What Evidence Supports Comparing WPI2-R and WPI2 Scores?

Short answer: what evidence supports comparing WPI2-R and WPI2?

Yes, the publisher supports comparing the Work Personality Index 2.0 Revised (WPI2-R) with WPI2 in a standardized reporting frame. The *Work Personality Index 2.0 Revised Technical Manual* makes the stronger claim directly: using the original normative group to standardize WPI2-R scores into Sten scores, it says, creates equality between scores and allows them to be compared. That statement should be read as the manual’s intended interpretation, not reduced to a claim that the editions merely share themes.

The practical limit is narrower: the statement does not itself quantify how far apart two particular people’s paired scores may fall, how precise an individual cross-edition difference is, or what size of difference would count as meaningful change. So the manual gives affirmative grounds to compare standardized report scores as comparable metrics. If the question is whether a person changed, or by how much, the reviewed evidence does not settle that further inference. Treat a cross-edition difference as something to investigate, not as a quantified personal change on its own. That distinction preserves the publisher’s stated comparison while avoiding a conclusion about change that requires more than matching score labels.

Sources: Equating Test Scores (without IRT), Second Edition

What changed between the two editions?

WPI2-R is a shortened revision of WPI2, not an identical second printing. The *Work Personality Index 2.0 Revised Technical Manual* describes reducing the earlier assessment from 198 items to 113. It presents this as statistical item reduction: the revised form keeps selected material from the earlier instrument while reducing the number of questions a respondent completes. The 85-item difference matters when reading the forms: a total formed by adding responses across 198 questions cannot be treated as the same kind of raw total as one formed across 113 questions. A like-for-like comparison therefore needs to follow the reported scales and score interpretation, rather than subtracting or equating the two raw totals as if the forms had the same length.

The manual says WPI2-R introduced no new items or scales; its items and scales were drawn from WPI2. A scale is a reported dimension made up of assessment items. When the revised manual refers to a counterpart scale, the term means the corresponding scale in the other edition—the scale intended to represent the same named dimension across forms. This retained structure provides a basis for lining up corresponding report sections. It does not mean every question, item contribution, or response path stayed the same: selecting a shorter set necessarily changes which item responses contribute to the revised form. Readers should match documented counterpart labels, not assume that an old and revised item-by-item profile is identical.

The publisher also says it applied the revised 113-item model to WPI2’s original normative group of 8,360 people. A norm group is the reference population against which a standardized score is interpreted; here, “original” identifies the group used for WPI2, not a new population recruited solely for the revision. In the manual’s account, retaining this reference was intended to preserve consistency between editions. It gives the revised scores a shared reference context with the earlier assessment, while the item set itself has changed.

These design choices establish continuity in the publisher’s scale framework and reference group, alongside a shorter form whose item composition differs. They explain why edition labels and corresponding scales matter when placing two reports side by side. They do not, by themselves, demonstrate that a particular respondent would receive the same result on both forms: using the same reference group is not the same thing as having identical questions or observed responses. The manual’s claim about standardized-score comparability is a separate issue; the relevant distinction here is that the reports are related through retained scales and norms, but their raw item totals are not interchangeable simply because the scales share names.

Sources: Work Personality Index 2.0 Revised Technical Manual

What do the manual's cross-version correlations establish?

The *Work Personality Index 2.0 Revised Technical Manual* reports two high-correlation summaries, but they describe different analytic contexts. Appendix D gives matched-scale correlations from 0.87 to 0.98, with a mean of 0.92, for the manual’s initial research sample of 113,755 individuals using the 113-item revised raw-score model. Those figures show that, across the reported scales, higher revised scores tended to accompany higher scores on their WPI2 counterparts in that sample. The breadth and consistency of the coefficients are meaningful evidence that the shortened form retained substantial scale-level continuity; they are not a trivial result simply because they leave other questions open.

Appendix H reports a second set, ranging from 0.86 to 0.99, after the manual applies the revised model to the original WPI2 normative group of 8,360. The appendix identifies Persuasion at the low end and Persistence and Social Confidence at the high end. This analysis puts the revised model and its counterparts in the original reference-group context described by the publisher. It reinforces the continuity picture across the listed scales, while the range also shows that correspondence was not identical for every scale. Appendix H should not be described as another correlation calculation on the 113,755-person research sample: the manual presents the original norm group and revised-model application as its context.

