It also contributed to our understanding of one approach that could be used to ensure the reliability of the instruments used in our study. However, all women who received the six retranslated items also received the 20 original CESD items, and this may have influenced their responses. Finally, the CESD scores in both groups were very low. Despite our efforts to eliminate positive affect item bias by reversing positive items, we believe that women had a general tendency to under-report symptoms on this instrument, probably related to the stigma of the topic of depression. While this study reveals interesting patterns and illustrates potential issues, we recognize that these results would need to be replicated in larger and more complex psychometric studies. What we provide here is a mixed-method model to begin such inquiry.

Conclusion

Cross-cultural nursing research that aims to move beyond description to theory testing and interventional research relies on instrumentation to draw conclusions about phenomena of interest to nursing. While translation and back-translation are often the “gold standard,” we recommend ethnographic approaches that help researchers attend to conceptual and semantic equivalence in the concepts held within scales and instruments. This requires in-depth knowledge of the cultures under study, and examination of connotations and idioms that may be hidden in the original instruments, as well as those in the culture under study. Careful attention to protocols that bring this material to the surface can lead to more trustworthy results.

Biographies

Denise Saint Arnault, PhD, RN, FAAN, is an associate professor at University of Michigan. Her research centers on how cultural factors interact to influence illness and help seeking, examining the help-seeking trajectory in widely variable phenomenon, including mental and physical illness, homelessness, psychosocial distress, and trauma recovery.

Hiroyo Hatashita, PhD, RN, is a professor and Head of the PhD program in Human Science at Mei University. She is a public health nursing researcher examining women’s health in several Japanese rural communities for over thirty years. Her research includes interpretive phenomenology and mixed methods to study domestic violence, disaster preparedness, and Japanese Brazilian women who live in Japan.

Hitomi Suzuki, MSN, is a researcher and educator at Kyoto Gakuen University. Her research focuses on Japanese women’s health including the cultural and social help seeking for Brazilian Japanese in Japan.