TEST_3_18_Practical Applications: Response Biases

Motivations of Malingering:

Criminal proceedings, disability evaluations, personal injury claims ($$$) Drug-seeking behavior in clinical settings

Self-deception

test takers intentionally underreport negative aspects of themselves, or have unrealistically positive views of themselves. Trait-like

Prevalence of malingering (Berry et al., 2002)

~7.3% to 27% in general psychological evaluations ~ 31% to 45% in forensic evaluations

Measures of Individual Differences in Social Desirability

Marlowe-Crowne Social Desirability Scale - "Uncommon virtues" Before voting I thoroughly investigate the qualifications of candidates - "Common Flaws" On occasion I have had doubts about my ability to succeed in life - Dichotomous response options (Yes/No) Brief Inventory of Desirable Responding (Paulhus, 2002) - Self-Deceptive Enhancement Scale (20-Items) - Impression Management Scale (20-Items) - "I am a completely rational person" - Scored on a 7-point Likert-type scale

"Fake-good"

Amount people lie on personality test to look better Between persons ~.66 SD across all of the Big Five Between persons ~1.0 SD across social desirability scales Within persons ~.5-.9 SD across all of the Big Five Within persons ~2.26 SD across social desirability scales

Strategies of Managing the testing CONTEXT

Anonymity Minimize frustration Validity warnings

Validity scales

Evidence as to the efficacy of validity scales depends on the scale and the source of the scale Evidence suggests can differentiate between malingerers and honest respondents when the groups are known There is no response bias "silver-bullet" (nothing will fix this permanently)

Malingering

Faking-bad Tendency to exaggerate problems Respondents are motivated to appear more cognitively impaired, emotionally distressed, physically challenged, or psychologically disturbed than they actually are.

Anonymity

May increase accurate responding May also increase random responding

Neutral items

Neutral on social desirability E.g., "I am sometimes less friendly than other people"

Forced choice assessment

Pairs or sets of items that are equally socially desirable

Response bias

Patterns of responding to scale items that result in false or misleading information -The test talker themselves is introducing a source of error.

Individual differences in acquiescence

People tend to respond and inflate and deflate their scores in different ways. -They may want to be liked so they answer in the most desirable way. -Affects the interpretability of test scores

Extreme/Moderate responding

Tendency to avoid or endorse extreme response options Individual differences in willingness to endorse extreme

Sources of Social Desirability Bias

Test content itself Testing context Test-takers themselves: Individual differences in the extent to which people engage in socially desirable responding

Impression management

Test takers intentionally attempt to appear socially desirable State-like

Guessing

(Correct/Incorrect items) Some respondents may be "luckier" than others and answer items correctly

Random Responding

(Likert-type items) Carelessness or lack of motivation to respond meaningfully "Christmas-treeing"

Types of response bias

-Acquiescence bias -Extreme/Moderate responding -Social desirability bias -Malingering -Random Responding -Guessing

May be multiple forms of social desirability:

-Impression management -Self-deception

Strategies for Detecting Response Biases

-Validity scales E.g., including assessments of minor flaws and weaknesses most people are not willing to admit -Intervening Exclude data from analysis Retain with caution Retain and evaluate statistically (sensitivity analysis)

Validity Warnings

-Warn participants inaccurate responses can be detected -Effective against malingering -Works if respondents believe they can be detected

2 sources of Acquiescence bias

-Yea-saying" Agreeing to all items regardless of content -"Nay-saying" Disagreeing to all items regardless of content

Managing the Effects of Response Bias

Balanced scales: items are positively and negatively keyed (or reverse scored) E.g. test of agreeableness I get along well with others I don't like to spend time with others (R) I rarely rock the boat I am a good friend, not a nice friend (R) Must be scored appropriately -Cannot differentiate acquiescent responders from those who have moderate trait levels (trading a more severe problem for another, less severe one) Do not prevent acquiescence or its effects Implications for validity

Straightforward items

E.g., "I am honest" vs. "I try to be as honest with my friends, family, co-workers, and strangers as much as I can be"

Item keying

If all items are keyed in the same direction, the items are more susceptible to acquiescence -This is why test makers make reverse coded items.

Strategies of Managing the testing CONTENT

Straightforward items Neutral items Forced choice assessment

Social desirability bias

The tendency for a person to respond in a way that seems socially appealing, regardless of his or her true characteristics. -Example: employment personality assessment

Acquiescence bias

When an individual agrees (or disagrees) with a statement without regard for the meaning of those statements -Its problematic when the items are reverse scored. The VALUES are messed up *Item phrasing is important here (Dominance modeling: when you have the most extreme of the question. Ex: answer "I Super Love ..." every time instead of "I like")

Minimize Frustration

When test-takers are fatigued or distracted they are more likely to randomly respond

In Extreme/Moderate responding the problem may be ___________________.

overstated -"Although an effect was found for item wording, the magnitude of nonequivalence was small and is unlikely to have a strong influence on scale-level measurement outcomes." (Nye et al., 2010, p. 806) -Basically were nitpicking. The errors will vance each other out and aren't as big of a problem as we think they are. Basically, Effects at the item level exist Effects are small Unlikely to influence scale score

What are personality tests ver vulnerable to?

socially desirable responding