The distinction between these contexts matters when weighing how far the tables travel. Appendix D’s sample size makes its summary broad in number, but the manual’s table does not, in the reviewed material, explain whether the same individuals completed both editions, whether the forms were administered under comparable conditions, or how the sample was assembled for this particular cross-version analysis. Appendix H answers a different descriptive question by locating the revised model against the original norm group. Because the appendices do not provide matching uncertainty estimates or a direct table of paired score gaps, their ranges should not be treated as a before-and-after experiment or as a confidence interval for any one respondent. The figures tell readers about the relationships the manual reports, not the unreported mechanics behind each person’s pair.

Read together, the appendices support a practical conclusion about the instrument versions: corresponding WPI2-R and WPI2 scales have strong reported associations in both the large research-sample summary and the original-norm-group analysis. That is affirmative evidence for continuity and helps explain why the publisher presents the editions as comparable. It does not show the exact score gap expected for a particular respondent or establish that every person would receive nearly the same paired result. The coefficients are presented at table level; in the reviewed sections, the manual does not supply a clear paired administration design, timing or conditions, confidence intervals, a distribution of individual differences, agreement limits, or a threshold for interpreting change. Without those details, readers can use these results as evidence of group-level correspondence, but cannot infer the precision or meaning of one individual’s cross-version difference from the ranges alone.

Sources: Work Personality Index 2.0 Revised Technical Manual

What does the manual mean when it says the Sten scores are equal and comparable?

The *Work Personality Index 2.0 Revised Technical Manual* says that standardizing WPI2-R scores into Sten scores using the original normative group creates equality between scores and permits them to be compared. That wording deserves its full force: the publisher is explicitly presenting the revised Sten scores as comparable with the earlier edition’s scores, rather than merely saying that the forms cover related dimensions. In the manual’s intended reporting frame, a reader may place standardized scores for corresponding scales side by side and interpret them as scores expressed on a common, norm-referenced metric. The correlations reported in Appendices D and H support that reading by showing strong relationships between counterpart scales. Together, the stated interpretation, shared reference context, and correlations provide real affirmative support for comparing the report metrics.

A Sten score is a standardized score expressed on a ten-point scale; its meaning depends on the norm reference used to locate a result relative to a comparison group. Here, the manual says it applied the revised 113-item model to the original WPI2 normative group of 8,360 people, and links the Sten comparison to that group. This gives the publisher’s equality claim a specific frame: corresponding standardized scores are intended to be read using that common norm reference. It is therefore reasonable to compare what the two reports show in that defined sense—for example, whether the same counterpart scale is relatively higher or lower in each report—while keeping the edition and norm information visible. The fact that WPI2-R has fewer items does not erase this stated standardized-score interpretation; the common metric is precisely what the manual offers for score comparison.

This distinction also prevents two opposite readings. It would be too dismissive to say that the editions cannot be compared merely because the manual does not publish every possible analysis: the publisher makes a direct comparability claim and supplies a shared standardized reference, with counterpart associations as supporting evidence. But it would go beyond that claim to treat the same Sten number as proof that a person’s underlying work pattern remained fixed, or to treat a numerical gap as a measured amount of personal change. The manual’s equality language establishes the intended relationship among report scores; it does not, in the reviewed passages, define individual change precision or a decision threshold. In practice, the reader can compare the standardized metrics as the manual intends and separately reserve judgment about what an individual difference means.

The remaining question is more specific than whether the scores are comparable: how much confidence should a reader place in a particular person’s difference between editions, and what size of gap would justify calling it meaningful change? A common standardized metric alone does not quantify the uncertainty of a paired difference. The manual’s reviewed sections do not report person-level difference precision, agreement intervals, or a threshold that separates ordinary variation from meaningful change across these forms. Those omissions do not negate the publisher’s claim that the standardized report metrics can be compared. They mark the boundary between comparing scores in the manual’s intended reporting frame and interpreting a specific gap as an accurate measure of how much one individual changed. For that latter conclusion, the report statement and scale correlations are not enough by themselves; evidence about paired individual differences would be needed.

Sources: Work Personality Index 2.0 Revised Technical Manual

Does using a common norm group amount to formal score equating?

