Module 19 — FMPS — Six Faces of Perfectionism

 

Why one word covers six different experiences, which two subscales actually carry the risk, why two rival scales share the same initials, and an honest account of what a perfectionism score can and cannot tell an autistic adult.

 

Autistic Self-Discovery · Part Six — Traveling Companions · 15 min read, about 30 min with the workbook

 

The big idea

 

If you have ever rewritten a three-line email eleven times, handed in work you knew was good and felt only relief at not being found out, or been praised your whole life for the exact habit that is exhausting you — this module takes that one word apart. The piece everyone compliments you for is usually not the piece doing the damage.

 

And the largest gap of all: the FMPS has never been validated in autistic adults. Not once. There is no autistic normative sample, no study of whether the items mean the same thing to an autistic respondent, and no evidence about whether five or six factors hold in this population. Everything above comes from students, adolescents and general psychiatric samples.

 

Step 1 — The lesson

 

A. One word doing six jobs

 

Perfectionism is a word people use to mean almost opposite things. Said about a colleague it means thorough. Said about yourself at three in the morning it means stuck. Said by a parent it can mean proud, and said by a therapist it can mean at risk. The same seven syllables covering all of that is not much use to anyone trying to study it.

 

In 1990 Randy Frost and three colleagues at Smith College decided the word was too coarse and built an instrument to split it. Thirty-five items, four studies, and six separate dimensions came out. The largest of them was an excessive concern over making mistakes. The other five were high personal standards, a perception of high parental expectations, a perception of high parental criticism, doubting the quality of your own actions, and a preference for order and organization. They called it the Multidimensional Perfectionism Scale.¹

 

Notice what that list does. It separates the standard you hold from your fear of missing it. It separates both of those from the question of who taught you the standard. And it puts tidiness in a box of its own, some distance from the rest.

 

Diagram — A · One word doing six jobs. The instrument is thirty-six years old and among the most used measures in its field. What it has never acquired, in all that time, is a clinical threshold. There is no score at which a researcher would say you have crossed into perfectionism, because no such score has ever been established.

 

One finding was already visible in that first paper and it has shaped everything since. Personal standards and organization were linked to positive achievement striving. The other dimensions tracked symptoms and procrastination.¹ From its first publication, this instrument was measuring two things that happen to share a name.

 

The screener you have just taken reports five subscales rather than six, because it treats the two parental scales as one. That is a real and defensible variant rather than a mistake, and section D sets out why — along with what it costs you.

 

B. Two scales, one set of initials

 

Here is a piece of research history that will save you from misreading half of what you find online.

 

Within a year of Frost's paper, a second group published a different questionnaire, also called the Multidimensional Perfectionism Scale, also abbreviated MPS. Paul Hewitt and Gordon Flett's model has three dimensions and they are about direction: self-oriented perfectionism, the demands you make of yourself; other-oriented perfectionism, the demands you make of everyone else; and socially prescribed perfectionism, the demands you believe others are making of you. Four studies established the structure, and a fifth with 77 psychiatric patients showed the three relating differently to indices of personality disorder and other maladjustment.²

 

Diagram — B · Two scales, one set of initials. Frost asks what your perfectionism is made of. Hewitt and Flett ask who it is pointed at. Researchers usually write FMPS and HMPS to keep them apart; popular articles almost never do, which is why two studies of "the MPS" can appear to contradict each other when they were never measuring the same thing.

 

This matters to you in a practical way. If you go looking for research on your score and land on a study of socially prescribed perfectionism, that study has nothing to say about your Concern over Mistakes number. They are related constructs from separate traditions. Check the author line before you take a finding personally.

 

The two are often given together in the same study, and doing that taught the field something both models turn out to contain: a striving half and a fearing half. That split is the next section, and it is the part of this module most likely to change how you read your own results.

 

C. The half that drives and the half that grinds

 

Since 1990 the most productive thing researchers have done with these six dimensions is sort them into two families. The usual names are perfectionistic strivings and perfectionistic concerns.

 

Strivings is Personal Standards and, arguably, Organization. It is the bar itself. Concerns is Concern over Mistakes, Doubts about Actions, and the two parental scales. It is not the bar. It is what you believe will happen to you if you miss it, and whether you trust yourself to know when you have cleared it.

