In 1999, a Task Force consisting of research and policy-oriented SLPs, people who stutter, and an epidemiologist (Ken St. Louis, Bobbie Lubker, Scott Yaruss, Jaan Pill, and Charles Diggs, respectively) convened to develop the first prototype of a questionnaire to measure attitudes toward stuttering.
Known as the Public Opinion Survey of Human Attributes (POSHA-E), it is handled like public opinion polls from a national polling organization, with respondents selected at random. Participants first answer baseline questions before filling out a demographic survey. Finally, they respond to follow up questions.
Here are examples of the most recent experimental version of the POSHA-E. It asks the respondents their overall impression of a person who is in one of nine categories: "left handed, mentally ill, obese, addicted to alcohol, has a stuttering disorder, is multilingual, has epilepsy, has HIV/AIDS, or uses a wheelchair." The respondents give their impression of the person, ranging from very negative, somewhat negative, neutral, somewhat positive, or very positive, to not sure.
Then the questionnaire asks the respondent, "I would want to be a person who" fits in one of those nine categories. Finally, the respondent answers follow-up questions on up to three of the listed categories. For example, "People who stutter (or are obese, or multilingual) are nervous or excitable; are dangerous to others; can raise a family." The respondents indicate whether they agree, disagree, or are not sure.
These survey questions have been asked of more than 1,200 adult respondents in 27 nonprobability (nonrandom) pilot study samples in 11 countries (Brazil, Bulgaria, Cameroon, Canada, Denmark, Nepal, Nicaragua, Macedonia, South Africa, Turkey, and the U.S.). Respondents completed questionnaires in either English or one of six other languages (Bulgarian, Macedonian, Portuguese, Turkish, French, and Spanish).
Since 1999, the team has undergone changes and the POSHA-E has been revised three times. One major change in the last version is that attributes of "old," "intelligent," and "good talker" were replaced by "epilepsy," "HIV/AIDS," and "alcohol addiction." Also, "overweight" was changed to "obese."
The research initiative using the results of the POSHA-E is known as the "International Project on Attitudes Toward Human Attributes" (IPATHA; also and previously known as the "International Project on Attitudes Toward Stuttering," IPATS).
Like most other measures of attitudes, the POSHA-E samples a variety of beliefs, reactions, behaviors, and emotions that would identify societal ignorance, stigma, and/or discrimination (e.g., Blood et al., 2003; Gabel et al., 2004; Hulit & Wertz, 1994; Klein & Hood, 2004). Like many instruments, it uses a paper-and-pencil format that asks for scaled responses that can compare one group of respondents to another. And like some attitude instruments, developers have paid considerable attention to validity, reliability, standardization, and user-friendliness. Once completed, the POSHA will be one of very few instruments that can be carried out with probability (e.g., random) sampling schemes consistent with solid epidemiological research.
The POSHA-E has a number of unique features. First, it places attitudes toward stuttering-or any of the other human attributes-within the context of a range of potential reactions ranging from very positive to very negative. Second, it seeks to reduce response bias by not stating specifically that stuttering (or any of the others) is the target attribute.
Third, it is designed to minimize cultural and linguistic bias, thereby enhancing translation to other languages and cross-cultural standardization. (For example, common words for some attributes carry negative connotations themselves, e.g., the word "stuttering" in Arabic often implies mental deficiency. In colloquial Spanish, there is no apparent difference between "overweight" and "obesity"; both translate to "fat." All items are worded directly and neutrally, minimizing idiom, to facilitate reading comprehension, reduce cultural bias, and permit more direct and accurate translations. Of course, simple translation errors can occur. In one case, "raise a family" was inadvertently translated as "to increase a family." To reduce these problems, IPATHA guidelines will mandate a "back translation" from a foreign language to English by another translator, unfamiliar with the study and the original translation, so that the eventual translated versions are as accurate as possible.)