Not by itself. Using the same norm group can put two editions into a shared interpretive frame, while formal score equating describes a more specific operation: a documented mapping between scores from forms. In *Equating Test Scores (without IRT), Second Edition*, ETS explains equating as converting scores on a new form to corresponding scores on a reference form. That distinction helps name what the Work Personality Index 2.0 Revised Technical Manual documents and what the reviewed account leaves unspecified; it does not make equating the only legitimate basis for comparing reports.

The manual says the revised 113-item model was applied to WPI2’s original normative group of 8,360 and that standardizing WPI2-R results as Sten scores creates equality between scores and allows comparison. That is meaningful: readers can understand the publisher as offering a common norm-referenced reporting context for corresponding scales, rather than merely asserting that the editions have related content. A common standardization frame can make the reported values comparable in the manual’s intended sense. The shared norm group also supplies the comparison population against which those standardized values are interpreted.

ETS’s description points to a different level of method detail. A conversion account typically identifies which score on one form maps to which score on another, and whether the mapping concerns raw scores or scores on a reporting scale. The reviewed WPI2-R manual passage states the common norm reference and its intended consequence, but does not present that statement as a score-by-score conversion table or spell out a separate linking design. That observation is about what the cited material describes; it does not establish that no other technical work exists or that the publisher’s standardized comparison is invalid.

The distinction matters because ‘same reference’ and ‘formal conversion’ answer related but nonidentical questions. A common reference helps readers interpret the score locations in comparable population terms. A conversion function, when documented, tells how possible scores on one form are related to scores on another under a specified method. Neither phrase alone determines how close an individual’s paired results will be, or what a particular gap means. Here, the manual’s common Sten frame and explicit comparability claim deserve to be reported fully; the available passage simply does not expose the mechanics of a formal score mapping.

So the practical judgment is bounded: the common norm group provides a useful basis for the publisher’s stated standardized-score comparison, and it should not be dismissed because a raw-score crosswalk is not shown in the reviewed sections. At the same time, a reader asking whether an individual WPI2 result can be converted exactly into that person’s WPI2-R result is asking for more specific evidence than a shared norm reference alone supplies. ETS’s general methods text clarifies the distinction; it does not prescribe what WPI must do or negate its stated reporting frame.

Sources: Equating Test Scores (without IRT), Second Edition

Why can strong correlation coexist with person-level differences?

Correlation describes how two sets of scores vary together: when one score is higher, does its counterpart tend to be higher too? Agreement asks a different question: are paired values close enough to one another for the intended interpretation? A strong association can support continuity in ordering while leaving the size and spread of individual gaps unresolved. In its abstract, *A note on the use of the intraclass correlation coefficient in the evaluation of agreement between two methods of measurement*, Bland and Altman explain that correlation or a test of average difference alone does not establish interchangeability; paired differences are relevant to that judgment. Their paper is a general methods argument, not a WPI analysis.

This distinction makes the WPI2-R manual’s high counterpart correlations useful but limited evidence. As the earlier section explains, the reported coefficients support the claim that corresponding scores tend to move together across the analyses described. They do not directly show whether each pair of scores is numerically close, how individual gaps are distributed, or whether any gap falls within a tolerance suited to a particular use. Preserving rank order and matching individual values are separate properties: a form can retain which cases score relatively higher while shifting some paired values by differing amounts.

Arithmetic illustration only, with invented numbers and no connection to WPI data: imagine five people with scores on one scale of 10, 20, 30, 40, and 50. A second set might be 14, 24, 34, 44, and 54. The ordering is exactly the same and every second score is four points higher. Another hypothetical second set—12, 25, 31, 47, and 53—also preserves the ordering, although the paired gaps now vary. In both examples, knowing that the association is strong would not, by itself, tell a reader whether the paired gaps are acceptably small for a specific purpose. No empirical quantity is being estimated here; the numbers only separate ordering from closeness.

For a person comparing two reports, the relevant object is the pair: the earlier value, the revised value, their difference, and how such differences behave across people under relevant conditions. A group-level correlation summarizes co-variation across many pairs; it does not provide the distribution of those differences or a boundary for interpreting one pair. Bland and Altman’s general measurement-method point is to examine paired differences rather than infer interchangeability from association alone. The abstract does not specify a WPI threshold or establish that one particular statistic must be used for every assessment purpose.