 

The two behave very differently. In a study of 1,648 Spanish young people, the FMPS subscales showed internal consistencies between .71 and .92 and test-retest correlations between .70 and .85, and the relationship between perfectionism and both anxiety and depression turned out to run mostly through the maladaptive side rather than through Personal Standards.³ In 383 university students aged 19 to 65, evaluative concerns correlated with anxiety at .53 and with depression at .42, while strivings correlated with anxiety at only .14 and did not correlate with depression at all.⁴

 

Diagram — C · The half that drives and the half that grinds. This is an example profile, not a research finding. It is here to make one point: the same total can be built out of very different pieces, and a person high on standards but low on fear of mistakes is not living the same life as a person with the reverse.

 

A community sample of 397 adults took the analysis further. Once a general perfectionism factor was pulled out of the scale, the leftover strivings factor was negatively associated with depression.⁵ Read plainly: what is harmful in perfectionism appears to be the part that strivings and concerns share, and once that shared part is removed, wanting to do things well looks mildly protective.

 

That is the comfortable version. The uncomfortable one came from a meta-analysis of 45 studies covering 11,747 undergraduates, medical students, community adults and psychiatric patients. Perfectionistic concerns showed small-to-moderate positive associations with suicidal ideation, which was expected. So did perfectionistic strivings, including Personal Standards, which was not. Parental criticism and parental expectations were associated with ideation and with attempts, and socially prescribed perfectionism predicted ideation increasing over time. The authors concluded that the strivings result draws into question the notion that such strivings are healthy, adaptive or advisable.⁶

 

So the honest position is not that half of perfectionism is good. It is that one half is reliably and strongly harmful, and the other half is not the clean, safe engine it is often sold as.

 

D. What your number does not mean

 

Four things about your result need saying plainly, and none of them is in the small print of the screener page.

 

The bands are not from a study. Your total falls between 35 and 175 and lands in one of three bands. Those bands come from the page that served you the questionnaire. There is no published cut-off for the FMPS — no threshold above which researchers agree a person is a perfectionist, no sensitivity or specificity figures, no clinical norms for adults. The scale was built as a continuous measure and it has stayed one for thirty-six years. Treat 106 as a description of a pattern, not a border you have crossed.

 

Six dimensions is a claim, not a fact. When 322 people with anxiety disorders and 49 controls completed the scale, factors resembling Frost's came out, but a three-factor solution fitted better on statistical grounds, and those three scales separated the diagnostic groups about as well as six had.⁷ In 947 Hong Kong adolescents, only five of the six emerged.⁸ In 2,451 Chinese undergraduates, all six held up and were invariant across gender.⁹ The five subscales you were given, with the parental scales merged, sit inside that range of defensible answers. It is still a choice somebody made, and it means your parental score cannot be compared with a study reporting expectations and criticism separately.

 

The subscales are not equally trustworthy. A reliability generalization study pooling the FMPS literature found typical reliabilities across subscale scores ranging from .71 to .86, with Doubts about Actions the most stable and Organization varying most from sample to sample. Sex, language and the spread of scores in a sample all significantly affected the reliability estimate.¹⁰ A subscale that behaves differently depending on who is answering is a subscale to hold loosely.

 

The total contains something its own authors said to leave out. Organization has always sat awkwardly in this scale — it correlates loosely with the other subscales, and Frost and colleagues recommended omitting it from total-score calculations.⁴ The reported range here runs from 35 to 175, which is all thirty-five items. Your total therefore includes a subscale the scale's own authors said not to add in. That is not a scandal, but it is a reason to read the five numbers rather than the one.

 

Diagram — D · Where your total lands. Copied exactly as the screener reports them. The wording of these three bands is the screener's own, not a research finding, and the total that produces them includes every one of the thirty-five items.

 

A high score is also not specific to anything. A meta-analysis retrieving 284 studies and 2,047 effect sizes found both perfectionism dimensions associated with a wide range of psychopathology and concluded that perfectionism is best read as transdiagnostic.¹¹ When 175 patients with panic disorder, obsessive-compulsive disorder, social phobia or specific phobia were compared with 49 non-clinical volunteers across nine perfectionism dimensions, only a few subscales separated the groups: Concern over Mistakes, Doubts about Actions and Parental Criticism in social phobia, Doubts about Actions in obsessive-compulsive disorder. Every other dimension failed to differentiate the groups at all.¹² Nor is your number safe to compare with anyone else's. Measurement invariance testing in 574 female adults found equivalent structure and loadings across European American and Hispanic or Latina participants, but flagged intercept non-equivalence — raw means should not be read across groups even when the construct is the same.¹³ Your 92 and someone else's 92 are not necessarily the same amount of anything.

