Consistency Checks, Social Desirability Bias, and Built-in Validity Scales

Why the WSQ has validity scales

A self-report personality instrument only works if respondents answer honestly. If everyone could fake the 'perfect officer' profile, the instrument would measure nothing useful. To protect against this, the WSQ includes validity scales — built-in checks that run alongside the trait scoring and measure whether your response pattern is believable. If a validity scale flags your profile as inconsistent, extreme, or socially desirable beyond what real humans produce, the scoring engine can down-weight or invalidate your trait scores regardless of how 'good' they look.

This is the single most important thing to understand about the WSQ: your validity scores gate your trait scores. A candidate with a believable but modest trait profile can outscore a candidate with a 'perfect' trait profile that is flagged for faking.

Consistency checks and paired reverse-worded items

The most common validity check is the paired item. The WSQ measures each trait with multiple items, some of which are reverse-worded — they describe the opposite end of the trait. For example, a teamwork scale might include both:

• I enjoy working with others. (positively worded)
• I prefer to do things on my own without help. (reverse-worded)

Both items measure teamwork, but in opposite directions. If you mark Strongly Agree to the first item, the scoring engine expects you to mark Disagree or Strongly Disagree to the second. A candidate who marks Strongly Agree to both is responding inconsistently — they cannot simultaneously love working with others and prefer to do everything alone. The validity scale records this inconsistency.

A few inconsistencies across 135 items are normal — people are not perfectly logical, and some statements are genuinely context-dependent. But a high count of inconsistent paired responses signals one of two things:

1. Careless responding — you skimmed the statement and marked without reading. This is fixable by reading each item fully.
2. Faking without reading carefully — you tried to mark the 'good' answer without noticing that the item was reverse-worded. This is the more damaging pattern, because it looks like deliberate distortion.

The scoring engine cannot tell which of these produced the inconsistency, and it does not need to — either way, your profile is unreliable.

Extreme responding flags

Extreme responding is the tendency to mark only the endpoints of the scale (1 or 5) and rarely use the middle values (2, 3, 4). A candidate who marks Strongly Agree to 100 of 135 items and Strongly Disagree to the rest is extreme responding. The WSQ's extreme-responding scale counts the proportion of endpoint responses and flags profiles that exceed a threshold.

Extreme responding is a problem for two reasons:

• It is statistically distinguishable from honest responding. Real people use the middle of the scale; they Agree to some things, Strongly Agree to others, and Neutral to things that depend on context. A flat pattern of all-endpoint responses does not look like a real person.
• It often co-occurs with faking good. A candidate trying to look perfect tends to mark Strongly Agree to every positive statement and Strongly Disagree to every negative one, which is the extreme-responding signature.

The fix is to use the full 1-5 range honestly. If a statement is mostly but not perfectly true of you, mark Agree rather than Strongly Agree. If a statement is sometimes true, mark Neutral. A balanced use of the scale is more believable and scores better than an all-endpoint pattern.

Social desirability bias and the faking-good scale

Social desirability bias is the tendency to mark the answer that makes you look good rather than the answer that is true. On the WSQ, this shows up as overclaiming virtue — marking Strongly Agree to every integrity, conscientiousness, and stress-tolerance item, and Strongly Disagree to every item that admits a flaw.

The WSQ detects this with a social desirability (faking-good) scale built from items that describe behaviors almost no one consistently does or does not do. Classic examples (paraphrased, not real items):

• I have never told a lie. — Almost no one has never told a lie. Strongly Agree here is an overclaim.
• I have never been late for anything. — Almost no one has never been late. Strongly Agree is an overclaim.
• I have never lost my temper. — Almost no one has never lost their temper. Strongly Agree is an overclaim.

A candidate who marks Strongly Agree to a high proportion of these never-statements is flagged for faking good. The scoring engine then down-weights the trait scores, because if you are overclaiming on the validity items you are probably overclaiming on the trait items too.

This is why the strategy in Section 4.4 emphasizes honest responding over trying to guess the 'right' answer. The WSQ is designed to reward honesty and detect faking. A believable profile with a few admitted imperfections scores better than a flawless profile that is flagged for social desirability.

How validity flags interact with trait scores

A validity flag does not automatically zero out your WSQ score. The scoring model is more nuanced:

• A mild inconsistency or extreme-responding flag may slightly down-weight your trait scores without invalidating them.
• A moderate social-desirability flag may move your profile toward the average (regression to the mean), pulling extreme scores back toward the middle.
• A severe flag — for example, a pattern that looks like random responding or blatant faking — can invalidate the WSQ score entirely, so your WSQ contributes nothing or is referred for follow-up.

The practical implication: protecting your validity scores is as important as protecting your trait scores. A candidate who answers honestly and consistently, uses the full 1-5 range, and admits minor imperfections will have clean validity scores and a believable trait profile. A candidate who tries to game the instrument will have impressive-looking trait scores and a validity flag that tells the scoring engine to ignore them.