High correlations are therefore not empty evidence. They indicate that counterpart scores vary together and strengthen a continuity argument, especially alongside the manual’s retained scales and common reporting frame. Their limit is the question they answer: association is evidence about co-movement, while an individual interpretation needs evidence about paired closeness and the amount of difference that matters for the intended decision. A reader can accept the continuity signal without treating it as a personal change estimate.

Sources: A note on the use of the intraclass correlation coefficient in the evaluation of agreement between two methods of measurement

What evidence would make a score difference interpretable?

To decide whether a gap between WPI2 and WPI2-R reflects more than uncertainty, an administrator would need evidence about the precision of that gap, not only evidence that each scale is internally consistent. Measurement error is the uncertainty around a score that arises because a test is an imperfect measure; reliable change is a claim that an observed difference is large enough, under a specified method and purpose, to be distinguished from that uncertainty. The *Standards for Educational and Psychological Testing* treats precision and measurement error as relevant to score interpretation, while treating validity as support for a particular interpretation and use. These are related but separate questions: a consistent scale does not automatically yield a precise difference between two editions.

For this comparison, the useful technical request is for version-specific precision information and an explanation of how uncertainty in a paired WPI2/WPI2-R difference was estimated. The *Standards for Educational and Psychological Testing* provides general guidance, not a WPI formula or threshold. An administrator should ask whether the manual or provider has data on paired differences or agreement, how the uncertainty was calculated, and what interpretation the evidence supports. A statistic about each form's score reliability may help describe each score separately, but it cannot be silently substituted for the precision of subtracting one edition's score from the other's. The difference is its own quantity, potentially affected by error in both scores and by their relationship.

Interpretation also depends on who supplied the evidence and under what conditions. Documentation should identify the population studied, whether participants completed both forms, the interval between administrations, and relevant administration or scoring conditions. Those details help an administrator judge whether the evidence resembles the person and circumstances being interpreted. If the samples, timing, or procedures differ, the reported uncertainty may not transfer cleanly to a particular pair. This is a request for context needed to evaluate the inference, not an assertion that one design or statistic is universally required.

The reviewed *Work Personality Index 2.0 Revised Technical Manual* reports internal consistency and other psychometric analyses for WPI2-R, and its cross-version tables report counterpart correlations. Those findings matter: they inform how the revised scales function and support continuity across the forms. But the reviewed material does not report a person-level difference-score precision estimate, agreement interval, or a rule for deciding when a cross-edition gap is meaningful change. It would therefore be inaccurate to say that the manual establishes such a rule—or that the absence of one in the reviewed sections proves no such evidence exists elsewhere.

Finally, the administrator should ask what conclusion the available evidence permits: comparison of standardized report scores, a description of a recurring pattern, or a claim of personal change. A change interpretation needs a supported decision rule tied to the evidence and purpose; there is no single threshold that can be declared mandatory without knowing the use and method. Until relevant paired precision and interpretation guidance are available, the responsible description is that the reports differ by a stated amount, while the meaning of that gap remains uncertain.

Sources: Standards for Educational and Psychological Testing

What can the comparison support at work?

For private reflection or coaching, a difference between corresponding WPI reports can be treated as a prompt to ask what was happening in the person's work life, how they approached a specific task, or whether the report's description fits a recent example. If a similar counterpart pattern appears in both reports, it may help frame a question about observed behavior. That is a reversible inquiry: the person can check it against examples, consider other explanations, and revise the interpretation. It is not evidence that a trait caused an outcome, nor does a cross-version comparison predict which job someone should take.

The *Standards for Educational and Psychological Testing* frames validity around the interpretation and use being proposed. The *Principles for the Validation and Use of Personnel Selection Procedures* likewise addresses evidence for selection use. Together, these sources support a boundary: evidence that two editions have related scales or comparable standardized reporting does not, by itself, validate using their scores to select, promote, set compensation, or rate performance. Each of those decisions makes a different consequential inference about a person. Evidence relevant to that exact use and the conditions in which the decision is made would be needed before a score could carry that role.

This remains true even though WPI is a work-focused assessment and its publisher presents the editions as comparable. Occupational framing can make a report relevant to a conversation about work; it does not show that cross-version differences predict performance or identify a suitable role. The WPI2-R manual's stated Sten-score comparison and its counterpart correlations support reading the standardized reports in the publisher's stated common frame. They do not establish that a particular gap is a valid hiring score, or that a person's apparent movement between reports should change an opportunity decision.