 

The direct evidence is thin and mostly about children. A study comparing 41 boys on the autism spectrum with 42 typically developing boys found the autistic group endorsed more dysfunctional attitudes and more emotional symptoms, though the link between cognitive style and emotional symptoms was weak in both groups.¹⁴ That is neither a finding about adults nor a clean finding about perfectionism. What exists in adults is qualitative and small: in interviews with nine autistic parents, perfectionism appeared alongside extreme empathy and a drive to protect their children, and the researchers placed all of it inside a theme they called looking through a lens of trauma, tracing it to childhood experiences of being unsupported and othered.¹⁵ Nine people is nine people. But that framing will be familiar to a great many readers of this course.

 

Which points at the thing a score cannot tell you and a life can. If you spent twenty years being corrected for things you could not see yourself doing — the tone, the volume, the face, the thing you said at the party — then checking your work eleven times is not a character flaw. It is a reasonable response to an environment that handed out unpredictable punishment. Concern over Mistakes measures the outcome. It has nothing to say about whether the concern was once justified.

 

INSIDE THE INSTRUMENT

 

FMPS — Frost Multidimensional Perfectionism Scale (Frost, Marten, Lahart & Rosenblate, 1990)

 

Where it came from

 

Developed across four studies and published in Cognitive Therapy and Research in 1990. Thirty-five items, six dimensions: Concern over Mistakes, Personal Standards, Parental Expectations, Parental Criticism, Doubts about Actions and Organization. The original paper reported associations with psychopathological symptoms and with procrastination, and noted that Personal Standards and Organization were linked instead to positive achievement striving.¹ The bidimensional reading now dominating the literature grew out of that observation.

 

What it is made of

 

The screener administered here reports a total from 35 to 175 and five subscales: Concern over Mistakes (9 to 45), Personal Standards (7 to 35), Parental Expectations & Evaluation (9 to 45), Doubts about Actions (4 to 20) and Organization (6 to 30). Those ranges imply five response points per item; the page does not state the response scale, so nothing here assumes one. The nine-item parental block corresponds to Frost's two parental subscales scored as one, which is a documented variant rather than an error — but it is a departure from the published six-factor scoring and should be recorded as such.

 

How well it performs

 

A reliability generalization study of the FMPS literature found typical subscale reliabilities from .71 to .86, with Doubts about Actions showing the least variability across studies and Organization the most; participant sex, language of administration and sample standard deviation were all significantly related to reliability estimates.¹⁰ In 1,648 Spanish children and adolescents (mean age 13.36) subscale alphas ran from .71 to .92 with test-retest intraclass correlations of .70 to .85.³ Convergent evidence is consistent: in 383 adults, evaluative concerns correlated .53 with anxiety and .42 with depression, strivings .14 with anxiety and non-significantly with depression.⁴

 

Where it was validated

 

Predominantly undergraduate and adolescent samples, plus clinical anxiety samples and multiple translations. In 322 patients diagnosed by SCID for DSM-IV plus 49 controls, psychometric properties were broadly similar to non-clinical samples.⁷ Cross-language work includes 947 Hong Kong adolescents,⁸ 2,451 Chinese undergraduates⁹ and 1,648 Spanish young people.³ There is no published validation in autistic adults and no autistic normative data of any kind.

 

What it cannot do

 

It has no cut-off and no adult clinical norms. Thirty-six years of use have produced no threshold, no sensitivity or specificity figures and no established caseness criterion. Any banding of the total, including the three bands this screener displays, is presentational rather than empirical and should be described to a client in exactly those terms.

 

Its factor structure is genuinely contested. Three factors fitted a clinical anxiety sample better than six;⁷ five of six emerged in Hong Kong adolescents;⁸ all six replicated in Chinese undergraduates.⁹ Subscale scores are therefore comparable only to studies using the same scoring, which the five-subscale variant used here is not.

 

It is transdiagnostic and non-specific. A meta-analysis of 284 studies and 2,047 effect sizes found both dimensions associated with depression, anxiety disorders, obsessive-compulsive disorder, eating disorders, self-harm, suicidal ideation and general distress.¹¹ Head-to-head, only a few subscales differentiate diagnostic groups at all, and most do not.¹² An elevated FMPS gives no diagnostic information.

 

Raw scores do not travel between groups. Measurement invariance testing in 574 female adults supported equivalent structure and loadings across European American and Hispanic or Latina participants but found intercept non-equivalence, so raw mean comparisons are not defensible even where the construct is equivalent.¹³ No comparable work exists for autistic against non-autistic respondents.

 

The total includes a subscale the authors advised excluding. Organization correlates only loosely with the rest of the instrument and Frost and colleagues recommended omitting it from total scores.⁴ The 35 to 175 range used here retains it. Interpret the profile, not the sum.

