The methodology used in collecting data from the three samples surveyed in this paper is described briefly below. Further details can be found in previous publications [17, 18] (Sample 1) and [15, 19] (Samples 2 & 3). Ethics approval was granted for the collection of all datasets by the Human Research Ethics Committee of the Australian National University.
Sample 1 (National)
Attitudes and information relevant to depression were collected from a total of 1001 Australian adults in a national face-to-face household survey during 2003 and 2004 of 3,998 Australian adults aged over 18 years. Households were sampled from 250 census districts covering all Australian States and both rural and metropolitan areas. Interviewers made up to five callbacks in metropolitan areas and three in rural areas. Response rate, computed as a percentage of the total number of contactable and physically available qualified respondents was 34%.
Respondents to the survey were presented with a vignette describing a person with depression (see Appendix 1). Half of the participants were administered a male version of the vignette and the other half a female version. The vignette satisfied DSM-IV and ICD-10 criteria for a Major Depressive disorder.
Respondents were asked a series of questions about the disorder depicted in the vignette. Stigma associated with the disorder was measured using a vignette version of the Personal and Perceived scales of the Depression Stigma Scale (DSS) [15, 18]. The DSS-Personal subscale comprises 9 items and is concerned with the respondent's personal attitudes to depression and the DSS-Perceived stigma scale comprises 9 items assessing the respondents beliefs about the attitudes of others to depression. Scores for each subscale range from 0 to 36. Higher scores indicate greater stigma. The DSS scales have previously demonstrated acceptable internal consistency and test-retest reliability . Respondents also completed a 5-item attitudinal Social Distance scale . This scale measured self-reported willingness to make contact with the person in the vignette. Respondents rated their willingness to 1) move next door to the person in the vignette; 2) spend an evening socialising with the person; 3) make friends with the person; 4) work closely on a job with the person; and 5) have the person marry into the family. The respondent rated each item on a 4-point scale: 'definitely willing', 'probably willing', 'probably unwilling' and 'definitely unwilling'. A participants score on the scale was the mean rating across the 5 items (range 1 to 4) with higher scores indicating greater social distance.
Respondent capacity to recognise the problem depicted in the vignette as depression was evaluated . Respondents were also asked if they had heard of Australia's national depression initiative, beyondblue , if they had 'ever had problems similar' to those of the character in the vignette, if their family had ever had such problems and if they had 'ever had a job that involved providing treatment or services' to a person with a problem like the character in the vignette. Information was collected about the respondent's sex, age category in years (18–19, 20–24, 25–29, 30–34, 35–39, 40–44, 45–49, 50–54, 55–59, 60–64, 65–69, 70–74, 75+), educational background, whether or not the respondent was born in Australia and the postcode of the respondent's residence. The latter was used to classify the locality of each participant according to the 2001 Australian Standard Geographical Classification (Major Cities, Inner Regional, Outer regional, Remote, and Very Remote) that applied to the majority of the population in the geographical region for the postcode. Respondents were also asked about their recall of media stories about depression and their current health status, as well as their views on which of a selection of options was the likely cause of the problem in the vignette, what might be helpful for treating the problem and their view on the likely prognosis for the person depicted in the vignette. However, these data are not the considered in the current paper.
Of the respondents, 59.9% were women, 22.5% had completed a Bachelor's degree or higher educational qualification, 67.8% resided in a major city and 26.2% were born outside of Australia. Median age was in the range 45 to 49 years. 33.1% reported that they had suffered from a problem similar to the character in the vignette (depression) and 15.1% indicated that they had suffered from 'depression' in the last month.
Sample 2 (Local community)
This sample comprised 6,134 respondents who returned a screening questionnaire in a mailout to 27,000 individuals who were randomly selected from amongst registrants aged 50 years or less on the Canberra region electoral roll. Registration on the electoral roll is compulsory in Australia. Of the 6,134 respondents, 562 proved to be older than 50 years and were therefore excluded from the analyses reported here. Thus, the final sample size was 5,572. The Sample 2 survey did not involve a vignette. Stigma was measured using the Personal and Perceived Scales of the DSS . Level of psychological distress was evaluated with the Kessler 10 (K10)  on a scale of 0 to 40, and exposure to depression with a modified version of the Level of Contact Report  on a scale of 1 to 12. Respondents were asked if they had 'ever been markedly depressed' and to indicate their age, gender, educational level, level of Internet access, whether they were receiving treatment from a mental health professional and their willingness to participate in an Internet intervention study.