Fourth, it is designed specifically to measure many of the responses that might be expected to change after public information campaigns designed to "improve attitudes" have been undertaken. For example, one series of questions asks where respondents have acquired their information. Even if attitudes may not have changed, it is likely that public awareness campaigns will result in different profiles of perceived knowledge acquisition. Fifth, consideration of the instrument's eventual recommended scheme for use must allow for probability sampling that is accurate but reasonably inexpensive for stakeholders.
Importantly, the POSHA-E is still in its developmental phase. It is not yet a user-friendly instrument, but we are making progress. For example, the response format has been progressively simplified in the three versions from an original 0-100 quasi-continuous scale, to a 1-9 equal-appearing interval scale, and finally to the above illustrated categorical "yes," "no," or "unsure" response mode (with a 1-5 scale for a few items).
This modification has progressively reduced respondent completion time and errors as well as data coding time and errors; both were achieved without appreciable loss in sensitivity to subtle differences in the attitudes of the nonprobability pilot samples studied. Also, as is the case with most standardized instruments, the POSHA-E contains many more items than the eventual version will. Forthcoming item analysis will identify those questions that are most discriminating and useful. After consideration of information available from pilot studies, redundant items will be eliminated. Also, confusing and ambiguous items will be eliminated or modified. The intent is to have a questionnaire that can be completed by most respondents in 10 minutes or less.
The next step is to carry out the difficult task of field-testing the near-final version of the POSHA-E with representative samples. Like political polling, this component is necessary to improve accurate representation of the attitudes of any target sampling area. A number of plans under evaluation, but one that may be the most cost efficient for stakeholders, could involve a public school-based sampling plan as follows. A probability (i.e., random) sampling scheme would be used to select public schools as the units from which random samples of children and, ultimately, progressively less random samples of parents, grandparents, and other adults would be generated as prospective respondents. All children in selected classrooms would be potential survey respondents. So, too, would one of their parents or guardians, one of their grandparents, and one of their adult neighbors. Probability sampling would determine school and classroom selection, but adults would be recruited by parents of children in these classrooms.
We hope that one outcome of the IPATHA initiative will be a data archive into which results from new studies may be added and against which results from isolated samples may be compared. Finally, in concert with other stakeholders around the world, we plan to develop strategies designed to reduce stigma and to measure their effectiveness.
Sample Pilot Results
In the process of gathering pilot data, a large amount of questionnaire data has been amassed from more than 1,200 adults. We have attempted not to be distracted from our primary purpose of instrument development. Accordingly, we have inspected the comparative data for trends showing consistent versus variable results that might suggest the best items for eventual inclusion and elimination. Following are some illustrations of what the pilot data have shown that is relevant to SLPs and to those interested in international comparisons (Knudsen et al., 2004; St. Louis et al., in press; St. Louis et al., 2004).
The table below provides selected demographic information from nine samples, ranging in size from 14 to 188 respondents. The first two (columns 2 and 3) compare probability samples of SLPs holding specialty recognition in fluency disorders with a sample of SLPs nationwide. It was thought that board-recognized specialists in fluency disorders might provide a "gold standard" for public attitudes toward stuttering. These are compared with a nonprobability sample of students from the mid-Atlantic region of the U.S. (column 3). All three of these samples responded to the most recent version of the POSHA-E.
The other nonprobability samples were from six different countries, with those in Denmark (column 5), South Africa (column 6), and Nepal (column 7) being administered in English and those in the remaining three countries administered in respondents' native language: Brazilian Portuguese (column 8), Bulgarian (column 9), and Turkish (column 10), respectively. These groups responded to either the first version (drawing a vertical line on a horizontal scale marked on either end and the middle, with scores later converted to numbers from 0 to 100) or the second version (circling a number from 1-9 or a "?" for "I don't know."). The Table illustrates that the percentage of females to males, age, educational level, religion, and marital status varied widely across the nine samples.