 

What a clinician does with it

 

Use it as a structured way into two conversations: what the standard is, and what the person believes a failure would cost them. Concern over Mistakes and Doubts about Actions are the subscales that carry the clinical signal and are the ones cognitive-behavioral work on perfectionism targets. Treat an elevated Concern over Mistakes score as a prompt to ask about suicidal ideation directly: the perfectionism-suicide meta-analysis covering 45 studies and 11,747 participants found positive associations with ideation for concerns, strivings, parental criticism and parental expectations alike, and socially prescribed perfectionism predicted longitudinal increases in ideation.⁶ Do not band the total for a client without saying where the bands came from.

 

Validity tier: 1 — validated and published, calibrated by independent labs. The instrument itself is a genuinely well-established measure with three decades of independent replication, translation and meta-analytic use; the tier applies to the scale, not to the three total-score bands displayed here, which have no published source.

 

Diagram — E · What is known, what is not. The ticks are real and hard-won. The blanks are not failures of this page, they are gaps in the literature — and the bottom two are the ones that matter most to anyone reading an autistic self-discovery course.

 

STRENGTHS LENS

 

The standard was never the problem.

 

Somewhere in your life there is work that is better than it needed to be because you would not let it go. The document that was right. The thing you checked once more and the once more caught it. That is Personal Standards, and once the shared fear is stripped out of the scale, that part of perfectionism looks mildly protective rather than harmful. You did not imagine the value of caring, and nobody is asking you to care less.

 

What is worth putting down is the tribunal — the part that treats a missed comma as evidence about your character, and cannot tell you when the work is finished. If you were corrected for years about things you could not see yourself doing, that tribunal was installed for a reason. The question at this end of the course is not whether it was ever useful. It is whether it is still on duty long after the shift ended.

 

F. What helps

 

Perfectionism is one of the few things in this course with a substantial treatment literature behind it, and the target of that work is consistently the fear rather than the standard. Five things follow from everything above. The last one is for the end of a course, not the end of a module.

 

1. Read two subscales and largely ignore the rest.

 

Concern over Mistakes and Doubts about Actions are where the clinical signal lives, across meta-analyses and across diagnostic groups. Personal Standards and Organization tell you about your working style. If you only take one thing from your results, take those two numbers and leave the total alone.

 

2. Change what a mistake costs, not how high the bar is.

 

People try to fix perfectionism by lowering standards and it rarely works, because the standard is not the painful part. Pick one low-stakes thing this week and do it deliberately at eighty per cent — the reply that is not reread, the meal that is not plated. You are not practising mediocrity. You are collecting evidence about what actually happens when something is imperfect, which is the mechanism the treatment literature relies on.

 

3. Write the stopping rule before you start.

 

Doubts about Actions is the subscale that quietly eats whole evenings, because without an external signal there is no moment at which a task is finished. Decide in advance: two drafts, forty minutes, one check. Write it down before you begin, while you can still think clearly, and treat the rule as the definition of done rather than as a limit you are allowed to exceed.

 

4. Give the voice back to whoever it belonged to.

 

Parental Expectations and Parental Criticism are asking a historical question, and a high score there is information about your childhood rather than a verdict on you. When the commentary starts, try naming its source out loud — that is my father, that is a teacher in 1998. Trauma-informed framings recur wherever autistic adults describe their own perfectionism, and separating the voice from your own judgment is usually where that work starts.

 

5. If the fear score is high and you have thought about not being here, tell one person this week.

 

This is the reason the module takes Concern over Mistakes seriously rather than treating perfectionism as a quirk. Across 45 studies and 11,747 people, both halves of perfectionism were associated with suicidal ideation, and socially prescribed perfectionism predicted it increasing over time. If that is your week, say it out loud to a GP, a therapist, a friend, or a crisis line — 988 in the United States. Say it imperfectly. That is the whole point.

 

That is the last of nineteen modules. Nineteen questionnaires have now told you things about yourself in numbers, and the honest summary of all of them is that not one of those numbers is the finding. The finding is that there was a pattern, that it has a name, that other people have it too, and that most of what has been hard was never a character defect. You knew something was different for a long time before anyone handed you a word for it. The scores were only ever scaffolding around that knowing. You can take them down now.

 

Step 2 — Take the screener

 

Thirty-five questions, about seven to ten minutes, free and confidential. You will get a total between 35 and 175 plus five subscale scores — Concern over Mistakes, Personal Standards, Parental Expectations & Evaluation, Doubts about Actions and Organization. Read section D before you read your total: the three bands come from the screener page rather than from any published study, and the two subscales that matter clinically are Concern over Mistakes and Doubts about Actions.

