Lead section[edit | edit source]
The Attribution Questionnaire (AQ) is a self-report questionnaire that assesses public stigma toward people with mental illnesses, specifically one's sense of attribution, emotional reaction, and discriminatory responses. The AQ begins with a paragraph-long hypothetical vignette about Harry, a person with schizophrenia. There are different versions of the Harry vignette that can be used to test different forms of attribution. The measure follows with 27 items rated from 1 (not at all) to 9 (very much) assessing stigma towards Harry, creating 9 subscales that capture how one perceives and how one would respond to a specific person with mental illness. Although the questionnaire itself is not specifically meant for one age group, the items inquire about complex emotions and hypothetical actions and thus may be most appropriate for older adolescents and adults. The AQ was developed in 2003 by Dr. Patrick Corrigan and colleagues based in Illinois and has been further developed into smaller tests such as the Attribution Questionnaire-9 (AQ-9), the revised Attribution Questionnaire (r-AQ), and the children's Attribution Questionnaire (AQ-8-C) for ages 10-18, in which youths are presented both a simplified version of the vignette and simpler test items.
|Click here for instructions for lead section|
The lead section gives a quick summary of what the assessment is. Here are some pointers (please do not use bullet points when writing article):
Psychometrics[edit | edit source]
Steps for evaluating reliability and validity[edit | edit source]
|Click here for instructions|
Reliability[edit | edit source]
|Click here for instrument reliability table|
Reliability[edit | edit source]
Reliability refers to whether the scores are reproducible. Unless otherwise specified, the reliability scores and values come from studies done with a United States population sample. Here is the rubric for evaluating the reliability of scores on a measure for the purpose of evidence based assessment.
Validity[edit | edit source]
|Click here for instrument validity table|
Validity[edit | edit source]
Validity describes the evidence that an assessment tool measures what it was supposed to measure. There are many different ways of checking validity. For screening measures, diagnostic accuracy and discriminative validity are probably the most useful ways of looking at validity. Unless otherwise specified, the validity scores and values come from studies done with a United States population sample. Here is a rubric for describing validity of test scores in the context of evidence-based assessment.
Development and history[edit | edit source]
People may exhibit public stigma towards others with mental health issues through stereotypes, prejudice, and discrimination. Despite psychological research and statistical trends that suggest such prejudices may be invalid, public surveys have suggested people still possess ingrained beliefs about people with mental illness. The attribution theory posits that someone's behavior, such as a mental illness, may be perceived by others as a personal characteristic or something that they brought on themselves. Inferences about a mental illness lead to inferences about a person, and this attribution can lead to emotional reactions such as anger and pity towards that individual. Negative thoughts about people with mental illness may also influence discriminatory behavior such as refusal to help. Research supporting the attribution theory shows that internal attributions lead to more negative emotions and discriminatory responses. The original study by Corrigan, Markowitz, Watson, Rowan, and Kubiak validating the AQ explored the composition of public stigma: attribution, emotional reactions, and discriminatory thoughts. Their research thus revealed areas of interest where public stigma can be addressed and reduced.
In the original study in 2003, tested using a sample of students in community college, the researchers created the Attribution Questionnaire (AQ) that references prior theory on attribution, resulting emotional responses, and resulting discriminatory behavior. The original study manipulated the hypothetical vignette's level of dangerousness, specifying the man's non-violent nature to one group and a violent nature to another group. The questionnaire contained 21 items grouped into 6 subscales that theoretically aligned with prior research: personal responsibility, pity, anger, fear, helping/avoiding behavior, and coercion/segregation. By the following year, Corrigan used a different version of the AQ, now containing 27 items and 9 subscales. Dangerousness was included as a new subscale instead of being manipulated, and coercion, segregation, avoidance, and helping behavior were split into four separate subscales. The AQ-27 is the current version of the questionnaire.
The subscales of the AQ-27 and descriptions are listed below:
- Blame: perception that a person's mental illness is controllable and their own responsibility
- Anger: towards the person with mental illness
- Pity: towards the person with mental illness
- Help: extent to which people are willing to support people with mental illness through interpersonal or economic means
- Dangerousness: perception of people with mental illnesses as a threat to themselves, to others, or to the society
- Fear: towards the person with mental illness
- Avoidance: active evasion of people with mental illness
- Segregation: support for treating people with mental illnesses in isolation away from the community, usually in institutions
- Coercion: support for mandatory treatment for people with mental illnesses
Versions[edit | edit source]
The AQ-9[edit | edit source]
A shorter version of the AQ-27, called the AQ-9, was created by Corrigan and colleagues in 2003. This version contains 9 items, each representing one of the 9 subscales existing in the longer version. These items were chosen for the shorter form for their strong factor loadings from the AQ-27.
The r-AQ[edit | edit source]
The revised Attribution Questionnaire (r-AQ) was created by Watson and colleagues using the AQ-9 as a basis and adjusting the measure for an adolescent audience. This version contains the same number of items as the AQ-9, but the items have been verbally adjusted along with the subject of the vignette to describe a younger person with mental illness.
Impact[edit | edit source]
As of now, there is not a widespread use of the Attribution Questionnaire. Current studies analyzing its use are limited, and a gold standard on combating mental health stigma has yet to be established. As stigma becomes increasingly researched, the AQ may be a springboard to reach more audiences and clarify which factors make up stigma. The AQ could be applied to a clinical setting, but such implementation has not been planned or brought into practice. Moreover, validity studies need to be conducted to analyze how the child and adolescent versions of the AQ may map onto the adult AQ-27 and if this measure could be used longitudinally.