Preliminary results indicate that multilingual, as well as intelligent and good talkers in the earlier versions of the POSHA, are regarded as desirable characteristics. Left handed-as predicted-is neutral. Old (in earlier versions) is neutral to somewhat undesirable. The remaining attributes, wheelchair use, obesity or overweight, mental illness, alcohol addiction, epilepsy, HIV/AIDS, and stuttering, are all regarded as conditions that respondents would not want to have or be. The only significant exception was that specialists in fluency disorders were less likely to rate stuttering as something they would not want to have. Respondents in the six countries, responding in English as a second language or in their native language showed similarity overall to profiles of adults in the U.S., but showed some interesting variations. For example, stuttering was the lowest item scored for the Turkish respondents while Brazilian and Bulgarian adults both scored mental illness lowest.
Data collected also confirm the well-known public confusion about the cause of stuttering. The data from specialists and generalist SLPs reveal general agreement for a genetic causal component and less certainty about psychological or learning components. Among the lay public, only the Danes believed that psychological etiology was not the strongest causal component; neither did they believe stuttering is learned. By contrast, all the other lay groups rated psychological etiology the strongest. Among all groups except fluency specialists, a disturbing minority regarded stuttering as an act of God, especially in South Africa, Nepal, and Turkey. The fact that ghosts, demons, or spirits as causal agents for stuttering was not completely rejected-even by all SLPs in the U.S.-suggests that stigmatizing beliefs still occur.
By contrast, most to nearly all respondents believe that people who stutter can lead normal lives, though least so in Nepal. While most respondent groups indicated that people who stutter could communicate effectively, they were less optimistic about whether or not they should work in jobs requiring a great deal of talking. An unsubstantiated stereotype holds that people who stutter are nervous, shy, and fearful (e.g., Blood, 1999; Shapiro, 1999), and the data here confirm that only the fluency specialists and Danes soundly rejected the nervousness component.
As for stuttering being related to shyness or fear, these same two groups, along with SLPs, were least likely to make a connection. Most of the other groups were unsure, but the Brazilians and Bulgarians confirmed the stereotype. Like the ghosts-demons-spirits question, the reduced intelligence question was included to identify the likelihood of potentially strongly stigmatizing attitudes. Nepalese and Bulgarian results suggest attitudes still are present that associate stuttering with reduced intelligence.
The best news about what people report they would do or feel if they found themselves talking to someone who stutters is that most (and virtually all SLPs and specialists who responded) would wait patiently and ignore the stuttering. (In fact, the latter question was changed in the last revision to "not make an issue of" rather than "ignore" because it is plausible that some might not make an issue yet not ignore the condition.) Specialist and generalist SLPs (and Danes) would be unlikely to fill in words or give advice to "slow down" or "relax" while talking with a person who stutters. By contrast, Turks and to a lesser extent, Nepalese, likely might behave in these ways. None of the groups are likely to joke about stuttering, but the chances for joking might be highest in Turkey or Bulgaria.
Feeling comfortable or relaxed during a conversational partner's stuttering varied the most for the different groups, ranging from being very to somewhat likely for specialists and generalists, respectively, to being moderately unlikely for Nepalese, Brazilians, Bulgarians, and Turks. Curiosity was quite uniformly neutral; pity ranged from slightly above neutral to highly unlikely (for SLPs). Reported embarrassment, anger, or fear were unlikely.
Taken together, all of the data from the pilot studies analyzed to date show that groups of people around the world perceive stuttering almost as negatively as mental illness and obesity (overweight), but more negatively than wheelchair use. The studies also suggest that the POSHA identifies subtle differences among sample groups, such as urban versus rural/suburban, student versus non-student adults, African Americans versus Caucasian Americans, speech-language pathology specialists in stuttering versus speech-language pathology generalists, and samples from low- to middle-income nations versus higher-income nations.
Responses indicated that the POSHA-E can be translated into other languages with results on attitudes that are similar to those from the English versions. Finally, the pilot studies showed that adult respondents could complete the questionnaire with relative ease and efficiency.