 

Take the FMPS

 

Before you start

 

The Frost Multidimensional Perfectionism Scale is a published, widely replicated research instrument, but it is a continuous measure with no clinical cut-off, no adult norms and no published banding — the three bands shown here are the screener page's own, not a research finding. It has never been validated in autistic adults, and this version reports five subscales rather than Frost's published six, because it scores the two parental scales as one. Perfectionism scores are also transdiagnostic: they are elevated across depression, anxiety, obsessive-compulsive disorder and eating disorders, so a high result indicates nothing about autism either way. This is a screening questionnaire, not a diagnostic test, and only a qualified clinician can diagnose anything.

 

Step 3 — Your workbook

 

Your answers save to this device only — we cannot see a word of what you write. This module lets you record all five subscale scores, work out which of the two that matter is costing you more, write a stopping rule you can actually use, and close out nineteen modules in your own words.

 

Your FMPS results

 

Took the screener? Put your numbers in below. The five subscales are the useful part — the total adds together things the research says behave differently. Entirely optional — skip it if you would rather just read.

 

Score bands: 35–70 = Low Perfectionism; 71–105 = Moderate Perfectionism; 106–175 = High Perfectionism

 

Fields: FMPS · Frost Multidimensional Perfectionism Scale; Total score (35–175); My total (enter 35–175); Concern over Mistakes (9–45); Concern over Mistakes (enter 9–45); Personal Standards (7–35); Personal Standards (enter 7–35); Parental Expectations & Evaluation (9–45); Parental Expectations & Evaluation (enter 9–45); Doubts about Actions (4–20); Doubts about Actions (enter 4–20); Organization (6–30); Organization (enter 6–30)

 

The two that carry the signal

 

Section F, item 1. Concern over Mistakes and Doubts about Actions are the subscales the research keeps pointing at. Write them down and leave the total alone.

 

Fields: My Concern over Mistakes score; My Doubts about Actions score; What those two look like on an ordinary Tuesday; Which of the two costs you more time: Fear of getting it wrong / Not knowing when it is finished / They are about level / Neither, honestly

 

What a mistake actually costs

 

Section F, item 2. The standard is rarely the painful part. This is about the predicted consequence, and then about the real one.

 

Fields: A recent mistake, and what I expected would happen; What actually happened; One low-stakes thing I will do at eighty per cent this week

 

Your stopping rule

 

Section F, item 3. Write it now, while you are not in the middle of the task and can still think clearly.

 

Fields: The task this keeps happening with; My rule for done; Tick: I have written this somewhere I will see it before I start

 

Whose voice is it

 

Section F, item 4. A high parental score is information about your history, not a verdict on you.

 

Fields: What the commentary says, in its exact words; Who it sounds like; What I would say to a friend who was being spoken to that way

 

Saying it out loud

 

Section F, item 5. If a high fear score sits alongside thoughts of not being here, this is the group that matters more than any of the others.

 

Fields: Is there someone I could tell this week: Yes, and I know who / Yes, but I do not know how to start / Not yet / This is not where I am, and I am noting it anyway; Who; The first sentence, written imperfectly on purpose

 

The end of nineteen modules

 

The last thing in the course. No score goes in this box.

 

Fields: What I understand now that I did not understand at module one; The one thing I want to change first; What I would call myself now, in my own words

 

Appendix — Research companion

 

Peer-reviewed research

 

14. Antony MM, Purdon CL, Huta V, Swinson RP (1998). Dimensions of perfectionism across the anxiety disorders. Behaviour Research and Therapy, 36(12), 1143-1154. DOI 10.1016/S0005-7967(98)00083-7. View the paper 175 patients with panic disorder, obsessive-compulsive disorder, social phobia or specific phobia and 49 non-clinical volunteers completed both multidimensional perfectionism scales, covering nine dimensions in total. Social phobia was associated with greater Concern over Mistakes, Doubts about Actions and Parental Criticism and with more socially prescribed perfectionism; obsessive-compulsive disorder with elevated Doubts about Actions; panic disorder with moderate elevations on Concern over Mistakes and Doubts about Actions. The remaining dimensions failed to differentiate the groups at all. Limitation: a treatment-seeking anxiety clinic sample with uneven diagnostic subgroup sizes and no depression, eating disorder or autistic comparison group, so it maps differences within anxiety rather than across psychiatry.