Use in other populations[edit | edit source]
The Attribution Questionnaire is not a commonly used measure and thus has not been translated into various languages. Currently, an Italian version of the measure exists.
Limitations[edit | edit source]
As the AQ is presented as a self-report, the pressure of social desirability may encourage subjects to endorse less of the items they think would be discriminatory towards those with mental illness. Furthermore, as the AQ is not normally built into a behavioral observation, the scores on this self-report questionnaire may not align with actual behavior. The AQ has only been validated in college volunteer students and adolescent students in research settings. More research is needed with the general public to see if the AQ could be used to assess stigma in other samples. Lastly, the components of mental stigma have yet to be fully revealed. Other components may not be well represented; others may be undeservedly so. The various existing studies have suggested the latter, with fewer factors being endorsed than the original 9.
OToPS Usage History[edit | edit source]
(when was measure added to OTOPS Survey?
(when was measure dropped from OTOPS survey?)
|Qualtrics scoring||Variable name of internally scored variable:
Notes on internal scoring:
- Is it piped?
- Is it POMP-ed?
- Any transformations needed to make it comparable to published benchmarks?
|Content expert||Name: Jane Doe, Ph.D.
Institution/Country: University of Wikiversity / Canada
Email: Type email out
Following page: Y/N
Scoring instructions and syntax[edit | edit source]
We routinely include scoring syntax in three major languages: R, SPSS, and SAS. All variable names are the same across all three (so we follow naming conventions that would work in any of them -- note that R is case sensitive), and all match the CSV shell that we provide as well as the Qualtrics export.
Hand scoring and general instructions[edit | edit source]
<Information about hand scoring and general instructions go here>
CSV shell[edit | edit source]
Because we are sharing the scoring syntax, we also want to share a shell data file that you could use in your own research. If you use this file, or make sure that your own file uses the same variable names, then the scoring code we provide should do the basic scoring for you. When different research projects and groups use the same variable names and syntax, it makes it easier to share the data and work together on integrative data analyses or "mega" analyses (which are different and better than meta-analysis in that they are combining the raw data, versus working with summary descriptive statistics).
Note that our CSV includes several demographic variables, which follow current conventions in most developmental and clinical psychology journals. You may want to modify them, depending on where you are working. Also pay attention to the possibility of "deductive identification" -- if we ask personal information in enough detail, then it may be possible to figure out the identity of a participant based on a combination of variables.
<Paste Link to CSV Shell here>
Syntax/code[edit | edit source]
|Click here for R code|
R code goes here
|Click here for SPSS code|
* Stigma scoring. * Done freehand with Di-Poo; no article in front of us. * Let's see how close we get to the real scoring! ;-). desc /var aq01 to aq27. compute aq07r = 10-aq07. compute aq08r = 10-aq08. compute aq26r = 10-aq26. compute aq16r = 10-aq16. compute aq20r = 10-aq20. compute aq21r = 10-aq21. * Note that Corrigan does *not* reverse these -- they form the Pity scale. compute aq09r = 10-aq09. compute aq22r = 10-aq22. compute aq27r = 10-aq27. reliability /var aq01 to aq06 aq07r aq08r aq09 aq10 to aq15 aq16r aq17 to aq19 aq20r aq21r aq22 aq23 to aq25 aq26r aq27 /sum total. * Not positive about scoring of #11; #5 shows craptacular corrected item-total. * Weird. Not reversing the Pity items makes the alpha worse, and low CITC (-.28!). compute aqtotaldi = (mean.24(aq01 to aq06, aq07r, aq08r, aq09r, aq10 to aq15, aq16r, aq17 to aq19, aq20r to aq22r, aq23 to aq25, aq26r, aq27r)-1)/8. compute aqtotal = (mean.24(aq01 to aq06, aq07r, aq08r, aq09, aq10 to aq15, aq16r, aq17 to aq19, aq20r, aq21r, aq22 to aq25, aq26r, aq27)-1)/8. compute aqblame = (mean(aq10, aq11, aq23)-1)/8. compute aqanger = (mean(aq01, aq04, aq12)-1)/8. compute aqpity = (mean(aq09, aq22, aq27)-1)/8. compute aqhelp = (mean(aq08r, aq20r, aq21r)-1)/8. compute aqdanger = (mean(aq02, aq13, aq18)-1)/8. compute aqfear= (mean(aq03, aq19, aq24)-1)/8. compute aqavoidance = (mean(aq07r, aq16r, aq26r)-1)/8. compute aqsegregation = (mean(aq06, aq15, aq17)-1)/8. compute aqcoercion = (mean(aq05, aq14, aq25)-1)/8. var labels aqtotaldi 'Diane scoring of AQ total (reverses Pity items, along with the 6 Corrigan flips)' /aqtotal 'Total Stigma -- Attitudes Questionnaire'. corr /var aqblame aqanger aqpity aqhelp aqdanger aqfear aqavoidance aqsegregation aqcoercion. * Blame = AQ10+ AQ11 +AQ23 * Anger = AQ1 + AQ4 + AQ12 * Pity = AQ9 + AQ22 + AQ27 * Help = AQ8 + AQ20 + AQ21 (Reverse score all three questions) * Dangerousness = AQ2 + AQ13 + AQ18 * Fear = AQ3 + AQ19 + AQ24 * Avoidance = AQ7 + AQ16 + AQ26 (Reverse score all three questions) * Segregation = AQ6 + AQ15 + AQ17 * Coercion = AQ5 + AQ14 + AQ25.
|Click here for SAS code|
SAS code goes here