The average age of the sample was 35.9 years (SD = 9.2), 65.2% were women, 44.3% had completed a tertiary qualification (Bachelor's degree or higher). 62.3% of the sample reported a history of depression.
Sample 3 (Local community Distressed subset)
This sample comprised the subset of 487 Sample 1 respondents who were aged less than 50 years and who satisfied the eligibility criteria for participation in an Internet intervention trial. Criteria for inclusion were a K10 score of 12 or greater, access to the Internet, willingness to participate in the trial and not currently receiving treatment from a psychologist or a psychiatrist. In addition to responding to the survey questions administered to Sample 2, these respondents completed a second mail survey approximately 2 weeks after the first. The second survey comprised questions designed to evaluate the severity of participants' depressive symptoms (Center for Epidemiologic Studies Depression Scale (CES-D, ) and their level of dysfunctional thoughts using the Automatic Thoughts Questionnaire (ATQ) . They also completed a 22-item Depression Literacy scale (D-Lit, maximum score = 22 ). In addition, the survey included some items that are not the subject of the current paper including the Cognitive Behaviour Therapy Literacy scale (CBT-Lit, ) and questions relating to Internet interventions and stage of change. The average age of this sample was 35.3 years (SD = 8.76), 43.5% had completed a tertiary qualification (Bachelor's degree or higher), and 72.1% were women. Average CES-D score was 21.6 (SD = 10.73) and 70.2% of the sample met the CES-D criteria (>16) for current depression. 93.2% of the sample had a self-reported history of depression.
Psychometric characteristics of the stigma scales
In order to explore if the DSS subscales were valid, each of the datasets was subject to a principal component analysis (PCA) followed by varimax rotation with the aim of identifying a component structure for the DSS that was simple, reliable and interpretable. The estimated number of components in the scale was initially determined using both parallel analyses (95th percentile) and the Velicer's Minimum Average Partial Method (MAP, ) using a script developed by O'Connor . Kaiser-Meyer-Olkin measures of sampling adequacy and Barlett's tests of sphericity were conducted to ensure that the data were suitable for principal component analysis. Variables (items) were included if their component loadings were at least 0.32 (see Tabachnick and Fidell ). Consistent with recommendations by Velicer and Fava  and our aims, we opted to retain only well identified factors, rerunning the analyses and extracting fewer components where we identified less than three high loadings on a component (cutoff of 0.6).
Predictors of depression stigma
The above analyses suggested that the DSS Personal and DSS Perceived stigma scales were valid (see Results below). Therefore, the status of demographic and other variables as predictors of depression stigma were analysed separately for the DSS Personal stigma and DSS Perceived stigma scales for each sample (Samples 1 to 3) using a series of two-step hierarchical regression analyses with entry of demographic variables in the first step and other predictor variables in the second (see Tables 1 to 3). A hierarchical regression analysis was also conducted on the Social Distance scale data for Sample 1 using the same predictor variables as for the DSS scales. In the case of Sample 1, variables were entered as follows: age (12 category); sex (females = 0, males = 1); Education (Not tertiary = 0, Tertiary = 1); Country of birth (Australia = 0, Overseas = 1); Major city vs elsewhere (No = 0, Yes = 1); Awareness of beyondblue: the national depression initiative (No = 0; Yes = 1); History of depression (No = 0, Yes = 1); Family member with depression (No = 0, Yes = 1); Service provider (No = 0, Yes = 1); and Recognition of depression vignette (No = 0; Yes = 1). Survey 2 and 3 variables were entered as 'continuous' measures except for sex (females = 0, males = 1).
Potential interactions between predictor variables and stigma types were investigated using a series of Mixed Between-within subjects Analyses of Variance (Between variable = Predictor; Within Variable = Stigma Type). For the purposes of the ANOVAs, predictors were dichotomized to ensure comparability across surveys. Thus the 12-category age measure in the Sample 1 survey was collapsed into 4 categories (18 to 24 years; 25 to 49 years; 50 to 64 years; 65 years and above) and the continuous age measures employed in the survey for Samples 2 and 3 (age range 18 to 50 years) were dichotomized into (18 to 24 years; 25 to 50 years). Educational categories were categorised into Not tertiary = 0 or Tertiary = 1 for each survey. Psychological distress was categorized as either low to mild (K10 < 19) or moderate to severe (K10 = 20 to 40), Contact was dichotomised into high contact (score > 6) versus low contact (score < 7) and Depression Literacy was dichotomised into pass (>10) and fail (<11).