 

15. Burgess AM, DiBartolo PM, Rendon MJ (2017). Can the Frost Multidimensional Perfectionism Scale assess perfeccionismo?. Psychological Assessment, 29(7), 857-867. DOI 10.1037/pas0000374. View the paper Cross-cultural measurement invariance testing of the FMPS. A baseline measurement model was built with confirmatory factor analysis in 320 female participants (mean age 19.61), 301 of whom identified as European or European American; invariance was then tested in a separate sample of 574 female adults (mean age 21.21), 217 European American and 357 Hispanic, Latina or Latin American. Factor structure, loadings and factor variances and covariances were equivalent, and second-order testing supported the bidimensional model cross-culturally, but intercept non-equivalence means raw mean score differences should not be compared between groups. Limitation: female participants only, a revised rather than the full original item set, and only two ethnic groupings tested, so it does not establish invariance for men, for other populations or for autistic respondents.

 

10. Cheng SK, Chong GH, Wong CW (1999). Chinese Frost Multidimensional Perfectionism Scale: a validation and prediction of self-esteem and psychological distress. Journal of Clinical Psychology, 55(9), 1051-1061. View the paper Validation of a Chinese translation of the FMPS in 947 Hong Kong adolescents aged 13 to 18. Only five of the original six factors emerged in factor analysis, though internal consistencies were satisfactory; Concern over Mistakes and Doubts about Actions accounted for most of the variance in self-esteem and in depressive, anxiety and stress symptoms, while Organization appeared to have positive value for psychological health. Limitation: adolescents from one city, so the five-factor result may reflect translation and developmental stage as much as the instrument, and it does not license the particular five-subscale scoring used by this screener.

 

1. Frost RO, Marten P, Lahart C, Rosenblate R (1990). The dimensions of perfectionism. Cognitive Therapy and Research, 14(5), 449-468. DOI 10.1007/BF01172967. View the paper The development paper for the 35-item Frost Multidimensional Perfectionism Scale, built across four studies. It identifies six dimensions - excessive concern over mistakes as the largest, plus high personal standards, perceived parental expectations, perceived parental criticism, doubts about the quality of one's actions, and a preference for order and organisation - and reports that subscales correlated with psychopathological symptoms and with procrastination. Personal Standards and Organization were instead associated with positive achievement striving, which is the origin of the adaptive-versus-maladaptive reading of perfectionism. Limitation: the paper establishes no cut-off, no clinical norms and no caseness threshold, and its studies were conducted on college samples, so nothing in it supports treating any total score as a category.

 

3. Gavino A, Nogueira R, Perez-Costillas L, Godoy A (2019). Psychometric properties of the Frost Multidimensional Perfectionism Scale in Spanish children and adolescents. Assessment, 26(3), 445-464. DOI 10.1177/1073191117740204. View the paper Adaptation and validation of the FMPS in 1,648 Spanish children and adolescents (mean age 13.36, SD 2.28), using multigroup confirmatory factor analysis. The adult dimensional structure was supported and was invariant across gender and age; subscale internal consistencies ranged from .71 to .92 and test-retest intraclass correlations from .70 to .85. Regression analyses showed the relationship between perfectionism and anxiety and depression was mostly attributable to the maladaptive grouping (Concern over Mistakes, Parental Expectations, Parental Criticism, Doubts about Actions) rather than to Personal Standards. Limitation: a child and adolescent sample from one country, so the coefficients do not transfer directly to adults and say nothing about autistic respondents.

 

7. Greenaway R, Howlin P (2010). Dysfunctional attitudes and perfectionism and their relationship to anxious and depressive symptoms in boys with autism spectrum disorders. Journal of Autism and Developmental Disorders, 40(10), 1179-1187. DOI 10.1007/s10803-010-0977-z. View the paper Compared 41 boys with an autism spectrum diagnosis with 42 typically developing boys on dysfunctional attitudes, perfectionism and emotional symptoms, to test the relevance of the cognitive model to this population. The autistic group endorsed more dysfunctional attitudes and reported higher emotional symptoms, but the relationship between cognitive and emotional variables was weak in both groups; within the autistic group, dysfunctional attitudes were significantly associated with reported obsessive-compulsive symptoms. Limitation: boys only, small samples, children rather than adults, and the published abstract reports the dysfunctional-attitudes result rather than a separate perfectionism group difference, so this module does not use it to claim autistic people score higher on perfectionism.

