Research reference
How TDF Was Developed and Studied
TDF's research manual documents how the inventory was developed and how its results were examined. This guide explains the main lines of evidence and what each can tell you.
How did the model become an inventory?
TDF began with a model of three perceptual lenses: facts, choices and connections. Developing an inventory required a further step: selecting self-description items and examining how their responses related to the six proposed patterns.
The TDF Pattern Inventory Research Manual describes successive revisions from an earlier forced-ranking format to adjective ratings. In the development work it reports, candidate words were examined in an adult business sample and selected for their relationships with the pattern scales. Source and research context.
This history distinguishes the framework from any one questionnaire used to explore it. Changes to items, language, response options or scoring create questions about how earlier findings apply. The manual documents the versions it studied; it does not by itself establish the performance of every later implementation.
What did the research examine?
| Area | What was examined? | How to read it |
|---|---|---|
| Internal consistency | Relationships among items within each pattern scale, across three samples. | Evidence about scale consistency in those samples. |
| Pattern structure | Relationships among the six scales and their proposed arrangement around the TDF wheel. | An examination of whether the scale relationships fit the model. |
| Other inventories | Relationships with measures of personality, vocational interests and interpersonal needs. | Evidence about overlap and distinctions between constructs. |
| MBTI classifications | Paired MBTI types and TDF patterns for 443 respondents. | A comparison of classifications, without a one-to-one conversion. |
These are different lines of inquiry. Agreement with part of a model, for example, is not a measurement of workshop effectiveness. Keep the research question attached to the finding when using or citing it.
What did the consistency studies find?
The manual reports Cronbach's alpha values from .85 to .92 across the six pattern scales in a development sample of 861 adult businesspeople. In a follow-up sample of 333 adult businesspeople, the values ranged from .80 to .90. A French-language sample of 129 participants produced values from .81 to .91.
These coefficients describe internal consistency. They are not percentages of correct pattern assignments. The authors also note that results from a sample used to select the items can look stronger than results in a later sample.
The manual states that test-retest reliability had not been directly assessed. Consequently, these findings do not establish how often an individual would receive the same pattern over time. The reliability and validity guide explains these distinctions with practical examples.
What do the comparisons with other assessments add?
The manual examines relationships with several inventories, including the Jackson Personality Inventory, Personality Research Form, Holland Vocational Preference Inventory and FIRO-B. These comparisons explore how TDF pattern scales relate to other descriptions of personality, interests and interpersonal needs.
The MBTI comparison uses a different form of evidence: reported types and patterns for the same participants. People sharing an MBTI type could have different TDF patterns. Read TDF and MBTI: How the Frameworks Compare for two concrete examples and guidance on interpreting both results.
The studies do not establish that a pattern determines a career, that an inventory predicts job performance, or that one assessment is superior to another. A question about those outcomes would require evidence designed to address it.
How should you use this evidence?
Use the research to understand what has been examined and to make your questions more precise. If you are choosing an assessment, ask how the studied version and population relate to your intended use. If you are reading your own result, compare the description with concrete experiences across settings.
The manual is an account produced within the TDF organization. This summary does not present it as independent peer-reviewed confirmation. The practical exercises elsewhere in Resources are applications of the framework; they are not interventions whose outcomes were established by these studies.
For help with your own report, start with how to read your TDF assessment results. It keeps the focus on understanding a result and choosing something useful to try.
Sources and research context
Primary source: Spencer, Boyd; Roberts, Bill; and Farr, David. TDF Pattern Inventory Research Manual. Second edition, Impact|TDF, copyright 2002. Page references below use the manual's printed page numbers.
- Development and lens definitions: pp. 1–5. The manual describes development beginning in the late 1970s and the adjective-based form developed in 1984.
- Internal consistency: pp. 5–6, Tables 1–3. Development sample: 1984, n=861. Follow-up English sample: 1989, n=333. French-language sample: 1989, n=129. Page 8 notes a five-point response format for the follow-up English sample, compared with the six-point format described elsewhere.
- Structure and inventory comparisons: pp. 7–14. Several comparison samples are small and occupationally restricted, which limits generalization.
- MBTI comparison: pp. 14–16, particularly Table 11, n=443. Data are attributed to JFA Consulting, Dallas. The data-collection year is not specified in that section; the manual's publication year should not be treated as the study year.
Notes on the printed tables
In Table 11, the listed DFT cells add to 97, while the printed column subtotal is 71. This is an arithmetic observation about the published table, not a correction verified against participant-level data. The examples used in our MBTI article concern other columns whose listed cells and subtotals agree.
Some percentages and the response-style tables also contain inconsistencies. We have not reproduced those results or used them to build a predictive tool. The findings summarized here remain tied to their original versions and samples, with the limits described above.