If a score comparison could affect another person's prospects, the administrator should first identify the precise claim the decision would rely on and seek evidence supporting that interpretation for that purpose, along with a clear account of uncertainty. The general professional guidance cited here does not determine what is appropriate in every workplace or jurisdiction, and it does not validate WPI for any specific employment decision. Where that purpose-specific support is not established, do not turn a report difference into a verdict about employability or potential. Keep the comparison descriptive and seek other relevant evidence through the applicable decision process.

In a low-stakes conversation, the practical use is narrower: choose one concrete work situation and ask whether the report suggests a question worth checking against what the person actually did. A score can help organize reflection when treated as a hypothesis, with room for context and disagreement. Once the interpretation is asked to justify an opportunity, rating, or reward, that exploratory role is insufficient; the evidence must support the consequential inference itself.

Sources: Standards for Educational and Psychological Testing; Standards for Educational and Psychological Testing, 2014 Edition

How should you read two WPI reports side by side?

A practical side-by-side reading starts with the report pages themselves. First, check the edition and completion date printed on each page; keep the earlier WPI2 and revised WPI2-R labels attached to their respective scores. Next, match only scale labels that the documentation identifies as counterparts. Similar wording is not enough to assume that two labels refer to the same scale. Then read each value in the Sten and norm context printed or described for that report. The manual’s stated common standardized frame supports comparing corresponding scores in that context. Record what the pages show before deciding what it means.

For example, imagine a reader has two reports with a documented counterpart scale on both. They could write down each edition and date, copy the matched scale names, and note the two printed Sten values alongside the relevant norm information. If the values differ, the reader can describe the observed difference plainly, without translating it into a cause: the pages show different scores at those administrations. The pages alone do not establish whether the gap reflects a lasting change, a change in circumstances, measurement uncertainty, or some combination. This hypothetical illustrates a reading process; it does not describe a respondent or empirical result.

The next useful move is to connect the report pattern to one real situation rather than attach a broad label to the person. A reader might ask, “In the planning meeting I have in mind, what did I actually do when the deadline changed?” or “What did I observe about my response to feedback in that project?” The question makes the score comparison a prompt for checking behavior and context. If examples fit one report better than the other, record those examples and the conditions around them; do not assume that fit explains the numerical gap. Keep the publisher’s comparison and the reader’s interpretation distinct: the former concerns how corresponding standardized scores may be read, while the latter asks what a particular difference means in an individual’s life.

This sequence preserves the useful comparison the reports allow while keeping the conclusion proportionate. It yields a concrete description of corresponding values, the context shown on the pages, and a question that can be checked against experience. A reader should not make the sequence more elaborate than the available documents permit: verify, match, read, note, then ask. If the question is simply how two reported scores sit in the publisher’s stated common frame, the side-by-side reading addresses it. If the question is why the scores differ or whether that difference should change a consequential decision, the pages and this reflection exercise do not settle it; that next question requires more specific evidence and a suitable decision process.

Next step: turn the report into one observation

If your purpose is reflection, choose one recent planning, feedback, decision, or collaboration situation. Write down what happened, what you did, and what you would want to try next time. Keep those observable actions separate from the report’s description, then ask whether the pattern offers a useful question to examine. One specific example can make a broad work concern easier to discuss without treating a score as a verdict.

If the difference could affect a consequential decision, pause before interpreting the gap. Ask the administrator for version-specific evidence about the precision or agreement of individual WPI2 and WPI2-R differences, and for evidence supporting the intended use. A general reliability figure or the fact that reports can be compared in a standardized frame does not answer those individual and purpose-specific questions. Until the relevant evidence is available, describe the scores and the uncertainty rather than assigning a cause or consequence.

For private reflection, the live Work Pattern Report at [/assessment](/assessment) offers a separate way to examine your own responses across ten work-pattern continuums. It is a non-validated self-reflection tool: it supplies no norms or cutoffs, and it cannot resolve whether WPI editions are statistically interchangeable. It is not a hiring score, diagnosis, or job recommendation. Use it to turn a vague work question into observations, not to decide what a WPI difference proves.

Questions readers ask