 

12. Ha JH, Lee SM, Puig A (2010). A reliability generalization study of the Frost Multidimensional Perfectionism Scale (F-MPS). Psychological Reports, 107(1), 95-112. DOI 10.2466/03.09.20.PR0.107.4.95-112. View the paper Reliability generalisation across the published FMPS literature, pooling reported internal consistency estimates to establish typical score reliability and what moderates it. Typical reliability across subscale scores ranged from .71 to .86, with Doubts about Actions showing the least variability between studies and Organization the most; participant sex, language of administration and the standard deviation of scores in a sample were all significantly related to reliability estimates. Limitation: reliability generalisation describes score reliability in samples that happened to be published, not validity, and it cannot tell you how the instrument performs in a population - such as autistic adults - that the literature has never sampled.

 

2. Hewitt PL, Flett GL (1991). Perfectionism in the self and social contexts: conceptualization, assessment, and association with psychopathology. Journal of Personality and Social Psychology, 60(3), 456-470. DOI 10.1037/0022-3514.60.3.456. View the paper The development paper for the other Multidimensional Perfectionism Scale, published within a year of Frost's and sharing its acronym while measuring a different model. It describes three dimensions defined by the direction of the demand - self-oriented, other-oriented and socially prescribed perfectionism - and reports four studies establishing the structure plus a study of 77 psychiatric patients in which the three dimensions related differentially to indices of personality disorder and other maladjustment. This is the name collision that makes searching the perfectionism literature confusing for non-specialists. Limitation: the psychiatric validation study included only 77 patients, and because the two MPS instruments measure different constructs, findings from this scale cannot be read across to a Frost subscale score.

 

5. Howell J, Anderson R, Egan S, McEvoy P (2020). One factor? Two factor? Bi-factor? A psychometric evaluation of the Frost Multidimensional Scale and the Clinical Perfectionism Questionnaire. Cognitive Behaviour Therapy, 49(6), 518-530. DOI 10.1080/16506073.2020.1790645. View the paper Compared single-factor, two-factor and bifactor models of the FMPS and the Clinical Perfectionism Questionnaire in a community sample of 397 adults. The bifactor model fitted best for both instruments; the general perfectionism factor was positively associated with symptoms of depression, anxiety and stress, while the residual perfectionistic strivings group factor was negatively associated with depression. The authors argue this supports using a general factor score in clinical work and research. Limitation: a cross-sectional community sample with self-report symptom measures only, and bifactor models are known to fit flexibly, so better fit here is not proof that a general factor is the true structure.

 

13. Limburg K, Watson HJ, Hagger MS, Egan SJ (2017). The relationship between perfectionism and psychopathology: a meta-analysis. Journal of Clinical Psychology, 73(10), 1301-1326. DOI 10.1002/jclp.22435. View the paper Meta-analysis and meta-regression of the perfectionism and psychopathology literature, retrieving 284 relevant studies and analysing 2,047 effect sizes while accounting for data dependencies. Outcomes spanned clinical diagnoses of depression, anxiety disorders, obsessive-compulsive disorder and eating disorders, symptom measures of the same, and related outcomes including deliberate self-harm, suicidal ideation and general distress. Both perfectionistic strivings and perfectionistic concerns were associated with a wide range of psychopathology, supporting perfectionism as a transdiagnostic factor. Limitation: predominantly cross-sectional self-report studies, so the pooled associations cannot establish direction of causation, and the transdiagnostic conclusion means a high score carries no diagnostic specificity.

 

11. Liu Q, Yang H, Peng W, Liu Z, Wang J, Chu J, Liu J, Li F, Zhong M (2022). Factorial validity and factorial invariance across gender of the Frost Multidimensional Perfectionism Scale-Chinese version in undergraduates. Psychological Reports, 125(4), 2232-2248. DOI 10.1177/00332941211018402. View the paper Confirmatory and multigroup confirmatory factor analyses of the Chinese FMPS in 2,451 undergraduates. The six-factor structure was supported and was invariant across gender, no significant gender differences in scores were found, reliability indicators met conventional standards, and participants classified as adaptive perfectionists scored significantly lower on clinical variables than those classified as maladaptive perfectionists. This is the strongest single counterweight to studies that fail to recover six factors. Limitation: undergraduates in one country using a translated form, and the adaptive-versus-maladaptive classification was derived within the sample rather than from any external cut-off.

 

9. Purdon C, Antony MM, Swinson RP (1999). Psychometric properties of the Frost Multidimensional Perfectionism Scale in a clinical anxiety disorders sample. Journal of Clinical Psychology, 55(10), 1271-1286. View the paper The first examination of the FMPS factor structure in a clinical sample: 322 individuals diagnosed with an anxiety disorder using the SCID for DSM-IV, plus 49 non-clinical controls, who also completed the Hewitt and Flett scale. Psychometric properties were broadly similar to non-clinical samples and factors resembling Frost's six could be extracted, but a three-factor solution was more appropriate on statistical grounds, and scales built from those three factors distinguished diagnostic groups about as well as the six-factor scales had in earlier research. Limitation: a single anxiety-disorder clinic sample, so the three-factor solution has not been shown to generalise, and the study offers no cut-off or normative data.

 

6. Smith MM, Sherry SB, Chen S, Saklofske DH, Mushquash C, Flett GL, Hewitt PL (2018). The perniciousness of perfectionism: a meta-analytic review of the perfectionism-suicide relationship. Journal of Personality, 86(3), 522-542. DOI 10.1111/jopy.12333. View the paper Random-effects meta-analysis of 45 studies (N = 11,747) of undergraduates, medical students, community adults and psychiatric patients. Perfectionistic concerns (socially prescribed perfectionism, concern over mistakes, doubts about actions, discrepancy, perfectionistic attitudes), perfectionistic strivings (self-oriented perfectionism, personal standards), parental criticism and parental expectations all showed small-to-moderate positive associations with suicidal ideation; socially prescribed perfectionism predicted longitudinal increases in ideation beyond baseline, and concerns, parental criticism and parental expectations were also associated with suicide attempts. The authors conclude that the strivings result draws into question the idea that such strivings are healthy or adaptive. Limitation: mostly cross-sectional correlational data with self-report measures of ideation, so it establishes association and premorbid risk rather than causation, and no included sample was autistic.

 

4. Woodfin V, Binder PE, Molde H (2020). The psychometric properties of the Frost Multidimensional Perfectionism Scale - Brief. Frontiers in Psychology, 11, 1860. DOI 10.3389/fpsyg.2020.01860. View the paper Confirmatory factor analysis and bifactor modelling of the 8-item FMPS-Brief in 383 Norwegian university students (mean age 27, range 19 to 65; 78.9% female). The two-factor model of strivings and evaluative concerns fitted best (CFI 0.94, RMSEA 0.09) with an overall alpha of .83; evaluative concerns correlated .53 with anxiety and .42 with depression, while strivings correlated .14 with anxiety and did not correlate with depression. The paper also records that Organization correlates loosely with the other subscales and that Frost and colleagues recommended excluding it from total-score calculations. Limitation: a student sample and an 8-item abbreviation, so its coefficients describe the brief form rather than the 35-item scale used by this screener.

 

Lived experience

 

8. Smit S, Hopper J (2023). Love, joy, and a lens of childhood trauma: exploring factors that impact the mental health and well-being of autistic parents via iterative phenomenological analysis. Autism in Adulthood, 5(1), 63-75. DOI 10.1089/aut.2021.0101. View the source Qualitative study of nine autistic parents using remote semi-structured interviews in participants' chosen communication modalities, analysed with Iterative Phenomenological Analysis. Three superordinate themes emerged - Identity and Purpose, Looking Through a Lens of Trauma, and External Factors - with participants describing extreme empathy, perfectionism and a drive to protect their children from the trauma they had experienced, alongside pervasive sensory overload. It is one of very few sources in which autistic adults describe perfectionism in their own words rather than through a questionnaire. Limitation: nine self-selecting participants in one qualitative study, with no measurement of perfectionism and no comparison group, so it describes experience rather than establishing prevalence or severity.

 

Further reading — general background

 

Stober J (1998). The Frost Multidimensional Perfectionism Scale revisited: more perfect with four (instead of six) dimensions. Personality and Individual Differences, 24(4), 481-491. DOI 10.1016/S0191-8869(97)00207-9. View the paper A reanalysis of the FMPS factor structure arguing, as its title states, that the scale is better represented by four dimensions than by the original six. Listed here as further reading for anyone following the long-running argument about how many dimensions this instrument actually has, alongside the three-factor, five-factor and six-factor results reported elsewhere in this module. Limitation: only the bibliographic record could be verified for this entry, so no sample size, fit statistic or specific factor grouping from this paper is quoted anywhere on this page.

 

Peer-reviewed = checked by independent experts before publication. Clinical model = an established professional framework, not a single study.

 

Up next

 

All modules in Autistic Self-Discovery

 

The bar can stay high. The tribunal can go. This course was built by clinicians who are part of the New Path Family. Perfectionism is one of the few things covered here with a substantial treatment literature behind it, and a high Concern over Mistakes score is worth talking through with someone rather than filing away. Therapy for clients in California and coaching worldwide, all by telehealth, are offered by our sister company New Path Family of Therapy Centers, Inc. A conversation costs nothing and there is no pressure. Saving this for later counts too. Talk with the New Path team